Humanized immunoglobulin reacting specifically with Fas ligand or active fragments thereof and region inducing apoptosis originating in Fas ligand

[0002] This application claims the benefit of Japanese application No. 8/231742 filed on Sep. 2, 1996 and Japanese Application No. 8/271546 filed on Sep. 20, 1996 which are incorporated herein by reference.

[0003] This invention relates to novel humanized immunoglobulins, and is based on information about a site on Fas ligand which is important to suppress apoptosis which is induced in Fas-expressing cells through the Fas-Fas ligand interaction. Specifically, the invention relates to humanized immunoglobulins or active fragments thereof which are specifically reactive to Fas ligand. Said immunoglobulins and said site are useful in clinical applications to diseases which are provoked by apoptosis induced by the physiological reactions between Fas antigen and Fas ligand, and for example, are useful in elucidation of Fas system in cell death, in immunological therapy and diagnosis, in detection for Fas ligand, or in the related industrial fields.

[0004] The homeostasis of multicellular organisms is maintained by the growth and death of cells, which are delicately controlled. During the processes of ontogeny, a large number of cells are eliminated through cell death. Organs in adults also maintain their functions by continuously keeping the balance of the growth and death of the organ-constituted cells. Such cell death is referred to as “programmed cell death”, which is a predetermined death, and is discriminated from accidental cell death [Raff, M. C., Nature, vol. 356, p.397-400, 1992].

[0005] These two cell deaths are different from each other in process. It has been understood that the programmed cell death is caused by apoptosis process whereas the accidental cell death is caused by necrosis process resulting in cell death [Kerr, J. F., Brit. J. Cancer, vol. 26, p.239-257, 1972].

[0006] Fas antigen is a cell surface protein which mediates the programmed cell death, apoptosis, and the cDNA of the antigen has been cloned [Nagata, et al., Cell, vol. 66, p.223-243, 1991]. The structure of the resulting cDNA has revealed that human Fas antigen is a transmembrane type protein which consists of 319 amino acid residues containing one transmembrane domain. The extracellular domain of Fas antigen consists of 157 amino acid residues and is rich in cysteine residues. Mouse Fas antigen consists of 306 amino acid residues, and shows 49.3% homology to human Fas antigen.

[0007] The structure of the extracellular domain in Fas antigen which is cysteine-rich has been demonstrated to be a well conservative structure also found in low-affinity receptors for nerve growth factor (NGF) and receptors of tumor necrosis factor (TNF), showing that Fas antigen is a cell surface protein which belongs to a family of NGF/TNF receptor. In 1993, a group of Dr. Nagata, Shigekazu and his coworkers identified the ligand molecule of rat Fas antigen [Nagata, et al., Cell, vol. 75, p.1169-1178, 1993], which had been expected to exist within the living body in the light of the fact that most proteins of this family co-exsit with their ligands in body. Subsequently, the same group identified the molecules of mouse and human Fas ligands [Nagata, et al., Int. Immunol., vol.6, No. 10, p.1567-1574, 19941.

[0008] Dr. Nagata et al. showed that Fas ligand is a protein consisting of 278 amino acids with a molecular weight of 31,138, and that it contains four N-glycosidic linkage sites and thus is a glycoprotein [Nagata, et al., Cellular Engineering, vol.13 No. 8, p.738-744, 19941]. Further, it has been shown that a soluble Fas ligand molecule induces apoptosis in target cells expressing Fas antigen on their cell surfaces (Nagata, et al., J. Exp. Med., vol. 179, p.873-879, 1994].

[0009] Hanabuchi et al. reported that the analysis on mechanism for the cytotoxicity of target cells by killer T cells via Fas antigen reveals that transmission of apoptosis signals via Fas antigen on the target cells may be responsible for the cytotoxicity of the target cells by CD4-positive T cells (CTLs) which does not express perforin, and thereby it was revealed that Fas ligand exists on the cell surface of CD4-positive CTLs [Hanabuchi, et al., Proc. Natl. Acad. Sci. USA, vol.91, No. 11, p.4930-4934, 1994].

[0010] As described above, Fas antigen appears to transmit a signal causing “death” to cells, and it has been shown that the inactivation of the proteins mediating apoptosis such as Fas antigen and Fas ligand causes the overgrowth of cells, while, on the contrary, the extraordinary activation of such proteins causes a certain inflammatory reaction.

[0011] For example, it has been reported that there is a mutation in the Fas gene of a mice which has lpr (lymphoproliferation) mutation causing an autoimmune disease-like symptom, whereas there is a mutation in Fas ligand itself of a mice which has gld (generalized lymphoproliferative disease) mutation also causing an autoimmune disease-like symptom [Nagata, et al., Cell, vol. 76, p.969-979, 1994].

[0012] Further, recent investigations have shown that the physiological reactions between Fas antigen and Fas ligand may cause various diseases.

[0013] For example, it has been reported that tat protein derived from HIV, the AIDS-causative virus, expedites the expression of Fas ligand to induce apoptosis of the T cells expressing Fas antigen, via the interaction of Fas-Fas ligand [Westerndrop, M., et al., Nature, vol. 375, p.497-500, 1995], and the expression of Fas on HIV-infected T cells has been actually found [Kobayashi, et al., Proc. Natl. Acad. Sci. USA, vol. 87, p.9620-9624, 1990]. These reports show that apoptosis induced by the interaction of Fas-Fas ligand may be one of the mechanisms of elimination of CD4-positive T cells in AIDS. Further, there are reports, each of which shows that the mice to which anti-Fas antibody (Jo-2) has been administered intra-abdominally develop fulminant hepatitis to lead death [Ogasawara, et al., Nature, vol. 364, p.806-809, 1993], that the Fas expression is observed in viral hepatitis [Hiramatsu, et al., Hepatology, vol. 19, p. 1354-1359, 1994], and that the Fas expression is also observed in diabetes mellitus and autoimmune diseases such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA), and therefore, it is speculated that these diseases as mentioned above would be caused by Fas ligand, which is reactive to Fas antigen.

[0014] Under the circumstances, creation of agents exhibiting suppressive (inhibitory) activity against the binding between Fas and Fas ligand to lead to suppression of apoptosis will be extremely significant for advanced exploration of the above studies and in particular, treatment of diseases in a future clinical application.

[0015] The inventors of the present invention found the mouse monoclonal antibodies which are specifically reactive to Fas ligand to suppress (inhibit) the physiological reactions between Fas and Fas ligand, and filed a patent application therefor (the Japanese Patent Application No. 303492/1995). abandoned, and to which priority is claimed by PCT Application No. WO96/29350. It has been believed that since the monoclonal antibodies bind to Fas ligand more strongly than Fas, the antibodies may inhibit the physiological reactions of Fas-Fas ligand in vivo.

[0016] It is natural that such highly active monoclonal antibodies are desirable in clinical application, but, unfortunately, non-human immunoglobulins such as the aforementioned mouse monoclonal antibodies suffer disadvantages in application to human, which are provided below. The non-human immunoglobulins show the relatively shorter half life in vivo, and require more frequent administrations to maintain the predetermined level in blood compared to human antibody. It should be noted that the non-human immunoglobulins contain an amino acid sequence which may exert antigenicity on the administration to human. Thus, in the case of frequent administration of the non-human immunoglobulins to human, an immune response elicited by the administration eliminates the immunoglobulins administered at the later stages, and ultimately may cause anaphylaxis-like shocks.

[0017] In order to resolve the above problems, so called chimeric antibody, an immunoglobulin having a constant region from human immunoglobulin combined to a variable region from mouse immunoglobulin has been created [LoBuglio, et al., Proc. Natl. Acad. Sci. USA, vol. 86, p.4220-4224, 1989]. This described that the half life of the chimeric antibody in human was prolonged six times compared to the mouse antibody, but the period is still only about ⅕ of that of the common human immunoglobulin. Also, the article reported that the immune response to the chimeric antibody was observed in one of three patients receiving the chimeric antibody. This immune response was believed to be derived from the variable region from the mouse immunoglobulin since the response could be absorbed when anti-mouse immunoglobulins were reacted with the chimeric antibody.

[0018] To resolve the problem which could not be overcome by the chimera technique with human immunoglobulin, Winter, et al. reported the process for constructing humanized immunoglobulin which is more analogous to the human immunoglobulin than the chimeric antibody, which comprises transplanting the complementarity determining region (hereinafter may be referred to as CDR) directly binding to an antigen, which is within the variable region (V region), into the CDR in the variable region of human immunoglobulin according to genetic engineering techniques [Winter, et al., Nature, vol. 321, p.522-525, 1986].

[0019] As reviewed in Bendig, A Companion to Methods in Enzymology, vol. 8, p.83-93, 1995, diverse humanized immunoglobulins have been previously prepared by procedures as mentioned above. However, compared to the original mouse immunoglobulins, which are donors of the CDR, most of humanized immunoglobulins involve significant decrease in their activity, and several maintain the same activity, above all, very few show the higher activity. Needless to say, no reports describe any preparation of a humanized immunoglobulin directed to Fas ligand according to the present invention.

[0020] The recent investigation has reported a putative three-dimensional structure of mouse Fas ligand [Manuel C. P., et al., Molecular Immunology, vol. 32 (10), p.761-772, 1996]. In this report, the authors believed that mouse Fas ligand should form a trimer structure to exhibit its biological activity similarly to TNF-α and TNF-β (hereinafter referred together to as TNF, simply), and, making reference to the structure of TNF trimer and the interaction site between TNF and TNF receptor, they predicted a interaction site between mouse Fas ligand monomers, and a interaction site between mouse Fas ligand and Fas antigen, after preparation of a trimer model for mouse Fas ligand. However, this prediction has not been supported therein any more.

[0021] On the other hand, any putative three-dimensional structure model for human Fas ligand as shown in the above literature has not been reported, and there have been no reports that show the interaction site between the Fas ligand monomers and the interaction site of the Fas-Fas ligand. Under the circumstances, it has been still unknown which regions in Fas ligand or Fas should be targeted to suppress (or inhibit) apoptosis induced by the physiological reactions of Fas-Fas ligand.

[0022] The decrease in the activity of humanized immunoglobulin compared to the original mouse immunoglobulin is believed to be mainly due to less antigen-binding activity induced by structural change in the CDR to be transplanted, which change is on the basis of the difference in stereochemical structure of the framework region between mouse immunoglobulin as CDR donor and human immunoglobulin as CDR acceptor. In order to avoid such decrease in the activity, any modification is required in the process for humanization of an antibody by CDR transplantation.

[0023] The monoclonal antibodies specifically reactive to Fas ligand which are encompassed within the scope of the invention of the Japanese Patent Application No. 303492/1995, have the useful activity to suppress (or inhibit) the physiological reactions between Fas and Fas ligand. However, the antibodies are a mouse monoclonal antibody, and therefore, cannot be practically applied to the clinical field of human in the light of the safety (induction of antigenicity) and the efficacy (shortening of half life).

[0024] As a major embodiment, the present invention provides humanized immunoglobulins or active fragments thereof which are specifically reactive to Fas ligand, and in particular, provides humanized immunoglobulins which are specifically reactive to Fas ligand to be able to suppress (or inhibit) the physiological reactions between Fas and Fas ligand.

[0025] So far, the interaction site between human Fas and Fas ligand has not been shown, and a modelling or a crystal structure analysis for the Fas-Fas ligand complex has not been performed, and therefore, it has been impossible to identify a site to effectively suppress (or inhibit) apoptosis induced by the physiological reactions of Fas-Fas ligand.

[0026] As another embodiment, the present invention demonstrates for the first time a site on human Fas ligand associated with apoptosis-suppressing activity, by means of examining the reactivities of diverse Fas ligand molecules introduced with an amino acid substitution (Fas ligand variants) with diverse anti-Fas ligand monoclonal antibodies having the high apoptosis-suppressing activity, constructing a molecular model for the human Fas ligand trimer, and further confirming the distribution of the site identified in the above experiment within the human Fas ligand trimer.

[0027] FIG. 1 depicts the nucleic acid and amino acid sequence of the variable region of the heavy chain of NOK2 antibody (SEQ ID NO: 142 and SEQ ID NO: 143), wherein the underlined portions represent primers used in the cloning (SEQ ID NO: 141 and 144).

[0028] FIG. 2 depicts the nucleic acid and amino acid sequences of the variable region of the light chain of NOK2 antibody (SEQ ID NO: 146 and SEQ ID NO: 147), wherein the underlined portions represent primers used in the cloning (SEQ ID NO: 145 and SEQ ID NO: 148).

[0029] FIG. 3 depicts (a) the expression plasmid for the heavy chain of chimeric NOK2 antibody, and (b) the expression plasmid for the light chain of chimeric NOK2 antibody.

[0030] FIG. 4 shows the result of ELISA for assaying the ability of chimeric NOK2 antibody to recognize Fas ligand.

[0031] FIG. 5 depicts the amino acid sequences of the variable regions of the heavy chains of NOK2 antibody (SEQ ID NO: 149), SGI antibody (SEQ ID NO: 150), and humanized NOK2 (RNOK2) immunoglobulins (RNOK2VHver11 (SEQ ID NO: 151) and RNOK2VHver12 (SEQ ID NO: 152)).

[0032] FIG. 6 depicts the amino acid sequences of the variable regions of the light chains of NOK2 antibody (SEQ ID NO: 153), REI antibody (SEQ ID NO: 160), huVL19 (SEQ ID NO: 155) and huVL31 (SEQ ID NO: 154) immunoglobulins from human, and RNOK2 immunoglobulins (RNOK2VLver1 (SEQ ID NO: 161), RNOK2VLver21 (SEQ ID NO: 156), RNOK2VLver22 (SEQ ID NO: 157), RNOK2VLver23 (SEQ ID NO: 158) and RNOK2VLver24 (SEQ ID NO: 159)).

[0033] FIG. 7 depicts the amino acid sequences of the variable regions of the light chains of NOK2 antibody (SEQ ID NO: 153), REI antibody (SEQ ID NO: 160), huVL19 (SEQ ID NO: 155) and huVL31 (SEQ ID NO: 154) immunoglobulins from human, and RNOK2 immunoglobulins (RNOK2VLver1 (SEQ ID NO: 161), RNOK2VLver21 (SEQ ID NO: 156), RNOK2VLver22 (SEQ ID NO: 157), RNOK2VLver23 (SEQ ID NO: 158) and RNOK2VLver24 (SEQ ID NO: 159)) continued from FIG. 6.

[0034] FIG. 8 illustrates the annealing of primers (#01 (SEQ ID NO: 162), #02 (SEQ ID NO: 163) and #03 (SEQ ID NO: 164)) with a template wherein the primers are used in preparation of the genes of the variable regions of the heavy chains of RNOK2 immunoglobulins (RNOK2VHver11 and RNOK2VHver12), and the template is the variable region of the heavy chain of RC25 immunoglobulin.

[0035] FIG. 9 illustrates the annealing of primers (#02 (SEQ ID NO: 163), #03 (SEQ ID NO: 164) and #04 (SEQ ID NO: 165)) with a template wherein the primers are used in preparation of the genes of the variable regions of the heavy chains of RNOK2 immunoglobulins (RNOK2VHver11 and RNOK2VHver12), and the template is the variable region of the heavy chain of RC25 immunoglobulin.

[0036] FIG. 10 illustrates the annealing of primers (#05 (SEQ ID NO: 166), #06 (SEQ ID NO: 167) and #07 (SEQ ID NO: 168)) with a template wherein the primers are used in the preparation of the gene of the variable region of the light chain of RNOK2 immunoglobulin (RNOK2VLver1), and the template is the light chain of NOK2 antibody.

[0037] FIG. 11 illustrates the annealing of primers (#08 (SEQ ID NO: 169), #09 (SEQ ID NO: 170) and #10 (SEQ ID NO: 171)) with a template wherein the primers are used in the preparation of the gene of the variable region of the light chain of RNOK2 immunoglobulin (RNOK2VLver1), and the template is the light chain of NOK2 antibody.

[0038] FIG. 12 illustrates the annealing of a primer (#11 (SEQ ID NO: 172)) with a template wherein the primer is used in preparation of the genes of the variable regions of the light chains of RNOK2 immunoglobulins (RNOK2VLver21, RNOK2VLver22, RNOK2VLver23 and RNOK2VLver24), and the template is the light chain of NOK2 antibody.

[0039] FIG. 13 illustrates the annealing of a primer (#12 (SEQ ID NO: 173)) with a template wherein the primer is used in preparation of the genes of the variable regions of the light chains of RNOK2 immunoglobulins (RNOK2VLver21, RNOK2VLver22, RNOK2VLver23 and RNOK2VLver24), and the template is the light chain of NOK2 antibody.

[0040] FIG. 14 shows the result of ELISA for assaying the ability of RNOK2 immunoglobulins to recognize Fas ligand.

[0041] FIG. 15 shows the result of inhibitory activities of RNOK2 immunoglobulins and mouse NOK2 immunoglobulin against apoptosis mediated by Fas ligand and Fas.

[0042] FIG. 16 depicts parts of the nucleic acid sequence (upper line) (SEQ ID NO: 174) and amino acid sequence (lower line) (SEQ ID NO: 175) of Fas ligand. Numerical order of the amino acid is in accordance with Nagata, et al. Int. Immunology, vol. 6, p.1567-1574, 1994.

[0043] FIG. 17 shows, in part, the result of ELISA for assaying the ability of anti-FLAG antibody to recognize the expressed native Fas ligand and the Fas ligand variants.

[0044] FIG. 18 shows, in part, the result of ELISA for assaying the ability of anti-Fas ligand antibody (NOK1 antibody) to recognize the expressed native Fas ligand and the Fas ligand variants.

[0045] FIG. 19 shows the relative ability of NOK and humanized NOK2 antibodies to recognize each of the Fas ligand variants. N.D. means “not done”.

[0046] FIG. 20 shows the relative ability of NOK and humanized NOK2 antibodies to recognize each of the Fas ligand variants. N.D. means “not done”.

[0047] FIG. 21 depicts the alignment of the amino acid sequence of Fas ligand with that of TNF-α and -β. The first line represents the amino acid sequence of the extracellular domain of human Fas ligand as shown in Nagata, et al., Int. Immunology, vol. 6, p.1567-1574, 1994 (SEQ ID NO: 176), the second line represents amino acid sequence of the region undergoing the modeling (SEQ ID NO: 177), and the third and fourth lines represent amino acid sequences of TNF-α (1TNF) (SEQ ID NO: 178) and TNF-β (1TNR) (SEQ ID NO: 179), respectively.

[0048] In FIG. 22, the amino acids on the two molecules of Fas ligand trimer, which are recognized by NOK1 antibody are indicated by circles. Shaded circles show the amino acids resided on the same side, whereas open circles show any others.

[0049] In FIG. 23, the amino acids on the two molecules of Fas ligand trimer, which are recognized by NOK2 and humanized NOK2 antibodies are indicated by circles. Shaded circles show the amino acid resided on the same side, whereas open circles show any others. Circles with dotted line show the amino acids especially recognized by humanized NOK2 antibody.

[0050] In FIG. 24, the amino acids on the two molecules of Fas ligand trimer, which are recognized by NOK3 antibody are indicated by circles. Shaded circles show the amino acid resided on the same side, whereas open circles show any others.

[0051] FIG. 25 depicts that alignment of the VH region of NOK2 antibody (SEQ ID NO: 180) with that of 1FOR antibody (SEQ ID NO: 181). The constant region which resides downstream of the VH region is from 1FOR antibody.

[0052] FIG. 26 depicts the alignment of the VL region of NOK2 antibody (SEQ ID NO: 182) with that of 1TET antibody (SEQ ID NO: 183). The constant region which resides downstream of the VL region is from 1TET antibody.

[0053] In FIG. 27, the amino acids on the two molecules of Fas ligand trimer, which are recognized by NOK and humanized NOK2 antibodies are indicated by circles. Shaded circles show the amino acids resided on the same side, whereas open circles show any others. The circle encompassing the amino acids indicated by the shaded circles is of about 17 Å (angstrom) radius, which scale is near the CDR of antibody.

[0054] In FIG. 28, the amino acids presented on the surface of Fas ligand (which may be recognized by the antibody, etc.) among all amino acids encompassed by the circle of about 17 Å radius as shown in FIG. 27 are indicated by circles. Shaded circles show the amino acids recognized by NOK and humanized NOK2 antibodies.

[0055] FIGS. 29-31 are diagrams representing a van der Waals contact model for the two molecules of Fas ligand trimer, which are indicated by white and black, respectively. Hatched areas in FIGS. 29, 30, and 31 show the amino acids recognized by NOK1 antibody, those recognized by NOK2 and humanized NOK2 antibodies, and those recognized by NOK3 antibody, respectively.

[0056] FIGS. 32 and 33 are diagrams representing the site on the Fas ligand trimer recognized by NOK2 antibody, viewing from the side. In FIG. 32, the upper represents the Fas ligand trimer, the lower represents the Fab portion of NOK2 antibody, and the intermediate represents the complementarity determining region (CDR) of NOK2 antibody. The size of the CDR of NOK2 antibody as shown in FIG. 33 is nearly equal to the circle of about 17 Å radius.

[0057] FIG. 34 is the diagram representing a van der Waals contact model similar to in FIGS. 29-31, and the hatched area shows the amino acids having a presenting surface of 15 angstrom square or above within the circle of about 17 Å radius as shown in FIG. 33.

[0058] Practical usage of the mouse monoclonal antibodies which are specifically reactive to Fas ligand, within the invention of the Japanese Patent Application No. 303492/1995 requires to convert them into a human-like immunoglobulin which both retains the property to suppress (inhibit) the physiological reactions between Fas and Fas ligand by specifically reacting with Fas, and does not induce antigenicity in human body. The possible method therefor includes alteration of the mouse immunoglobulin to form the human immunoglobulin by genetic engineering techniques as detailed in, for example, Winter, et al., Nature, vol. 321, p.522-525, 1986. Many humanized antibodies prepared according to such method as this, however, decrease significantly in their activity compared to the original mouse antibody, a donor for CDR, as described above and therefore, there have been a demand for an improved method on the basis of CDR implantation to prepare the humanized antibody retaining the high activity.

[0059] Previously, it has been suggested that the amino acids within a framework region (FR) (hereinafter may be referred to as FR amino acids) would reside near the CDR in terms of either primary structure or stereochemistry, which are responsible for maintaining the stereochemical structure of CDR leading to retaining the activity of immunoglobulin, in the reports such as Winter, et al., Nature, vol.332, p.323-327, 1988 as to humanization of mouse immunoglobulin, and Chothia, et al., J. Mol. Biol, vol. 196, p.901-917, 1987 as to the stereochemical structure of immunoglobulin CDR.

[0060] In humanization of the monoclonal antibody specifically reactive to Fas ligand according to the present invention, the FR amino acids which are selected using the computer modelling together with the CDR were transplanted into a recipient human antibody on the basis of the above information, so that the humanized antibody having the activity equal to or above the original mouse antibody successfully have been prepared.

[0061] First, computer modelling of antibody was performed to predict significant amino acids in the FR which are responsible for maintaining the stereochemical structure of the CDR leading to retaining the activity of antibody. For example, QUANTA/CHARMm or Modeler which is a software for molecular design starting on Silicongraphics (both are produced by Molecular Simulations Inc.) can be used for the computer modelling. On the basis of the data for the three-dimensional structure of antibody constructed using such software, the FR amino acids which are responsible for maintaining the stereochemical structure of CDR are selected. Specifically, the amino acids which form hydrogen bond or undergo energy contact directly or indirectly to any CDR amino acid are selected using calculation program installed in the software. Amino acids to be transplanted together with the CDR are selected from these amino acids, provided that the amino acids should not lead to any amino acid sequences which are not found in human antibodies and may elicit antigenicity in human body in the course of transplantation.

[0062] The relationship in each amino acid in the variable regions between mouse and human antibodies is readily understood from the classification of Kabat, et al., Sequence of Proteins of Immunological Interest, 4th ed., Public Health Service, NIH, Washington D.C., 1987, which is a standard in the art.

[0063] In accordance with the above procedures, the present invention has been accomplished. Thus, the invention provides humanized immunoglobulins which can be specifically reactive to Fas ligand to suppress (inhibit) the physiological reactions between Fas and Fas ligand without inducing antigenicity in human, and which have the activity equal to or above the original mouse antibody.

[0064] Nowadays, the interaction site between human Fas and Fas ligand has not been shown, and a modelling or a crystal structure analysis for the Fas-Fas ligand complex has not been also performed, and therefore, it has been unknown which site in Fas ligand or Fas antibody should be blocked to effectively suppress (or inhibit) apoptosis induced by the physiological reactions of the Fas-Fas ligand.

[0065] Then, the recognition sites which are recognized by various anti-Fas ligand monoclonal antibodies having the high apoptosis-suppressing activity was examined in order to identify a site on human Fas ligand which is important to suppress (inhibit) apoptosis. Specifically, the inventors prepared diverse Fas ligand molecules incorporated with an amino acid substitution (i.e., Fas ligand variants), examined their reactivities to anti-Fas ligand monoclonal antibodies produced by the hybridomas, NOK1, NOK2, NOK3, and NOK4 as well as the humanized immunoglobulin of the present invention, and identified the site on the human Fas ligand molecules recognized by the above antibodies, so as to demonstrate the important site to suppress apoptosis; hereinafter the aforementioned anti-Fas ligand monoclonal antibodies may be referred to as NOK1, NOK2, NOK3, and NOK4 immunoglobulin (antibody), respectively; the aforementioned hybridomas have been deposited at the National Institute of Bioscience and Human Technology, Ministry of International Trade and Industry (1-1-3 Higashi, Tsukuba shi, Ibaraki) under Accession Nos. FERM BP-5044, FERM BP-5045, FERM BP-5046, and FERM BP-5047; and the antibodies generically may be referred to as NOK antibody.

[0066] FERM BP-5045 was deposited on Mar. 20, 1995, under the terms of the Budapest Treaty. Applicants have deposited NOK 2 with the National Institute of Bioscience and Human-Technology Agency of Industrial Science and Technology (NIBHTAIST), 1-3, Higashi 1 chome Tsukuba-shi Ibaraki-ken, 305 Japan. The NOK 2 samples deposited with the NIBHTAIST are taken from the same deposit maintained by Sumitomo Electric Industries, Ltd., 5-33, Kitahama 4-chome, Chuo-ku, Osaka-shi, Osaka 541 Japan since prior to the filing date of this application. The deposits of the NOK 2 samples will be maintained without restriction in the NIBHTAIST depository for a period of 30 years, or 5 years after the most recent request, or for the effective life of the patent, whichever is longer, and will be replaced if the deposit becomes non-viable during that period.

[0067] Further, the inventors constructed a human Fas ligand trimer model, and confirmed the distribution of the site identified as shown above within the human Fas ligand trimer, thereby demonstrating for the first time the site on the Fas ligand associated with apoptosis-inhibitory activity.

[0068] As described above, Fas ligand (hereinafter may be referred to as FasL) is a ligand for Fas antigen which is a cell surface protein which mediates programmed cell death (apoptosis), and Fas ligand has been identified from human, rat, and mouse, so far. The present invention may cover a general Fas ligand, and among them, Fas ligands from human and mouse are preferred. Thus, the invention preferably provides humanized immunoglobulins and active fragments thereof which are specifically reactive to a ligand for human Fas antigen or a ligand for mouse Fas antigen.

[0069] As used herein, “immunoglobulin” means a protein which consists of one or more polypeptides essentially encoded by a immunoglobulin gene. The immunoglobulin genes include the genes of a kappa, a lambda, an alpha, a gamma, a delta, an epsilon, and a mu constant regions and the gene of myriad immunoglobulin variable region. The immunoglobulin of the present invention also encompasses active fragments. The active fragments mean fragments of antibody having an antigen-antibody reaction activity, and include F(ab′)₂, Fab′, Fab, Fv, and recombinant Fv.

[0070] As used herein, “chimeric antibody” means an antibody derived from a gene which is combined by segments of the immunoglobulin genes wherein the light and heavy chain genes are derived from different species, typically according to genetic engineering techniques. For example, the segment of the variable region (hereinafter referred to as V region) gene from mouse monoclonal antibody may be combined with the segment of the human constant region (hereinafter referred to as C region) gene, e.g., γ1 or γ4. Accordingly, the typical chimeric antibody for therapeutic purpose includes a hybrid protein which composes of the V region or antigen-binding region from the mouse antibody and the C region or effector region from the human immunoglobulin, although any other mammalian species can be used.

[0071] As used herein, “framework region” means parts of the variable regions of the light and heavy chains of immunoglobulins, which is relatively conserved among various immunoglobulins within a single specie, or which is other than CDR. “Human framework region” is substantially the same (about 85% or above) as a frame work region of a naturally-occurring human immunoglobulin or the same as a common sequence among such immunoglobulins. “Humanized immunoglobulin” means an immunoglobulin which contains the human framework and at least one CDR from non-human immunoglobulin, and of which the constant region is substantially identical, i.e., about 85-90% identical, preferably, at least 95% identical to that of the human immunoglobulin in terms of the amino acid sequence of the protein. Accordingly, all parts of the humanized immunoglobulin except the CDR may be substantially the same as the corresponding parts of one or more natural human immunoglobulins.

[0072] Humanized immunoglobulin possesses at least three possible advantages over mouse antibody from the standpoint of therapeutic application to human.

[0073] (1) Humanized immunoglobulin is compatible with other aspects of the immune system in human since the effector part thereof is from human, and for example it can cause more effective dissolution of the targeted cells by complement-dependent cytolysis (CDC) or antibody-dependent cell-mediated cytotoxicity (ADCC);

[0074] (2) Human immune system does not recognize the framework or the C region of humanized immunoglobulin, and therefore, the level of undesirable immune responses to humanized immunoglobulin is lower than mouse antibody in which the whole is xenogenic and the chimeric antibody in which the part is xenogenic; and

[0075] (3) It has been reported that mouse antibody has the much shorter half life than common human antibodies when infused in human (Shaw, D R et al., J. Immnol., vol. 138, p.4534-4538, 1987). On the other hand, humanized immunoglobulin has the half life nearer to naturally-occurring human antibodies compared to mouse antibody, and therefore, can be expected to provide the lower dose and the lower frequency of the administration.

[0076] The humanized immunoglobulins of the present invention may be limitless as long as they are specifically reactive to Fas ligand, and those which can inhibit (suppress) the physiological reactions between Fas ligand and Fas are especially preferred. As used herein, an immunoglobulin which inhibits the physiological reactions means an immunoglobulin which can specifically bind to the binding site of Fas ligand to be bound to Fas so as to interfere with Fas ligand in binding to Fas under the circumstances that cells expressing Fas ligand or the solubilized Fas ligand (hereinafter may be referred to as sFas ligand or soluble Fas ligand) are bound to cells expressing Fas to afford signals causing death by apoptosis to the Fas-expressing cells. In other words, immunoglobulins which inhibit the physiological reactions between Fas ligand and Fas prevent the Fas ligand-expressing cells or sFas ligand from killing the Fas-expressing cells. Preferably, the immunoglobulins used in this situation are those having the higher ability to bind to Fas ligand than Fas. Specifically, the binding ability may be examined by using as an indicator a chimeric molecule comprising Fas combined by the Fc of IgG (hereinafter may be referred to as Fas-Ig). The Fas-Ig can bind to Fas ligand as strongly as Fas in body. Thus, when an immunoglobulin directed to Fas ligand can inhibit the binding between Fas ligand and Fas at the lower concentration than that by the Fas-Ig chimera molecule, then the immunoglobulin would be practically able to inhibit various actions in body caused by Fas ligand.

[0077] The humanized immunoglobulins of the present invention which are specifically reactive to human Fas ligand include, for example, RNOK201, RNOK202, and RNOK203 antibodies, which are obtained by humanizing NOK2 antibody, as a donor antibody, which is included in the invention of the Japanese Patent Application No. 303492/1995 and is produced by hybridoma NOK2, which has been deposited at the National Institute of Bioscience and Human Technology, Ministry of International Trade and Industry under Accession No. FERM BP-5045 (hereinafter, the antibodies humanized from NOK2 antibody may be referred to as humanized NOK2 antibody, humanized NOK2 immunoglobulin, RNOK2 antibody or RNOK2 immunoglobulin).

[0078] The following is presented by way of illustration of certain suitable embodiments of steps for humanization of NOK2 antibody, which should not be construed as limiting to this, but any antibodies leading to the humanized immunoglobulin which is specifically reactive to Fas ligand may be used.

[0079] First, the cloning of the gene of variable region (V region) from NOK2 antibody is necessary to convert NOK2 antibody into a molecule having the amino acid sequence of human immunoglobulin by genetic engineering techniques.

[0080] The V region gene may be cloned by conventional gene manipulation techniques. For example, the techniques include the cloning of the V region gene from chromosomal DNA of cells according to the conventional method such as T. Maniatis, Molecular Cloning Harbor Laboratories, 1982, and the cloning of the V region gene by preparing cDNA from mRNA in cells according to the conventional method such as D. M. Glover ed., DNA Cloning, vol. 1, IRL press, 1985.

[0081] In either method, DNA probes which are synthesized making reference to the nucleic acid sequence of the mouse immunoglobulin gene as previously reported such as Sakano et al., Nature, vol. 286, p.676, 1980 and E. E. Max et al., J. Biol. Chem. vol. 256, p.5116, 1981, may be utilized as probes for the cloning of the V region gene. Also, the cloning by polymerase chain reaction (PCR) can be utilized (R. Orlandi et al., Proc. Natl. Acad. Sci. USA, vol. 86, p.3883, 1989; W. D. Huse et al., Science, vol. 246, p.1275, 1989). According to these typical methods, the gene of the V region of NOK2 antibody may be cloned from the NOK2 antibody-producing hybridoma to determine the base sequence of the gene and the amino acid sequence corresponding thereto. See FIGS. 1 and 2.

[0082] The chimeric NOK2 antibody gene may be prepared by ligating the V region gene segment to the upstream of the constant region of human immunoglobulin. ELISA reveals that said chimeric NOK2 antibody binds to the Fas ligand molecule in a concentration-dependent manner, confirming that the gene of the variable region of immunoglobulin cloned by the present inventors should encode the anti-Fas ligand activity. See FIG. 4.

[0083] The gene of humanized NOK2 antibody, which is more closely related to human immunoglobulin than the chimeric antibody may be prepared according to the method of Winter et al., Nature, vol. 321, p.522-525, 1986 by transplanting the complementarity determining regions (CDR) which reside within the V region and directly bind to antigens into the CDR in the variable region of human immunoglobulin. Previously, it has been suggested that the amino acids within the framework (FR) (hereinafter may be referred to as FR amino acids) which are responsible for maintaining the stereochemical structure of the CDR and responsible for retaining the activity of immunoglobulin would reside near the CDR on either primary structure or stereochemistry, in the reports such as Winter, et al., Nature, vol. 332, p.323-327, 1988 as to re-shaping of mouse immunoglobulin, and Chothia, et al., J. Mol. Biol, vol. 196, p.901-917, 1987 as to the stereochemical structure of the immunoglobulin CDR. On the basis of these findings, the FR amino acids which have been suggested to be responsible for maintaining stereochemical structure of the CDR and responsible for retaining the activity of immunoglobulin are also transplanted together with the CDR. The FR amino acids to be transplanted are selected from those which are shown by the computer modelling of NOK2 antibody to interact directly or indirectly with the amino acids of the CDR (hereinafter may be referred to as CDR amino acids).

[0084] That is to say, first, the three-dimensional structure of NOK2 antibody may be estimated with the modelling using QUANTA/CHARMm or Modeler which is a software starting on Silicongraphics (both are produced by Molecular Simulations Inc.). For example, the QUANTA/CHARMm or Modeler, a software for molecular design, can be used according to the manufacture's instructions to estimate the structure of NOK2 antibody by utilizing as a template for the three-dimensional structure the variable region of the H chain (VH) of PDB ID:1FOR and the variable region of the L chain (VL) of PDB ID:1TET recorded on Brookhaven Protein Data Bank (PDB), which show the high homology to the VH and the VL of NOK2 antibody, although the used template is not limited to such variable regions of the antibody, and other data on the higher-order structure of an antibody which shows homology to the variable region of NOK2 antibody may be used if available. Further, not only QUANTA/CHARMm or Modeler, but also a software for the molecular design which is available for any other proteins as a whole may be used as a software.

[0085] Then, by using the installed program, the amino acids within the FR are selected which form the hydrogen bond to the CDR of the H and L chains in the above estimated three-dimensional structure of NOK2 antibody (the first group), and further, the amino acids within the FR which form the hydrogen bond to the resulting amino acids (the second group) are selected. Similarly, using the installed program, the amino acids within the FR are selected which form the energy contact to the CDR in the H and L chains of NOK2 antibody (the first group), and further, the amino acids within the FR which form the energy contact to the resulting amino acids (the second group) are selected. The energy contact herein includes so-called electrostatic interaction and van der Waals forces. The amino acids which are transplanted into the FR in the variable region of human immunoglobulin together with the CDR amino acids are selected from a group consisting of the first group and the second group as shown above. It should be noted that when the transplantation of any FR amino acids among the above FR amino acids into the relevant site in human immunoglobulin raises any sequences which can not be cited as the variable amino acid sequences of human immunoglobulin in the classification of Kabat, et al., Sequence of Proteins of Immunological Interest, 4th ed., Public Health Service, NIH, Washington D.C., 1987, and a software for information retrieval, Entrez (trademark), which has developed by and National Center for Biotechnology Information: NCBI, then such FR amino acids should not be transplanted. This reduces the possibility to elicit the antigenicity as much as possible when the humanized immunoglobulin is administered to human. With respect to the amino acid sequence of the CDR of NOK2 antibody, the sequences of the heavy chains CDR1, CDR2, and CDR3 are each described in SEQ ID Nos: 1, 2, and 3 in the sequence listing, and the sequences of the light chains CDR1, CDR2, and CDR3 are each described in SEQ ID Nos: 4, 5, and 6 in the sequence listing. The condition of the spans of FR and CDR herein is in accordance with the classification of Kabat, et al., as shown above.

[0086] It is preferred that the amino acid sequence of the variable region of human immunoglobulin which receives the transplantation has the high homology to those of the variable region of mouse immunoglobulin to be humanized. Empirically, it is desirable that the amino acid sequence of the framework region of the recipient human immunoglobulin has at least 60% homology to those of the framework region of the donor antibody (e.g. a mouse immunoglobulin). Usually, any human immunoglobulin having the high homology to the mouse immunoglobulin to be humanized is selected using the available database and utilized. The preferred embodiment of the invention involves transplantation of the CDR of NOK2 antibody into the variable region of human immunoglobulin. Specifically, the CDR in the VH region (the variable region of the heavy chain) of NOK2 antibody is transplanted into SGI (SEQ ID No. 7: distributed by Dr. Bendig of MRC Collaborative Center, United Kingdom), which is the VH region containing the FR (framework) region of human subgroup II. The CDR in the VL region (the variable region of the light chain) of NOK2 antibody is transplanted into REI as previously reported, which is the VL region containing the FR of the human κ chain (W. Palm et al., Physiol.Chem., vol. 356, p167, 1975), and into the VL region containing the FR of the κ chain cloned from the cDNA library derived from human peripheral blood lymphocytes (huVL-19: SEQ ID No. 8 and huVL-31: SEQ ID No. 9). Practical transplantation of the amino acids is performed at the genetic level by PCR mutagenesis whereby mutations are introduced by PCR using the VH gene of RC25 antibody, a humanized anti-HIV immunoglobulin (the International Publication No. WO94/20632) as a template for the VH, and using the VL gene of the chimeric NOK2 prepared as shown above as a template for the VL. For example, with respect to the genes of the H and L chains of humanized NOK2 antibody, two kinds of the H chain, and five kinds of the L chain, which combination affords 10 kinds of humanized NOK2 antibody, are prepared.

[0087] The immunoglobulins of the present invention which encompass the binding fragment and any other derivatives thereof may be readily prepared by various recombinant DNA techniques, and ultimately be expressed in the transfected cells, preferably the immortalized eucaryotic cells such as myeloma and the hybridoma. The polynucleotide which consists of a set of the first sequence encoding the framework of the humanized immunoglobulin and the second sequence encoding the complementarity determining region of the desired immunoglobulin is prepared synthetically or by the combination of the suitable cDNA and genomic DNA segments.

[0088] The genes which encodes the H and L chains for the construction of the humanized antibody and are prepared with the manner as described above are expressed in host cells after combining the sequence to an expression control sequence so as to work. The expression vectors are typically replicable in host cells as episome or a significant portion of the chromosomal DNA. The expression vectors usually contain selectable markers such as a gene for tetracycline resistance, G418 resistance or mycophenolic acid resistance, or HSV-tk, which permits to detect the cells into which the desired DNA sequence is transformed.

[0089] Escherichia coliis a prokaryotic host cell which is especially useful for the cloning of the DNA sequence of the present invention. Additionally, Bacilli such as Bacillus subtilis, and any other enterobacteria such as Salmonella, Serratia, and various Pseudomonas species may be used as host cells. The prokaryotic host cells may be used in the construction of expression vectors, which typically contain an expression control sequence which is suitable to the host cells. Also, the expression vectors may contain any number of diverse known promoters such as lactose-promoter system, tryptophan (trp)-promoter system, β-lactamase-promoter system or the promoter system from λ phage. Promoters regulate the expression, sometimes together with the operator sequence, and comprise the ribosome-binding site sequence, or the like, which permits to initiate and terminate the transcription and translation.

[0090] Other microorganisms such as yeast, for example, may be used as an eukaryotic host cell for the expression. Saccharomyces, which contain a suitable vector having an expression control sequence such as a promoter comprising a gene for 3-phosphoglycerate kinase or any other glycolytic enzymes, and a replication origin, a terminal sequence and a desired analogue, is a preferable host cell.

[0091] In the preparation of the humanized immunoglobulin of the present invention, insect cell cultures, typically the expression system on Baculoviridae also may be used. The humanized immunoglobulin also may be prepared by expressing the polynucleotide sequence encoding the humanized immunoglobulin according to the method reported in ZuPutlitz. J. et al., Bio/Technology, vol. 8, p.651-654, 1990.

[0092] In addition to the aforementioned host cells, mammalian cell cultures also may be used to express and prepare the humanized immunoglobulin of the present invention. For example, the preferable mammalian host includes CHO cell line, diverse COS cell lines, Hela cells, preferably myeloma cell line, or the transformed B cells or hybridoma, which are among many host cell lines which have been developed for secretion of whole immunoglobulin in the art. Expression vectors for these cells contain a expression control sequence such as a replication origin, a promoter, an enhancer and the necessary processing information site such as a ribosome-binding site, an RNA splicing site, a polyadenylation site, and a transcription terminator sequences. Preferable expression control sequences include an enhancer and a promoter which are derived from immunoglobulin gene, SV40, adenovirus, bovine papilloma virus, cytomegalovirus, chicken β-actin genes.

[0093] In the suitable embodiment of the invention, the genes prepared as described above encoding the H and L chains of humanized NOK2 antibody are each ligated into an expression vector which contains the gene of the constant region of human immunoglobulin (H chain: Cγ1, L chain: Cκ), the enhancer from cytomegalovirus, the promoter of chicken β-actin gene, splice acceptor site of rabbit β-globin gene, and a gene for a drug-resistant marker (H chain: neor, L chain: dhfr), and the resultant vector is introduced into a host cell by the well known methods involving any modifications depending on the type of host cell. For example, transfection with calcium chloride is usually applied to prokaryotic cells whereas the treatment by calcium phosphate, lipofection, biolysistics, transduction via virus, or electropolation are applied to any other cells.

[0094] Once the desired immunoglobulin is expressed, a whole immunoglobulin of the present invention, dimers thereof, individual light and heavy chains, or any other types of the immunoglobulin are purified according to a conventional method in the art. The method includes precipitation by ammonium sulfate, various ion-exchange chromatography, and affinity chromatography.

[0095] For the purpose of pharmaceuticals, the immunoglobulin is preferably substantially pure with at least about 90-95% homogeneity, and more preferably, 98-99% or above homogeneity.

[0096] The humanized immunoglobulin of the invention at a concentration of 0.06 μg/ml or above (i.e., effective concentration) can inhibit apoptosis of the Fas-expressing cells induced by the soluble Fas ligand, by 90% or above of apoptosis inhibition rate. The apoptosis inhibition rate herein means a viable rate of target cells to which the immunoglobulin was added in the cytotoxic reaction assay which comprises the steps of:

[0097] reacting both effector molecules which are the soluble Fas ligand prepared from the supernatant of culture of cells transfected with the Fas ligand gene, and target cells which are the cells transfected with the Fas gene, in a 100 μl reaction system on a 96-well plate; and 16 hours thereafter, determining the viable rate of the target cells by the usage of a reagent for counting vital cells.

[0098] In the case of the usage of either the aforementioned RNOK201, RNOK202 or RNOK203 as humanized immunoglobulins, the viable rate of the target cells, i.e., the apoptosis inhibition rate will be 90% or above, when the soluble Fas ligand contained in the supernatant of the cell culture transfected with the Fas ligand gene is used as an effector molecule whereas the cells transfected with the Fas gene (Fas/WR19L) are used as target cells, and 25 μl of the dilution of the effector molecule, 50 μl of 2×10⁵cells/ml liquid of the target cell, and 25 μl of the supernatant of the hybridoma culture containing any aforementioned immunoglobulin are all combined together, and the mixture is reacted at 37° C. for 17 hours.

[0099] The apoptosis inhibition rate of the humanized immunoglobulin of the invention is higher than the mouse immunoglobulin, which is the original donor antibody, at the same concentration as shown in FIG. 15. It is apparent that the humanized immunoglobulin of the present invention is a rare case of success, in the light of the fact that the humanized immunoglobulins previously reported have the greatly lowered activity compared to the original non-human immunoglobulins.

[0100] Further, the apoptosis-inhibitory activity of the present humanized immunoglobulin may be higher than the Fas-Ig chimeric molecule. This is because the mouse antibody, which is a donor antibody for the preparation of the humanized antibody of the present invention has been shown to exhibit the higher apoptosis-inhibitory activity at a concentration of 0.01-8 μg/ml (effective concentration) compared to the Fas-Ig chimeric molecule at the same concentration, as described in the PCT Application No. WO96/29350, therefore, it is apparent that the humanized antibody of the present invention, which has the higher apoptosis-inhibitory activity than the original mouse antibody exhibits the same or higher activity compared to the Fas-Ig chimeric molecule at a concentration of 0.01-8 μg/ml (i.e., effective concentration)

[0101] The humanized immunoglobulin of the present invention is not only useful for the immunological study but also for the immunological therapy and diagnosis. To accomplish these purposes, a whole immunoglobulin is not necessary, and a portion of the molecule can be used as long as the portion has the activity, which portion may be preferred in a certain case. This will be understood by those skilled in the art. Accordingly, the present invention encompasses the active fragments of the anti-Fas ligand immunoglobulins. Antibody is a homogeneous immunoglobulin which recognizes a specific antigenic substance. Active fragment means a fragment of the immunoglobulin showing the antigen-antibody reaction activity, and includes F(ab′)2, Fab′, Fab, Fv, and recombinant Fv. F(ab′)2 fragment is a fragment which is obtained by the digestion of immunoglobulin, IgG, with pepsin. Digestion of IgG with pepsin at a pH of near 4.0 causes the cleavage of the H chains at the hinges thereof to provide the fragment with a molecular weight of about 100,000. The cleavage occurs down to the C-terminal from the disulfide bonds between the H chains. This fragment has two binding sites, and therefore can cause precipitation and aggregation reactions. Fab′ fragment is a fragment with a molecular weight of about 50,000 which is obtained by the cleavage of the S—S bonds between the H chains by reducing F(ab′)2 fragment with a reagent such as 2-mercaptoethanol and alkylating the reduced material with monoiodoacetic acid.

[0102] Fas fragment (antigen-binding fragment) is a fragment which is obtained by the digestion of IgG with papain. Digestion of IgG with papain in the presence of cysteine causes the cleavage of the H chains at the site toward the N-terminal from the disulfide bonds between the H chains at the hinge so as to provide two Fabs and one Fc (i.e., crystallizable fragment). Fab fragment is a fragment with a molecular weight of 45,000 wherein the Fd fragment corresponding to the half of the H chain at the N-terminal side (i.e. VH domain+CH1 domain) and the L chain are combined via disulfide bond. Fab fragment has one antigen-binding site. Fv fragment is an antigen-binding fragment which consists of the variable region of the H chain (VH) combined with the variable region of the L chain (LH) via non-covalent bond.

[0103] Recombinant Fv can be obtained by determining the base sequences encoding the VH and LH of the DNA from the immunoglobulin-producing hybridoma and then introducing the DNA fragments into a vector so as to prepare a monovalent active fragment of antibody having the structure: VL-linker-VH. The VH and the LH of IgG, Fab or F(ab′)2 are combined together via S—S bond, whereas the recombinant Fv fragment has the linker inserted between the VH and LH to assume the stereochemistry similar to S—S bond. The latter fragment is also referred to as merely Fv, or scFv (single-chain Fv). The recombinant Fv also may be expressed in a microorganism such as E. coli, or bacteriophage.

[0104] The active fragments can be used alone, or if necessary, can be used as a new complex in which the fragments are conjugated with an agent such as albumin, and polyethylene glycol. In general, such complex often exert its effects to maximum capacity without decomposition for a prolonged period of time. The method for conjugating the active fragment with the agent such as albumin, and polyethylene glycol has been described in, for example, Antibodies, A. Laboratory Manual, Cold Spring Harbor Laboratories, 1988. Generally, divalent reactive reagents such as SPDP (Pharmasia) can be used to conjugate readily the active fragment with albumin or the like.

[0105] The immunoglobulin and the pharmaceutical composition thereof of the present invention are especially suitable for parenteral administration such as subcutaneous, intramuscular or intravenous administration, and are usually dissolved in an acceptable carrier, preferably in an aqueous carrier. For example, water, buffer, phosphate buffered saline (PBS), 0.4% physiological salt solution, 0.3% glycine, human albumin solution, or the like may be used as the aqueous carrier. The solutions are sterile, and generally contain no particulate. The composition is sterilized according to the conventional, well known methods for sterilization. If necessary, the composition may contain a pharmaceutically acceptable addictive, for example, a pH modulator and a buffering agent such as sodium acetate, sodium chloride, potassium chloride, calcium chloride, and sodium citrate for adapting to the physiological condition. The concentration of the immunoglobulin in the composition may be changed broadly, i.e., in the range from about 1 to 20% by weight, and is mainly determined on the basis of volume, viscosity and the like of the composition, depending on the choice of the type of administration.

[0106] The immunoglobulin of the present invention is frozen or lyophilized for storage, if necessary, and is reconstituted in a suitable dissoluble liquid prior to use.

[0107] Many Fas ligand variants in which an amino acid substitution is introduced into the amino acid sequence of Fas ligand can be used to identify the recognition site on Fas ligand which is bound to an anti-Fas ligand antibody. The inventors prepared 45 Fas ligand variants, each of which is introduced with one amino acid substitution into the region which is predicted to be recognized by NOK2 immunoglobulin, which may be referred to as NOK2 antibody, described in the PCT Application No. WO96/29350, said region encompassing 45 amino acids including 14 amino acids spanning Leu at position 207 to Gln at position 220 numbered from the N-terminus of Fas ligand, wherein the substitution is that the amino acids other than Ala are replaced by Ala, and Ala is replaced by Gly, and wherein the numerical order of amino acids is in accordance with Nagata, et al. Int. Immunology, vol. 6, p.1567-1574, 1994, unless otherwise indicated. Then, the Fas ligand variants were examined for the binding activity to the humanized NOK2 antibodies of the present invention as well as anti-Fas ligand antibodies (which represent NOK1 antibody, NOK2 antibody, NOK3 antibody, and NOK4 antibody, and generically may be referred to as NOK antibody) which have the high inhibitory activity to apoptosis occurred via the interaction of Fas-Fas ligand as described in the above application.

[0108] Consequently, it has been shown that the aforementioned NOK and humanized NOK2 antibodies can be bound to any position in the amino acid sequence included in the region spanning Arg at position 198 to Met at position 238 of Fas ligand. Further, with respect to NOK1, NOK2 and NOK4 antibodies, the mapping of the recognition site using the synthetic peptides revealed again that the aforementioned site of Fas ligand contains any amino acids which play an important role for NOK1, NOK2 and NOK4 antibodies to recognize and/or bind to the antigen.

[0109] The aforementioned NOK antibody is monoclonal antibodies which have been prepared by immunizing mouse with Fas ligand-expressing cell (see Example 1 at the detailed explanation of the invention in the specification of the PCT Application No. WO96/29350). Further, the aforementioned NOK antibody and humanized NOK2 antibody have the strong inhibitory activity to apoptosis induced in Fas-expressing cells by the interaction of Fas-Fas ligand (see Examples 1 and 2 at the detailed explanation of the invention in the specification of the Japanese Patent Application No. 303492/1995, and Example 4 at the detailed explanation of the invention in the specification of the present application). In other words, NOK antibody is the first anti-Fas ligand monoclonal antibodies which have been prepared by immunizing with Fas ligand having the native structure and which have the high inhibitory activity to apoptosis. These monoclonal antibodies are different from each other in terms of amino acid sequence of the CDR (i.e., complementarity determining region) and the class and subclass of antibody, and therefore, they are the independent antibodies from each other.

[0110] As described above, the site recognized by the different antibodies occupies a broad region which is composed of the amino acid sequence spanning Arg at position 198 to Met at position 328 of Fas ligand, and therefore, it can be concluded that the anti-Fas ligand antibodies having the high apoptosis-inhibitory activity should generally recognize the amino acids included in the aforementioned region as a recognition site. Thus, the aforementioned site, which resides on the Fas ligand molecule should arise the anti-Fas ligand antibody having the high apoptosis-inhibitory activity. This suggests that the aforementioned site is an important region for Fas ligand to exert the apoptosis-inducing activity, and an agent which can recognize and bind to this site would have apoptosis-inhibitory activity.

[0111] Features of the site to be bound to the aforementioned antibody having the high apoptosis-inhibitory activity can be demonstrated exhaustively when the model for the stereochemistry of Fas ligand molecule constructed using the commercially available software for molecular design is adapted to the data obtained in the analysis of the recognition site using the aforementioned Fas ligand variants and synthetic peptides.

[0112] Modeling of the monomer of Fas ligand molecule can be performed by using Modeler, which is a software for molecular design according to the manufacture's instructions utilizing as a template an X-ray crystal structure data of TNF-α (PDB ID:1TNF) and TNF-β (PDB ID:1TNR) having the high homology to Fas ligand, which have been recorded on Brookhaven Protein Data Bank (PDB). Model for the Fas ligand trimer can be constructed by superimposing the above Fas ligand monomer model on each segment of the TNF monomer on the atomic coordinates of PDB ID:1TNF, TNF-α trimer, using the software for molecular design, QUANTA/CHARMm (produced by Molecular Simulations).

[0113] Although Modeler and QUANTA/CHARMm (both are produced by Molecular Simulations) are used herein as a software for the molecular design, they are not exclusive, and any other software produced by any manufactures also can be used which is available as a software for general proteins. Further, TNF-α and -β are used herein as a template protein, but they are not exclusive, and any other protein which belongs to the TNF family also can be used.

[0114] Superimposition of the site to be bound to NOK and humanized NOK2 antibodies on the model obtained as shown above has revealed surprisingly that the antibodies bind to a broad region striding the two molecules of Fas ligand, which appears after the Fas ligand trimer is formed. Such finding with respect to the relationship between Fas ligand and anti-Fas ligand antibody is novel and has not been reported so far.

[0115] In general, it has been reported that the antigen-binding region in antibody is a region including six sites referred to as CDR (complementarity determining region), and the size of the region is in the range of from about 700 angstrom square (Sasatsuki, Tatehiko ed., Immunobiology, Nankodo, p. 128, 1995) to about 900 angstrom square (Tulip W. R., et al., J. Mol. Biol., vol. 227, p.122-148, 1992) for protein antigens with normal mass although the size varies depending on the particular antigen to be bound. The Fas ligand molecule usually occurs on the cell surface as a transmembrane type of protein with about 40 kD, and further, the molecule also may occurs as a free type of molecule with about 27 kD owing to the processing of the extracellular domain of the molecule. Additionally, it has been reported that the three molecules of Fas ligand monomer are associated with together to form a Fas ligand trimer. The protein is so large that the antibody comes into contact with the antigen surface-to-surface using the whole of CDR (Sasatsuki, Tatehiko ed., Immunobiology, Nankodo, p.128, 1995).

[0116] Consequently, the site associated with the apoptosis-inhibitory activity which resides on the Fas ligand trimer, which has been discovered by the present inventors can be illustrated as a plane. In general, the plane may be depicted as a flat plane defined by three points A (XA, YA, ZA), B (XB, YB, ZB), and C (XC, YC, ZC) which are not on a straight line.

[0117] The plane of the antigen defined on the Fas ligand trimer model may be estimated by assigning to the expression the data for the atomic coordinate of the amino acids recognized by antibodies, which is obtained from the Fas ligand trimer model. Any three amino acids which are optionally selected from the amino acids recognized by the antibodies are required to define such plane, and the preferred embodiment comprises Gln at position 200 which is on one of two adjacent Fas ligand molecules, and Asn at position 203 and Gln at position 220 which are on the other Fas ligand molecule. The plane defined by the combination of atomic coordinates of the α-carbons (Cα) of these amino acids can cover the region composed of the amino acids recognized by the antibodies. The effective plane which comes into contact with antibodies or the plane through which atoms of the antibodies approach to the Fas ligand trimer is believed to be a plane which resides opposite to the Fas ligand molecule being apart from the atomic coordinates of the Cα atoms by the length of the side chains of the amino acid residues plus about 4.1 Å of the cut off value of the common van der Waals contact.

[0118] Further, the size of the CDR of anti-Fas ligand antibody can lead to demonstration of amino acids which compose the region on the Fas ligand trimer, which is associated with apoptosis-inhibitory activity. It is believed that the CDR of anti-Fas ligand antibody is not different to some extent from the common antibodies in terms of size in the light of the length of amino acid sequence of the variable region. Stereochemistry model of NOK2 antibody is used as the representative of the anti-Fas ligand antibodies to demonstrate the size of the CDR of anti-Fas ligand antibodies, but any other NOK and humanized NOK2 antibodies can be used to demonstrate the size since the numbers of amino acid in the CDR of these antibodies are not different among them. Further, mere determination of the size of CDR will not be limited to usage of only these antibodies.

[0119] Investigation of the CDR of NOK2 antibody revealed that the CDR has substantially the same size as the circle with about 17 Å (angstrom) radius. Comparison between the circle with about 17 Å radius and the region composed of the amino acids on the Fas ligand trimer model, which are recognized by the antibodies has shown that the amino acids on the Fas ligand timer, which are recognized by NOK and humanized NOK2 antibodies are entirely covered within the circle with about 17 Å radius. This means that the anti-Fas ligand antibody having the high apoptosis-inhibitory activity in common recognizes the region composed of the amino acids encompassed in the circle, and in other words, the amino acids of the Fas ligand trimer within the circle could interact with the CDR amino acids of anti-Fas ligand antibody. A candidate for the amino acids which can interact with the CDR amino acids of the anti-Fas ligand antibody may be any one of amino acids which are within the circle, and are exposed on the Fas ligand trimer, and preferred amino acid is one selected from a group consisting of the amino acids, each of which is exposed on the surface of the Fas ligand trimer, and has an exposed area with 15 angstrom square or above. In particular, preferred amino acids include Ser at position 153, Tyr at position 166, Ile at position 168, Arg at position 198, Gly at position 199, Gln at position 200, Gln at position 237, Met at position 238, Arg at position 241, Ser at position 242, Phe at position 269, Glu at position 270, Glu at position 271, and Ser at position 272 of one of adjacent two Fas ligand molecules, as well as Ser at position 157, Met at position 158, Pro at position 159, Glu at position 161, Lys at position 178, Gly at position 179, Asn at position 203, Asn at position 204, Leu at position 205, Pro at position 206, Ser at position 208, Lys at position 210, Tyr at position 212, Arg at position 214, Tyr at position 218, Pro at position 219, Gln at position 220, Asp at position 221, Leu at position 222, Val at position 223, Lys at position 228, Met at position 230, Tyr at position 232, His at position 256, Tyr at position 258, Asn at position 260, Ser at position 262, Glu at position 263, Leu at position 264, and Ser at position 265 of the other Fas ligand molecule. Interaction with the amino acids encompassed in the region on the Fas ligand trimer is important for proteins, peptides, especially antibodies or analogues thereof to show the high apoptosis-inhibitory activity.

[0120] Agents which exert the apoptosis-inhibitory activity according to the present invention include, in addition to the humanized immunoglobulins which are specifically reactive to Fas ligand, substantially any agents which can specifically recognize and are able to be reactive to the aforementioned domain, region or the amino acids thereof which are on the Fas ligand and are important for the apoptosis-inhibitory activity. Particularly, agents which recognize and bind to the defined amino acid on the protein molecule such as immunoglobulin are preferable, and preferred agents include a fragment derived from immunoglobulin or an analogue thereof, and a receptor altered to be able to recognize the aforementioned region, and enzymes which recognize and react with the defined amino acids, and the like. These proteinaceous agents may be prepared by genetic engineering techniques, and therefore, they can be referred to as recombinant proteins. Additionally, there can be exemplified as a candidate for such agents, by a peptide synthesized based on the amino acid sequence derived from the above proteinaceous agents, a peptide having an absolutely novel sequence, which is obtained by the screening such as phage display system for its ability to recognize the domain, region or amino acids which are on Fas ligand and are important for the apoptosis-inhibitory activity from a peptide library consisting of random sequences, and a peptide-like substance which is composed of a non-naturally occurring synthetic peptide prepared based on these peptides.

[0121] As described above, recent investigations have shown that the physiological reaction between Fas and Fas ligand is associated with diverse diseases such as AIDS, graft-versus-host diseases in bone marrow transplantation (GVHD), fulminating hepatitis, autoimmune diseases (SLE, RA), and therefore, the humanized immunoglobulin of the present invention, which has ability to inhibit such physiological reactions can be used as a therapeutical agent for treating the above diseases.

[0122] Further, the site on Fas ligand required to induce apoptosis, which has been found for the first time by the present inventors is also significant to elicit anti-Fas ligand antibody having the high apoptosis-inhibitory activity. This finding afford the important information to design various other molecules for inhibiting apoptosis. Thus, the finding may be utilized to find a new therapeutic or diagnostic agent by creating a recombinant protein or peptide which is specifically reactive to the amino acids within the aforementioned site to inhibit apoptosis.

[0123] The following examples are presented for purpose of further illustration of the invention, and such examples are not intended to be limiting the invention in any respect.

[0124] NOK antibody described in Examples herein is included in the invention of the PCT Application No. WO96/29350 which is incorporated herein by this reference. The preparation is detailed in the application, and the outline is provided below.

[0125] First, the human Fas ligand gene was amplified by PCR with the primers prepared on the basis of the report of Nagata, et al., Cell, vol.75, p.1169-1178, 1993, using as a template the cDNA prepared conventionally from human killer T cells expressing Fas ligand. The restriction site which had been incorporated into the primers was utilized to ligate the amplified genes into the expression vector BCMGSNeo (Karasuyama, Hajime, Experimental Medicine, Special number, Handbook for Genetic Engineering, Yodo Cor., p.297-299, 1992), providing the expression plasmid for Fas ligand. The expression plasmid was transformed into COS cell (ATCC CRL1650) by conventional DEAE-dextran method to provide the COS transformants for Fas ligand. The transformants were used to challenge MRL lpr/lpr mice intra-abdominally three times at intervals of a week, and the spleen was excised three days after the final challenge. The splenocytes from the excised spleen were fused with 8-azaguanine-resistant cells from mice, P3X63Ag8.653 (ATCC CRL1580), by the conventional method using polyethylene glycol. The immunoglobulins which were contained in the cultured supernatants of clones from the hybridomas cloned by limiting dilution was screened for the apoptosis-inhibitory activity by the determination as shown below, and the hybridoma producing monoclonal antibody NOK2 which were specifically reactive to Fas ligand was obtained, which had been deposited at the National Institute of Bioscience and Human Technology, Ministry of International Trade and Industry under Accession No. FERM BP-5045.

[0126] Similarly, NOK1, NOK3, and NOK4 antibodies were obtained.

[0127] 1-1) Isolation of the Gene of the Variable Region of NOK2 Antibody

[0128] Isolation of the gene of the variable region (V region) of the mouse immunoglobulin was performed as shown below. First, total RNA was extracted from the NOK2-producing hybridoma using ISOGEN (trade name), RNA extraction reagent produced by Nippon Gene. The procedures were performed according to the accompanied protocol. The hybridoma producing NOK2 antibody was the same as FERM BP-5045 deposited at the National Institute of Bioscience and Human Technology, Ministry of International Trade and Industry. Second, the mRNAs were prepared from the total RNA using POLY (A) QUIK mRNA ISOLATION kit (trade name) produced by STRATAGENE. The procedures were performed according to the accompanied protocol. Subsequently, single-stranded cDNAs were synthesized from the mRNAs as templates using First Strand cDNA Synthesis kit (trade name) of Pharmasia Biotech. The oligo (dT) primers accompanied therein were used as primers, and the procedures were performed according to the protocol accompanied therein. Polymerase chain reaction (PCR) was performed using the resultant single-stranded cDNAs as templates with DNA primers (the heavy chain/MHL4.4 primer: SEQ ID No: 10, MHJ124 primer: SEQ ID No: 11; the light chain/MKL2.4 primer: SEQ ID No: 12, MKJ124 primer: SEQ ID No: 13), which had been prepared on the basis of the base sequences of V regions and J regions classified by Kabat, et al., Sequence of Proteins of Immunological Interest, 4th ed., Public Health Service, NIH, Washington D.C., 1987. Both V region primers and J region primers contain the HindIII and BamHI sites.

[0129] PCR was performed using the kit of PERKIN ELMER according to the accompanied protocol. The condition of PCR comprises 35 cycles consisting of 94° C. for one minute, 60° C. for two minutes, and 72° C. for two minutes. After the PCR, the DNA segments were cloned into pCRII vector (trade name) of Invitrogen. The procedures were performed according to the accompanied protocol.

[0130] 1-2) Base Sequence of the Gene of the V Region of Mouse NOK2 Antibody

[0131] The reaction product was applied to an autosequencer to determine the sequence of the V region gene incorporated into pCRII by using Dye Primer Cycle Sequencing kit (trade name) of Perkin-Elmer. The base sequences of the variable regions of the heavy chain (VH) and of the light chain (VL) of NOK2 antibody are shown in FIGS. 1 and 2 (VH: SEQ ID No: 14, VL: SEQ ID No: 15). Further, the amino acid sequences derived from the base sequences (VH: SEQ ID No: 16, VL: SEQ ID No: 17) are also shown in FIGS. 1 and 2. Both base sequences were indicated to contain a reconstitution which is unique in the V region gene, and to form an open reading frame (ORF).

[0132] 1-3) Construction of the Gene of Chimeric NOK2 Antibody (CNOK2H, CNOK2L)

[0133] Mouse-human chimeric antibody was prepared in order to determine if the isolated gene of the V region of NOK2 antibody encodes the V region which is responsible for the anti-Fas ligand activity. For the expression of the chimeric antibody, expression vectors PCAG-κ and pCAG-γ1 were used, which contain a promoter from the chicken β-actin gene, a splice acceptor sequence from the rabbit β-globin gene, and an enhancer from cytomegalovirus.

[0134] pCAG-κ had been prepared according to the usual genetic engineering procedures by ligating together the constant region gene of the human immunoglobulin κ chain and the polyadenylation site from RHC25 plasmid, an expression plasmid for the light chain of RC25, a humanized anti-HIV neutralizing monoclonal antibody described in the International Publication No. WO94/20632, the dhfr gene from pSV2-dhfr plasmid (Lee, F. et al., Nature, vol. 294, p.228-232, 1981) as a selectable marker suitable for eukaryotic cells, the SV40 promoter and polyadenylation site, the Ampr gene as a selectable marker suitable for prokaryotic cells described in the Japanese Patent Publication (kokai) No. 168087/1991, the enhancer from cytomegalovirus (CMV), the promoter from the chicken β-actin gene, and the splice acceptor site from the rabbit β-globin gene. The vector plasmid contains the HindIII site incorporated downstream from the splice acceptor site of the rabbit β-globin gene, and the BamHI site incorporated upstream from the constant region gene of the human immunoglobulin κ chain.

[0135] pCAG-γ1 had been prepared according to the usual genetic engineering procedure by ligating together the γ1 constant region gene of the human immunoglobulin and the polyadenylation site from HCMV-VH0.5β-γ1 plasmid described in Maeda, et al., Hum. Antibod. Hybridomas, vol.2, p124-134, 1991, the neor gene from pAd.RE.noe plasmid described in the Japanese Patent Publication (kokai) No. 5890/1990 as a selectable marker suitable for eukaryotic cells, the SV40 promoter and polyadenylation site, the Ampr gene as a selectable marker suitable for prokaryotic cells described in the Japanese Patent Publication (kokai) No. 168087/1991, the enhancer from cytomegalovirus (CMV), the promoter from the chicken β-actin gene, and the splice acceptor site from the rabbit β-globin gene. The vector plasmid contains the HindIII site incorporated downstream from the splice acceptor site of the rabbit β-globin gene, and the BamHI site incorporated upstream from the γ1 constant region gene of the human immunoglobulin.

[0136] The V region gene of NOK2 antibody prepared as described in 1-2) was digested with the restriction enzymes HindIII and BamHI (both are from Takara, and all of restriction enzymes used hereinafter are from Takara, unless otherwise stated), and the segments of the VH and the VL were incorporated into the HindIII-BamHI sites of pCAG-γ1 and pCAG-κ, respectively (CNOK2H and CNOK2L; FIG. 3).

[0137] 1-4) Expression of Chimeric NOK2 Antibody (CNOK2)

[0138] The plasmid DNAs, CNOK2H and CNOK2L prepared as described above were transformed into CHO-DG44 cell (Chasin, L. A., et al., Somatic Cell. Mol. Genet., vol. 12, p.555-566, 1986) using Lipofect ACE (trade name) from Gibco BRL to provide the transformants producing chimeric NOK2 antibody (CNOK2). The procedures were performed substantially according to the protocol accompanied in Lipofect ACE. Briefly, CHO-DG44 cells were transformed with a mixture which comprised Lipofect ACE and the PvuI-linearized products of both plasmid DNA 1 μg prepared using a kit of QIAGEN, and the transformants were cultured on an α-MEM medium (Gibco BRL) without nucleotides containing 1 mM G418 (Gibco BRL) and 10% dialyzed fetal bovine serum (Gibco BRL) in a condition of 5%CO₂at 37° C. for 2 weeks. The cells cotransformed with CNOK2H and CNOK2L plasmid DNAs were grown in the conditions. The resultant transformants were subculured and expanded on the above medium until reached confluent in a 75 cm²culture flask (CORNING), at this time, the medium was replaced by 20 ml of an ASF medium (trade name) (Ajinomoto Inc.), and the cells were cultured in a condition of 5%CO₂at 37° C. for seven days, before recovering the supernatant.

[0139] 1-5) Concentration of the Cultured Supernatant and Quantitative Analysis for CNOK2 Antibody Therein

[0140] The ASF cultured supernatant was concentrated 10-fold using Centricon-10 Spine Column (trade name) produced by Amicon. The chimeric antibody included in the concentrated supernatant was quantified by ELISA assay. In detail, 2 μg/ml of a goat anti-human IgG (Fc) antibody (Cappel) was first added to a 96-well Maxisorp Plate (trade name) produced by InterMed at 50 μl/well. After incubation overnight at 40C, the plate was rinsed three times in 0.01M PBS with 0.05% Tween20. Then, PBS with 1% BSA was added at 100 μl/well, and the plate was incubated at 37° C. for three hours. The plate was again rinsed three times in 0.01M PBS with 0.05% Tween20, then the aforementioned concentration of the cultured supernatant was added thereto at 50 μl/well, and the plate was incubated at 37° C. for an hour. Then, the plate was rinsed three times in 0.01M PBS with 0.05% Tween20, and the 5,000-fold dilution of the HRP-labeled anti-human Ig-Cκ antibody (produced by Southern Biotechnology Associate) in PBS with 1% BSA as a secondary antibody was added at 50 μl/well. The plate was incubated at 37° C. for an hour, and was rinsed five times in 0.01M PBS with 0.05% Tween20. Finally, a solution of chromophoric substrate (0.5 mM TMBZ+hydrogen peroxide) was added at 50 μl/well to allow the reaction, and when a suitable development was obtained, the reaction was quenched with the addition of 0.3N sulfuric acid at 50 μl/well, followed by determining an absorbance at a wavelength of 450 nm using Microplate Reader (Molecular Devices). Concentration of RNOK2 was determined on the basis of the calibration curve created for the standard human IgG with known concentrations in accordance with a similar procedure. As the standard human IgG, a purified product (purity: 98% or above) of RC25 antibody, anti-HIV neutralizing monoclonal antibody having the same Fc and Cκ as chimeric NOK2 antibody (the International Publication No. WO94/20632) was used.

[0141] 1-6) Preparation of Soluble Fas Ligand Molecule

[0142] The procedures of isolation of the Fas ligand gene, creation of the Fas ligand-expressing cells, and preparation of the soluble Fas ligand molecule are the same as those described in the PCT Application No. WO96/29350. Outline is provided below.

[0143] First, cDNAs were prepared from the mRNAs extracted from the human killer T cells which express human Fas ligand according to the conventional method. Second, PCR was performed using the cDNAs as templates with 5′ and 3′ primers each of which was incorporated with XhoI and NotI sites, respectively, described in Nagata, et al., Int. Immunol., vol. 6, No. 10, p.1567-1574, 1994, so as to provide the gene of human Fas ligand as an amplified product. The human Fas ligand gene was incorporated into the XhoI site and the NotI site of the expression vector BCMGSNeo (Karasuyama, hajime, Experimental Medicine, Special number, Handbook for Genetic Engineering, Yodo Cor., p.297-299, 1992), to provide human Fas ligand-BCMGSNeo, an expression plasmid for human Fas ligand. Subsequently, the plasmid was amplified in E. coli, and recovered, and the amplified plasmids were transformed into COS cells (ATCC CRL1650) by DEAE-dextran method (Experimental Medicine, Special number, Biomanual, series vol. 4, Yodo Cor., p.16-22, 1994) to provide the COS cells expressing Fas ligand. Then, the Fas ligand molecules released in the supernatant from the culture of the Fas ligand-expressing COS cells on a 10% FCS-DME medium were purified by an affinity chromatography on NOK1 antibody, the anti-Fas ligand antibody described in the PCT Application No. WO96/29350 which was produced by the hybridoma assigned with the accession No. FERM BP-5044, and sterilized by passage through 0.45 μm membrane filter so as to provide the soluble Fas ligand molecules. Observation of a single band on SDS-PAGE revealed that the Fas ligand was entirely purified. Concentration of the Fas ligand was determined by the absorbance at 280 nm.

[0144] 1-7) Affirmation of Binding of CNOK2 Antibody Against the Soluble Fas Ligand

[0145] Activity of the chimeric antibody included in the concentrated supernatant was quantified by ELISA using the soluble Fas ligand and anti-human IgG. Specifically, a 50 ng/ml solution of the soluble Fas ligand prepared according to the procedure of 1-6) was first added to a 96-well Maxisorp Plate (trade name) produced by InterMed at 50 μl/well. After incubation overnight at 4° C., the plate was rinsed three times in 0.01M PBS with 0.05% Tween20. Then, PBS with 1% BSA was added at 100 μl/well, and the plate was incubated at 37° C. for two hours. The plate was again rinsed three times in 0.01M PBS with 0.05% Tween20, then the concentration of the cultured supernatant prepared as described in the aforementioned 1-4) was added thereto at 50 μl/well, and the plate was incubated at 37° C. for two hours. Then, the plate was rinsed three times in 0.01M PBS with 0.05% Tween20, and the 5,000-fold dilution of the HRP-labeled anti-human Ig-Cκ antibody (produced by Southern Biotechnology Associate) in PBS with 1% BSA as a secondary antibody was added at 50 μl/well. The plate was incubated at 37° C. for an hour, and was rinsed five times in 0.01M PBS with 0.05% Tween20. Finally, a solution of chromophoric substrate (0.5 mM TMBZ+hydrogen peroxide) was added at 50 μl/well to allow the reaction, and when a suitable development was obtained, the reaction was quenched with the addition of 0.3N sulfuric acid at 50 μl/well, followed by determining an absorbance at a wavelength of 450 nm using Microplate Reader (produced by Molecular Devices). The expressed product by cotransformation with CNOK2H and CNOK2L plasmid DNAs was bound to Fas ligand depending on the concentration, suggesting that the isolated gene for the V region of NOK2 antibody certainly encodes the V region of an antibody having anti-Fas ligand activity (FIG. 4).

[0146] 2-1) Transplantation of the CDR of V Region of NOK2 Antibody Via PCR Mutagenesis

[0147] The CDRs (complementarity determining regions, SEQ ID Nos: 1-6) in the VH and VL of the cloned NOK2 antibody were each transplanted into the VH and VL regions of human immunoglobulin. The procedures were performed according to the preparation of humanized immunoglobulins (the Japanese Patent Publication (kokai) No. 141095/1992). The CDR in the VH region of NOK antibody was transplanted into SGI, which is the VH region containing the FR (framework region) of human subgroup II (distributed by Dr. Bendig of MRC Collaborative Center, United Kingdom). The CDR in the VL region of NOK2 antibody was transplanted into REI as previously reported, a VL region containing the FR of the human κ chain (W. Palm et al., Physiol. Chem., vol. 356, p167, 1975), and into the VL region containing the FR of the κ chain cloned from the cDNA library derived from human peripheral blood lymphocytes (huVL-19 and huVL-31).

[0148] In the humanization of NOK2 antibody, the putative three-dimensional structure of NOK2 antibody was first constructed using a computer modelling. Specifically, screening for the homology to Brookhaven Protein Data Bank (PDB) was performed on the amino acid sequences of the variable regions of the heavy chain and the light chain of NOK2 antibody to select the variable regions of the heavy chain (PDB ID:1FOR) and of the light chain (PDB ID:1TET) of the antibody, which would be used as templates. Then, modelling was performed using QUANTA/CHARMm which is a software running on Silicongraphics. The modelling steps were consisted of 1) copying the amino acid sequence of NOK2 antibody onto the atomic coordinate of the template of the aforementioned variable region amino acids, 2) energy-minimizing calculating (maximum gradient method), 3) chilling from 300K to 0K by taking 400 steps, 4) energy-minimizing calculating (maximum gradient method), 5) heating from 0K to 300K by taking 7,500 steps, 6) equilibrating by taking 30,000 steps (corresponding to 30 picoseconds), and 7) simulating by 20,000 steps (corresponding to 20 picoseconds). Next, one structure was extracted per 100 calculations to provide totally 200 structures, and they were divided into five groups using Cluster Analysis Program installed in the aforementioned software. Consequently, one structure obtained from the energy-minimizing calculation of the above 2), and five structures, each of which shows the minimum energy in the individual five groups as shown above were extracted to provide totally six structures and these six structures were assigned as an estimated three-dimensional structure of NOK2 antibody. Then, using the program installed in the same software, any amino acids within the FR which form the hydrogen bond to the CDR of the H chain and the L chain were selected in each of the aforementioned structures of NOK2 antibody (the second group), and further, any amino acids within the FR which form the hydrogen bond to them (the second group) were selected. Similarly, using the installed program, any amino acids within the FR which form the energy contact to the CDR were selected in the H chain and the L chain of NOK2 antibody (the first group), and further, any amino acids within the FR which form the energy contact to the resulting amino acids (the second group) were selected. The energy contact herein includes so-called electrostatic interaction and van der Waals forces. Initializations of the above program were used as parameters necessary for selection of any amino acids forming the hydrogen bond, and for the calculation of energy contact. Subsequently, from the FR amino acids of the first and second groups for the hydrogen bond, and the FR amino acids of the first and second groups for the energy contact, any FR amino acids which are included in FR amino acid clusters associated with four or more of the six estimated structures were selected as stereochemistry-responsible FR amino acids, and these were transplanted into the FR in the variable region of human immunoglobulin together with the CDR amino acids, provided that, when the transplantation of an FR amino acid among the selected FR amino acids into the relevant site in human immunoglobulin raises any sequences which can not be cited as the variable amino acid sequences of human immunoglobulin in the classification of Kabat, et al., Sequence of Proteins of Immunological Interest, 4th ed., Public Health Service, NIH, Washington D.C., 1987 and in a software for information retrieval, Entrez (trademark), which has been developed by National Center for Biotechnology Information: NCBI, then such FR amino acids were not be transplanted, so as to reduce the possibility to elicit the antigenicity as much as possible when the resulting humanized immunoglobulin is administered to human.

[0149] For humanized VH, two types of variants comprising two differences in FR3 amino acids were created (RNOK2VHver11: SEQ ID No: 18 and RNOK2VHver12: SEQ ID No: 19). For humanized VL, RNOK2VLver1 (SEQ ID No: 20) was created as a transplanted entity into the REI of the VL region of human immunoglobulin, whereas four types of variants (i.e., RNOK2VLver21: SEQ ID No: 21, RNOK2VLver22: SEQ ID No: 22, RNOK2VLver23: SEQ ID No: 23 and RNOK2VLver24: SEQ ID No: 24) comprising one or two differences in amino acids of the FR1 were created as a transplanted entity into huVL-19 and huVL-30 obtained from the cDNA library. So, totally five types of the variants were created for the VL.

[0150] The VH and VL amino acid sequences of NOK2 antibody, human immunoglobulin receiving transplantation, and humanized NOK2 antibody constructed are depicted in FIGS. 5 and 6. Large characters in the amino acid sequences represent the conservative sequence between NOK2 antibody and the human antibody receiving transplantation, whereas small characters represent the different amino acids from each other. The amino acids surrounded by box represent the FR amino acids to be transplanted together with the CDR amino acids, which were selected by the aforementioned procedure.

[0151] Actually, transplantation of the amino acids was performed at the genetic level by the PCR mutagenesis whereby mutations were introduced by PCR using the VH gene of RC25 antibody, a humanized anti-HIV immunoglobulin (the International Publication No. WO94/20632) as a template for the VH, and using the VL gene of chimeric NOK2 prepared as described previously as a template for the VL, so that the genes encoding the amino acid sequences of humanized NOK2 antibody as shown in FIGS. 5, 6 and 7 were constructed. FIGS. 8-13 show the annealing of the synthetic primers used in the PCR mutagenesis with the template VH and VL regions.

[0152] PCR was performed using a kit produced by Perkin-Elmer according to the accompanied protocol. The conditions of PCR comprise 30 cycles consisting of 95° C. for one minute, 60° C. for one minute, and 72° C. for two minutes.

[0153] In the case of the VH, the 5′ of the VH gene was amplified using primer pAGF (SEQ ID No: 25) and primer #01 (SEQ ID No: 26), and the 3′ of the VH gene was amplified using #02 (SEQ ID No: 27) or #03 (SEQ ID No: 28) and #04 (SEQ ID No: 29), provided that #02 was used for the preparation of RNOK2VHver11, and #03 was used for ver12, using plasmid RHC25 (the International publication No. WO94/20632) cloned with the variable gene of humanized C25 immunoglobulin as a template. Then, equal volumes of both amplified gene fragments obtained were combined together, and the combination was used as a template together with pAGF primer and #04 to perform PCR. The amplified gene fragments thus obtained were ligated into the KpnI site in the VH gene of humanized C25 antibody (the International Publication No. WO94/20632) to create RNOK2VHver11 and ver12, the VH genes of humanized NOK2 antibody.

[0154] In the case of VL, on the other hand, the 5′ of the VL gene was amplified using primer #05 (SEQ ID No: 30) and primer #06 (SEQ ID No: 31), the intermediate of the VL gene was amplified using #07 (SEQ ID No: 32) and #08 (SEQ ID No: 33), and the 3′ was amplified using #09 (SEQ ID No: 34) and #10 (SEQ ID No: 35), using the VL of chimeric NOK2 as a template. Then, equal volumes of three amplified gene fragments thus obtained were combined together, and the mixture was used as a template together with #05 and #10 to perform PCR. The amplified gene fragments thus obtained were ligated into the ApaI site in the leader region upstream from the VH gene of human immunoglobulin SGI to create RNOK2VLver1, the VL gene of humanized NOK2 antibody.

[0155] The remaining four variants of the VL were prepared in the following manner. First, PCR was performed with primer #11 (SEQ ID No: 36) and #12 (SEQ ID No: 37), using the VL of chimeric NOK2 as a template. Primer #11 comprises variations in the nucleic acid sequence to create the variants RNOK2VLver21, 22, 23, and 24. The amplified gene fragments thus obtained were ligated into the ApaI site in the leader region upstream from the VH gene of human immunoglobulin SGI to create RNOK2VLver21, 22, 23, and 24, the VL genes of humanized NOK2 antibody.

[0156] The base sequences of humanized VH and VL region genes constructed as described above were analyzed using a kit produced by Perkin-Elmer and an autosequencer in order to select a clone having the defined sequence.

[0157] The fragments of the humanized VH and VL region were digested with the restriction enzymes HindIII and BamHI, and the digested products were each introduced into the HindIII-BamHI site of pCAG-γ1 and pCAG-κ as described in the aforementioned preparation for the chimeric antibody (See Example 1) to provide the plasmids for expression of humanized NOK2 antibody genes, RHNOK2 and RLNOK2, which are generically referred to the variants of the humanized VH and VL, respectively.

[0158] 2-2) Expression of Humanized NOK2 (RNOK2) Immunoglobulin

[0159] For the preparation of transformants producing RNOK2 immunoglobulin, plasmid DNAs for expression of humanized NOK2 antibody genes, RHNOK2 and RLNOK2 were each introduced into CHO DG44 cells according to a procedure similar to the preparation of the chimeric antibody (See Example 1). This procedure was performed using Lipofect ACE as described in the case of chimeric antibody.

[0160] Since two types of the expression plasmids for the humanized variable region gene for the VH whereas five types for the VL were prepared as described in above, totally 10 types of humanized NOK2 immunoglobulin were expressed by the combination thereof. RNOK201, RNOK202, and RNOK203 described hereinafter are humanized NOK2 antibody which was each expressed in combinations of RNOK2VHver11 and RNOK2VLver1, RNOK2VHver11 and RNOK2VLver21, and RNOK2VHver11 and RNOK2VLver22, respectively. Hereinafter, these antibodies may be referred to as RNOK2 antibody(ies) or RNOK2 immunoglobulin(s).

[0161] After transformed with the expression plasmids, the transformants were cultured. ELISA of the cultured supernatants using the anti-human IgGγ1 and the anti-human IgGκ revealed that human immunoglobulin was produced in the cultured supernatants.

[0162] 3-1) Preparation of Purified RNOK2 Immunoglobulins

[0163] The cultured supernatants obtained by culturing RNOK2-expressing CHO-DG 44 cells prepared as described in Example 2 were recovered, and RNOK2 immunoglobulins included herein were purified. Specifically, the transformants expressing RNOK2 immunoglobulins obtained in Example 2 were first cultured on an α-MEM medium (Gibco BRL) without nucleotides containing 1 mM G418 (Gibco BRL) and 10% dialyzed fetal bovine serum (Gibco BRL) in a condition of 5%CO₂at 37° C. in a 75 cm²culture flask (CORNING) until reached confluent. Then, the culture was expanded to two 225 cm²culture flasks (CORNING), and again cultured in the same conditions until reached confluent. The two cultures were expanded to eight 225 cm²culture flasks, and the medium was replaced by 65 ml of an ASF medium (trade name) (Ajinomoto Inc.) at the point when they reached confluent, followed by culturing in a condition of 5%CO₂at 37° C. for seven days, before recovering totally about 500 ml of the supernatant.

[0164] From the cultured supernatant, only the protein G-adsorbed IgG was isolated using a protein G column (produced by Pharmasia Biotech) and a TPLC system (produced by Pharmasia Biotech). The concentration of the purified immunoglobulin was determined according to ELISA method described in Example 1. Further, SDS-PAGE electrophoresis was performed on the purified immunoglobulin in the reductive conditions, followed by CBB stain, observing no band of contaminant proteins but bands assuming to be the heavy chain and the light chain.

[0165] 3-2) Binding Activity of RNOK2 to Fas Ligand

[0166] The binding activities of RNOK201, RNOK202, and RNOK203 immunoglobulins among the purified RNOK immunoglobulins prepared as described in the above to Fas ligand were examined by ELISA method using the soluble Fas ligand and the anti-human IgG (Example 1), to confirm that RNOK2 immunoglobulins bind to Fas ligand in a concentration-dependent manner (FIG. 14).

[0167] 4-1) Preparation of Soluble Fas Ligand Molecule

[0168] Soluble Fas ligand molecule was prepared according to the same procedure as in Example 1-6).

[0169] 4-2) Preparation of RNOK2 Immunoglobulin Solution

[0170] The RNOK2 immunoglobulins purified with the manner as described in Example 3 were diluted in a 10%FCS RPMI1640 medium to prepare 12 solutions having the different immunoglobulin concentrations which are provided below. One hundred μl aliquots of the solutions each having immunoglobulin concentrations of 4 μg/ml, 2 μg/ml, 1 μg/ml, 0.5 μg/ml, 0.25 μg/ml, 0.125 μg/ml, 0.0625 μg/ml, 0.03125 μg/ml, 0.01563 μg/ml, 0.007813 μg/ml, 0.003906 μg/ml, or 0.001953 μg/ml were prepared.

[0171] The ¼ portions (i.e. 25 μl) of the solutions were finally added to 100 μl reaction systems, and therefore, the effective concentrations would be ¼ values of the above concentrations.

[0172] 4-3) Preparation of Purified Mouse NOK2 Immunoglobulin

[0173] Each of Hybridoma NOK2, which had been deposited at the National Institute of Bioscience and Human Technology, Ministry of International Trade and Industry under Accession No. FERM BP-5045 was grown in a RPMI1640 medium with 10%FCS to 3×10⁷cells. Such preparation of the 3×10⁷cells was performed on a scale for cell culture by adding a 30 ml culture to a 75 cm²flask (CORNING). Specifically, starting from culturing at concentration of 2×10⁵cells/ml, the cells were recovered when they reached 1×10⁶cells/ml.

[0174] The recovered hybridomas were suspended in 1.5 ml of PBS, and the 0.5 ml portions corresponding to 1×10⁷cells of the suspensions were administered to nude mice intra-abdominally. After feeding for 10 days, the ascites deposited in the visceral cavity were recovered giving 6.7 ml/animal. The 10 ml portions of the ascites were purified.

[0175] Ten ml of a saturated solution of ammonium sulfate was added dropwise to the equal volume of the ascites such that the mixture was first purified on the salting-out with ammonium sulfate. After stirring at 4° C. for two hours, the mixture was centrifuged at 10,000 g for 15 minutes. The supernatant was removed, and the pellet was dissolved in 5 ml of PBS. Then, the solution was dialyzed overnight against 3 liters of PBS. Subsequently, the dialyzed sample was recovered, and then only the protein G-adsorbed IgG was isolated by FPLC system using protein G column (Pharmasia). The sample was further dialyzed overnight against PBS. On the following day, concentration and purity of the protein were determined.

[0176] The quantification of the protein was performed using a protein quantification reagent of BIO RAD. The procedure was in accordance with the protocol accompanied therein. SDS-PAGE electrophoresis was performed on the purified immunoglobulin in the reductive conditions, followed by CBB stain, observing no band of contaminant proteins but bands assuming to be the heavy chain and the light chain.

[0177] In accordance with a similar procedure to the aforementioned case of RNOK2 immunoglobulin, 12 solutions were prepared having the different concentrations of the purified NOK2 immunoglobulin thus obtained.

[0178] 4-4) Preparation of Target Cells

[0179] As a target cell, WR19L cells which had been transformed with human Fas gene were used. The transformation of WR19L cells (ATCC TIB52) with human Fas gene was performed according to the conventional method. Specifically, the procedure was in accordance with Okumura, et al., Proc. Natl. Acad. Sci. USA, vol. 91, No. 11, p4930-4934, 1994. The resultant Fas-WR19L cells were cultured, and adjusted to 2×10⁵cells/ml in a 10%FCS-RPMI medium.

[0180] 4-5) Determination of Apoptosis-inhibitory Activity of RNOK2 Immunoglobulin

[0181] First, the soluble Fas ligand molecules prepared in the above 4-1) were diluted at 18.5 ng/ml with a 10%FCS-DME medium. The 25 μl portion of the dilution was added to each well on a 96-well flat-bottomed plate. Then, each of the solutions with the various concentrations of RNOK2 immunoglobulins (RNOK201, RNOK202 and RNOK203) prepared in 4-2), and of mouse NOK2 immunoglobulin prepared in 4-3) was added to the every three wells of the plate at 25 μl/well. Then, the plate was incubated in 5%CO₂atmosphere at 37° C. for an hour. Subsequently, the Fas-WR19L cell suspension prepared in the above 4-(4) as a target cell was added thereto at 50 μl/well, and the plate was incubated in 5%CO₂atmosphere at 37° C. for 17 hours. Then, Alamer Blue purchased from COSMO BIO was added thereto at 10 μl/well, and the plate was further incubated in 5%CO₂atmosphere at 37° C. for four hours. Subsequently, fluorescent strength was assayed at a determination wavelength of 590 nm using a Fluorescent Microplate Reader, Fluoroscan II (trade name) (produced by Titertek) with an excitation wavelength of 544 nm. The fluorescent strength represents viable cell numbers.

[0182] For a control of 100% viable, only a 10% FCS-RPMI1640 medium (neither soluble Fas ligand, RNOK2 immunoglobulin, nor mouse NOK2 immunoglobulin) was added to 50 μl of Fas-WR19L cells (target cell) in the well at 50 μl/well, whereas for a control of apoptosis, 25 μl of a 10% FCS-RPMI1640 medium and 50 μl of Fas-WR19L cells as a target were added to 25 μl of the soluble Fas ligand. The results are shown in FIG. 15.

[0183] FIG. 15 shows that all of humanized NOK2 immunoglobulins (RNOK201, RNOK202 and RNOK203) inhibit apoptosis by 90% or more in the range of concentrations of 0.06 μg/ml (effective concentration) or above. This demonstrates that all of the humanized immunoglobulins of the present invention are able to inhibit the apoptosis-inducing activity of the soluble Fas ligand against the Fas-expressing cells. Further, it should be especially noted that the activity of the humanized immunoglobulins of the present invention is equal to or more than the original mouse NOK2 immunoglobulin.

[0184] Further, mouse NOK2 immunoglobulin has been shown to have the much higher apoptosis-inhibitory activity at the antibody concentration of 0.01-8 μg/ml (effective concentration) compared to the Fas-Ig chimeric molecule at the same concentration as described in the PCT Application No. WO96/29350. Accordingly, it is apparent that the humanized immunoglobulins of the present invention have the higher affinity to Fas ligand and are more effective than the mouse Fas-Ig.

[0185] Additionally, this shows that either of the humanized immunoglobulins of the present invention bind to Fas ligand in body preceding the bind between Fas ligand and Fas, and therefore, it can be readily predicted that the humanized immunoglobulins of the present invention could inhibit satisfactorily the physiological reactions between Fas and Fas ligand in body.

[0186] 5-1) Construction of Expression Vector for Fas Ligand

[0187] In order to construct an expression vector for Fas ligand, the gene of the variable region of the light chain containing the leader sequence of C25 antibody described in the International Publication No. WO94/20632 was first excised with HindIII and BamHI, and ligated into the HindIII-BamHI site of the pCAG expression vector which was defective in the gene of the constant region of the κ chain in pCAG-κ expression vector described in Example 1-3). The plasmid was transformed into E. coliaccording to the conventional procedure, and the plasmid was recovered from the cultured transformants by the small scale plasmid preparation (Cold Spring Harbor Lab., Molecular Cloning, p.1.25, 1989). Subsequently, the leader sequence in part and the whole gene of the variable region gene which resided between the KpnI site in the leader sequence and the BamHI site downstream from the variable region of the C25 light chain were excised from the plasmid to provide the pCAG expression vector which was supplemented with the leader sequence from the light chain of C25 antibody.

[0188] 5-2) Creation of Fas Ligand Gene Variants by PCR Mutagenesis and Construction of Expression Plasmids Thereof

[0189] In order to identify a recognition site in the amino acid sequence of Fas ligand, which is a region to be bound by immunoglobulins having the high apoptosis-inhibitory activity, many variants having substitution of each one of the amino acid sequence in the extracellular domain of Fas ligand as shown in FIG. 16 with underline by Ala or Gly were prepared wherein the substitution in the amino acid of Fas ligand is that the amino acids other than Ala were replaced by Ala, and Ala is replaced by Glu. Hereinafter, the Fas ligands introduced with amino acid substitution are generically referred to as Fas ligand variants. Totally, 45 Fas ligand variants were prepared wherein the variant which has one amino acid substitution that the first Tyr of the underlined in FIG. 16 replaced by Ala is referred to as B1, afterwards, the variants prepared by the sequent replacement by Ala were referred to as B2, B3, . . . and the final variant prepared by the replacement of the last Ala by Gly was referred to as B45.

[0190] At first, PCR was performed with a sense primer, EP-02 primer (SEQ ID No: 38) and an anti-sense primer, EP-03 primer (SEQ ID No: 39), using as a template the Fas ligand gene of the expression vector (i.e., human Fas ligand-BCMGSNeo) described in Example 1-6), to provide the gene fragment of Fas ligand without amino acid substitution (hereinafter, referred to as native Fas ligand to avoid confusion).

[0191] PCR was performed using TAKARA Ex Taq (trade name) (Takara Co.), whereby preparing reaction solutions according to manufacture's instructions and carrying out the 30 cycles consisting of 95° C. for one minute, 60° C. for one minute, and 72° C. for two minutes. The amplified fragments were ligated into the pCAG expression vector described in 5-1) to provide the expression vector for native Fas ligand. Ep-02 primer contains the KpnI site for ligating the above PCAG expression vector, the remaining portion of the leader sequence gene of C25 antibody, the gene sequence encoding FLAG sequence described below, and the gene sequence encoding the sequence spanning the N-terminus to Ser at position 117.

[0192] FLAG sequence comprises eight amino acids, Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys. Tagging the sequence to a protein which is to be expressed facilitates to identify and quantify the intended protein using an antibody directed to the sequence. In the experiment, FLAG sequence was tagged at Phe of the third position from the N-terminus in the extracellular region of Fas ligand. EP-03 primer contains the gene sequence encoding the amino acid sequence downstream from Phe of the sixth position from the C-terminus of Fas ligand, the termination codon, and the BamHI site for ligating into the expression vector.

[0193] The genes encoding the Fas ligand variants were prepared by PCR mutagenesis. Specifically, the N-terminuses of the extracellular domain of Fas ligand variants were each amplified with a sense primer, MF primer (SEQ ID No: 40) corresponding to the sequence spanning from His at the fourth position to Gln at the 18th position from the N-terminus in the extracellular domain of Fas ligand, and with anti-sense primers, primers for variants introduced with the codon for Ala or Gly at the position to be substituted (the primers of variant Fas ligands B1-B45 are each described in SEQ ID Nos: 41-85). The MF primer sequence contains EcoNI site. Then, the C-terminuses of variant Fas ligands were each amplified using primers complementary to the aforementioned variant primers (the primers complementary to primers of SEQ ID Nos: 41-85 are each described in SEQ ID Nos: 86-130) as sense primers, and using an anti-sense primer, EP-03 primer. In order to prepare the 45 Fas ligand variants, 45 kinds of the aforementioned variant primers and the complement primers thereof per Fas ligand variant were synthesized. Next, PCR was again performed with EP-02 and EP-03 primers, using as templates mixtures containing equal volumes of both amplified gene fragments of the N- and C-terminuses of the Fas ligand variants. Such PCR was performed on each Fas ligand variant to provide the gene encoding the extracellular domain of each of the 45 Fas ligand variants, B1 to B45, which were introduced with the amino acid substitution in the underlined of FIG. 16.

[0194] The gene fragments of the Fas ligand variants thus obtained were inserted into the expression vector for the native Fas ligand at the EcoNI-BamHI portion in place of the native Fas ligand gene, to provide each of expression vectors for Fas ligand variants. When the vectors were transformed into host cells, and the transformants were cultured, then proteins tagged with FLAG sequence at Phe of the third position from the N-terminus of the extracellular domain of Fas ligand would be secreted into the cultured supernatants.

[0195] 5-3) Expressions of Fas Ligand Variants

[0196] The plasmids prepared as shown above were each transformed into E. coli, and the transformants were cultured. From the cultured transformants, the plasmids were recovered and purified using Plasmid Mini Kit (trade name) (produced by QIAGEN). Subsequently, the purified expression plasmids were each transformed into COS cells (ATCC CRL1650) and the transformants were cultured to provide the supernatants including the Fas ligand variants. Thus, a mixture of 1 μg of the above expression plasmid and Lipofect ACE was added to COS cells, and the cells were cultured overnight in a condition of 5%CO₂at 37° C. Then, 4 ml of an ASF medium (trade name) (produced by Ajinomoto Inc.) was added thereto, and the cells were cultured in a condition of 5%CO₂at 37° C. for four days, before recovering the supernatant. Additionally, the expression plasmid for the native Fas ligand was transformed into COS cells according to a similar procedure to provide the cultured supernatant. As a negative control, pCAG vector without the Fas ligand gene was transformed into COS cells according to a procedure similar to the above, and the transformants were cultured on the same medium for the same period of time before recovering the cultured supernatant.

[0197] 5-4) Relative Assay for the Fas Ligands in the Cultured Supernatant

[0198] Relative concentrations of the Fas ligand variant molecules in the cultured supernatants were determined by ELISA using as standard solution the cultured supernatant including the native Fas ligand prepared as described above. Specific procedures are as follows. First, serial twofold dilutions were performed up to seven times with an ASF medium on the cultured supernatants of the native Fas ligand, each of Fas ligand variant molecules, and the negative control, to prepare totally eight kinds of solutions per supernatant. Then, the solutions with various dilutions were each added to the eight wells in a lane of a 96-well Maxisorp plate (trade name) at 50 μl/well. The eight cultured supernatants including the Fas ligand variants were each added to the wells at the eight lanes of the 3th to 10th lanes from the left hand. Since lots of plates were used in this assay, the solutions including the native Fas ligand at various concentrations were each added to the eight wells at the second lane from the left-hand of the plates, and they were used as standard for determination of the relative concentration of the Fas ligand variant added to each plate. Further, in order to examine the non-specific chromophoric strength, the solutions of the supernatant of the negative control were each added to the eight wells at the 11th lane from the-left hand. Any end lane of the plate was not used.

[0199] Next, the plates were incubated overnight at 4° C., and then, the wells were rinsed three times in 0.01M PBS with 0.05% Tween20, followed by addition of PBS with 1% BSA at 100 μl/well. After incubation at 37° C. for an hour, the wells were rinsed three times in 0.01M PBS with 0.05% Tween20, and then 25 μg/ml anti-FLAGM2 mouse monoclonal antibody (trade name) (produced by Kodak) was added to each eight wells at 50 μl/well. After incubation at 37° C. for two hours, the wells were rinsed three times in 0.01M PBS with 0.05% Tween20, and then a 5,000-fold dilution of HRP-labeled anti-mouse Ig-Cκ antibody (produced by Southern Biotechonlogy Associate) in PBS with 1% BSA was added to each eight wells at 50 μl/well. Subsequently, the plates were incubated at 37° C. for an hour, and the wells were rinsed five times in 0.01M PBS with 0.05% Tween20, after which a solution of chromophoric substrate (0.5 mM TMBZ and hydrogen peroxide) was added at 50 μl/well, and when a suitable development was obtained, the reaction was quenched with the addition of 0.3N sulfuric acid at 50 μl/well, followed by determining an absorbance in each well at a wavelength of 450 nm using Microplate Reader (produced by Molecular Devices). In the wells including the cultured supernatants of the native Fas ligand and Fas ligand variant molecules, the developments were shown depending on the concentration of the anti-FLAGM2 antibody. No development was shown in the wells including the cultured supernatants of the negative control. The results in part are shown in FIG. 17. Consequently, it can be believed that the chromophoric strengths thus obtained (OD450 values) should be reflected to the concentrations of the native Fas ligand and Fas ligand variants tagged with FLAG sequence in the above cultured supernatants.

[0200] Next, relative volumes of the Fas ligand variants were determined using the chromophoric strength of the native Fas ligand as a standard. First, specific chromophoric strength in the well was determined by subtracting the OD₄₅₀values of the wells which include negative control from the OD₄₅₀values of the wells which include the native Fas ligand or the Fas ligand variant at the same dilution. Then, the OD₄₅₀, values were plotted against the dilutions of the cultured supernatant so as to select certain OD₄₅₀value from the range within which the OD₄₅₀values showed a linear decrease and every linear lines showed the same inclination, and the dilutions of the native Fas ligand and Fas ligand variant were determined from the plot against the selected OD₄₅₀value. Finally, relative dilution for the Fas ligand variant was evaluated on the basis of the dilution for the native Fas ligand to get relative concentration for the Fas ligand variant to the native Fas ligand. Such evaluation was performed on each plate to provide the relative concentrations for all Fas ligand variants to the native Fas ligand.

[0201] 5-5)

[0202] Among the relative concentrations obtained in 5-4), the lowest relative concentration was selected and used as a new standard, and the cultured supernatants having the higher concentrations of the Fas ligand variants were diluted to the same concentration as the standard with the aforementioned ASF medium. This adjusted the concentrations of the Fas ligand variants or the native Fas ligand in all solutions to the same ones.

[0203] 6-1) Preparation of Solutions of Anti-Fas Ligand Monoclonal Antibodies

[0204] Anti-Fas ligand monoclonal antibodies produced by the hybridomas, NOK1, NOK2, and NOK3 described in the international publication WO96/29350 (hereinafter may be referred to as NOK1, NOK2, and NOK3 immunoglobulin (antibody), respectively; they generically may be referred to as NOK antibody) were prepared and purified as described below.

[0205] First, hybridomas NOK1, NOK2, and NOK3 were each grew on an RPMI1640 medium supplemented with 10% FCS to 3×10⁷cells. Such 3×10⁷cells were prepared in a scale which comprises culturing 30 ml of the cultured cells in a 75 cm²culture flask (Pharcom). Specifically, starting from the culture at 2×10⁵cells/ml, the cells were recovered when reached 1×10⁶cells/ml.

[0206] The recovered hybridomas were suspended in 1.5 ml of PBS, and the 0.5 ml portions of the suspensions (1×10⁷cells) were administered per nude mouse intra-abdominally. After feeding for 10-18 days, the ascites deposited in the visceral cavity were recovered. The 10 ml portions of the ascites were purified. The purification was performed firstly by salting-out with ammonium sulfate thereby adding 10 ml of a saturated solution of ammonium sulfate dropwise to the equal volume of the ascites to combine together. After stirring at 4° C. for two hours, the mixtures were centrifuged at 10,000 g for 15 minutes. The supernatants were removed, and the sediments were dissolved in 5 ml of PBS. Then, the solutions were dialyzed overnight against 3 liters of PBS.

[0207] For NOK1 and NOK2 antibodies, only the protein G-adsorbed IgGs were purified by FPLC system using protein G column (trade name) (produced by Pharmasia) after recovering the dialyzed samples. The resultant samples were further dialyzed overnight against PBS. On the following day, concentration and purity of the protein were determined. For NOK3 antibody, gel-filtration was performed on the recovered dialyzed samples by FPLC system using Superdex200 column (trade name) (produced by Pharmasia) for gel-filtration to provide IgM, which had been eluted at void volume. Concentration and purity of the IgM were also determined.

[0208] The quantification of the protein was performed using a protein quantification reagent produced by BIO RAD. The procedure was in accordance with the protocol accompanied therein. SDS-PAGE electrophoresis was performed on the purified immunoglobulin in the reductive conditions, followed by CBB stain, observing no band of contaminant proteins but bands assuming to be the heavy chain and the light chain.

[0209] Serial twofold dilutions were performed with PBS containing 1% BSA on the purified NOK antibodies and the three purified humanized NOK2 antibodies (RNOK201, RNOK202, and RNOK203), specifically, using, as a starting liquid, 10 μg/ml for NOK1 antibody, 50 μg/ml for NOK2 antibody, 0.1 μg/ml for NOK3 antibody, 3.57 μg/ml for RNOK201, 1.86 μg/ml for RNOK202, and 3.35 μg/ml for RNOK203, so as to prepare totally eight kinds of antibody solutions per antibody.

[0210] 6-2) Identification of the Site Recognized by Each NOK Antibody with the Fas Ligand Variants

[0211] Solutions of 29 of the Fas ligand variants prepared according to the procedure of Example 5-5), as well as the native Fas ligand were used to examine the site on Fas ligand to be bound to NOK antibody and humanized NOK2 antibody by ELISA as described below. In this experiment, the relative chromophoric strength for each Fas ligand variant on the basis of that (OD₄₅₀value) for the native Fas ligand was determined by treatment with the same concentration of NOK antibody or humanized NOK2 antibody.

[0212] First, the solutions with the native Fas ligand and the Fas ligand variants as well as the solution of negative control, as prepared in 5-5) were each added to the eight wells in a lane of a 96-well Maxisorp plate (trade name) at 50 μl/well. The relationship in the location of lanes is the same as in 5-4). After incubation overnight at 4° C., the wells were rinsed three times in 0.01M PBS with 0.05% Tween20, and PBS with 1% BSA was added thereto at 100 μl/well. After incubation at 37° C. for an hour, the wells were rinsed three times in 0.01M PBS with 0.05% Tween20. Then, the solutions with various concentrations of NOK and humanized NOK2 antibodies prepared previously were each added to the eight wells at 50 μl/well. Subsequently, after incubation at 37° C. for two hours, the wells were rinsed three times in 0.01M PBS with 0.05% Tween20, and then a 5,000-fold dilution of HRP-labeled anti-mouse Ig-Cκ antibody or HRP-labeled anti-human Ig-Cκ antibody (both are produced by Southern Biotechonlogy Associate) in PBS with 1% BSA was added thereto at 50 μl/well. After incubation at 37° C. for an hour, the wells were rinsed three times in 0.01M PBS with 0.05% Tween20, and then a solution of chromophoric substrate (0.5 mM TMBZ and hydrogen peroxide) was added at 50 μl/well, after which, when a suitable development was obtained, the reaction was quenched with the addition of 0.3N sulfuric acid at 50 μl/well, followed by determining an absorbance in each well at a wavelength of 450 nm using Microplate Reader (produced by Molecular Devices). In the wells including the cultured supernatant of the native Fas ligand, the developments were shown depending on the concentration of NOK and humanized NOK2 antibodies. No development was shown in the wells including the solution of the negative control. The results in part are shown in FIG. 18. Consequently, it can be believed that the chromophoric strengths thus obtained (OD450 values) should be reflected to the specific binding of NOK and humanized NOK2 antibodies to the native Fas ligand in the aforementioned solutions.

[0213] Next, relative binding activities of NOK and humanized NOK2 antibodies to each Fas ligand variant were examined on the basis of the chromophoric strength for the native Fas ligand. First, specific chromophoric strength in the well was determined by subtracting the OD₄₅₀values of the wells in which the native Fas ligand or the Fas ligand variant was added to the wells including negative control from the OD₄₅₀values of the wells including the native Fas ligand or the Fas ligand variant at the same dilution. Then, the OD₄₅₀values were plotted against the dilutions of the cultured supernatant so as to select certain OD₄₅₀value from the range within which the OD₄₅₀values showed a linear decrease and every linear lines showed the same inclination, and the concentrations of NOK and humanized NOK2 antibodies were determined from the plots against the selected OD₄₅₀value. Finally, relative concentrations of NOK and humanized NOK2 antibodies for each Fas ligand variant were evaluated on the basis of the antibody concentration for the native Fas ligand to provide relative binding activities of NOK and humanized NOK2 antibodies to each Fas ligand variant.

[0214] The relative binding activities of NOK and humanized NOK2 antibodies to each Fas ligand variant thus obtained are summarized in FIGS. 19 and 20.

[0215] As shown in these figures, the decrease in the binding activity of NOK2 antibody was observed to the range from Arg at position 198 to Met at position 238 from the N-terminus of Fas ligand, and especially significant decreases were observed in Arg at position 198, Gly at position 199, Leu at position 205, Gln at position 220, Asp at position 221, Leu at position 222, Met at position 230, Gln at position 237 and Met at position 238, which numerical order is in accordance with Nagata, et al. Int. Immunology, vol. 6, p.1567-1574, 1994.

[0216] The three types of humanized NOK2 antibodies, RNOK201, RNOK202, and RNOK203 have nearly the same binding pattern, and the decrease in the binding activity was observed to the range from Arg at position 198 to Met at position 238, and especially significant decreases were observed in Arg at position 198, Gly at position 199, Leu at position 205, Gln at position 220, Asp at position 221, Leu at position 222, Lys at position 228, Met at position 230, Gln at position 237, and Met at position 238. On the other hand, this pattern was different from that for NOK2 in Arg at position 198, Lys at position 228, and Met at position 230, and these differences may be due to humanization of the former. From this findings, it is believed that the higher apoptosis-inhibitory activity of RNOK2 antibody than that of NOK2 antibody (as described in Example 4) should be due to alteration of quality and strength of the binding to the aforementioned recognized amino acids by the humanization, to cause increase in the binding activity to Fas ligand.

[0217] In the case of NOK1 antibody, the decrease in binding activity was observed from Gly at position 199 to Gln at position 237, and especially in Gly at position 199, Asn at position 203, Leu at position 205, Tyr at position 218, Gln at position 220, Asp at position 221, Leu at position 222, Gly at position 227, Lys at position 228, and Gln at position 237. In the case of NOK3 antibody, the decrease in binding activity was observed from Gly at position 199 to Met at position 238, and especially, in Gly at position 199, Gln at position 200, Asn at position 203, Leu at position 205, Tyr at position 212, Gln at position 220, Asp at position 221, Leu at position 222, Lys at position 228, Met at position 230, Gin at position 237, and Met at position 238.

[0218] The above NOK antibody is a monoclonal antibody obtained by immunization of mouse with Fas ligand-expressing cells as described in Example 1 of the Japanese Patent Application No. 303492/1995. NOK and humanized NOK2 antibodies are ones having the high inhibitory activity against apoptosis induced in the Fas-expressing cells via the interaction of Fas-Fas ligand, as shown in Examples 1 and 2 of the international publication WO96/29350 and in Example 4 herein. In other words, NOK antibody is the first anti-Fas ligand monoclonal antibody which has been obtained by immunization with Fas ligand having the native structure, and which has been shown a high apoptosis-inhibitory activity. These monoclonal antibodies are different from each other in terms of the amino acid sequences of the CDRs (complementary-determining regions), and the class and subclass, and therefore, they are entirely independent. Since the different antibodies recognize the broad region of the amino acid sequence of Fas ligand between Arg at position 198 to Met at position 238, as described above, it can be concluded that the anti-Fas ligand antibodies having the high apoptosis-inhibitory activity should generally recognizes the amino acid residues within the aforementioned region as a recognition site. Thus, the site must be a predominant region in the Fas ligand molecule, which can elicit an anti-Fas ligand antibody having the high apoptosis-inhibitory activity. Further, the above site is important for Fas ligand to exert apoptosis-inducing activity, and therefore, agents which can recognize and bind to the site may show apoptosis-inhibitory activity.

[0219] 1. Peptide library was prepared, which composes of 34 kinds of 15-mer peptides which comprises the 15-mer peptide starting from the N-terminus of the extracellular domain of Fas ligand (Gln at position 103 from the N-terminus of Fas ligand), the 15-mer peptide from 6 to 20 positions, the 15-mer from 11 to 25, 15-mer from 16 to 30, and the likes which were synthesized in turn by starting from downstream by five mers from the foregoing one, using PepSet (trade mark, produced by Chiron).

[0220] 2. Positions in Fas ligand to which Fas ligand-antibodies were reactive, were identified using the cultured supernatants of NOK1, NOK2, and NOK4 hybridomas according to the following manner.

[0221] (1) Each well on a 96-well plate (Maxisorp, trade name, produced by Nunc) was filled up with 4-fold dilution of Blocking solution (BlockAce, produced by Dainippon Pharmaceutical Con.), and the pin of PepSet in which the synthetic peptides were immobilized on its tip was immersed into each well on the plate to block the tip of the pin for two hours at room temperature.

[0222] (2) After completion of the blocking, the pin of PepSet was removed and washed with PBS.

[0223] (3) The cultured supernatants of the NOK1, NOK2, and NOK4 hybridomas were dispensed into a fresh 96-well plate at 100 μl/well. Antibody solution attached to PepSet was used as control. PepSet contains a pin as for both positive and negative control, and the antibody solutions against the same.

[0224] (4) Then, the pins were immersed into each well on the plate described in the aforementioned item (3), and the plate was reacted for two hours at room temperature.

[0225] (5) After removal of the pins from the plate of the aforementioned item (4), the pins were transferred to pads containing PBS to wash three times by shaking for 10 minutes.

[0226] (6) 1000-Fold dilution of HRP (horseradish peroxidase)-labeled anti-mouse IgG (produced by Cappel) with PBS was dispensed into a fresh 96-well plate at 100 μl/well, and the pins of PepSet were immersed into the wells to allow the reaction for two hours at room temperature.

[0227] (7) After the reaction, the pins of PepSet were removed, and washed three times in PBS by shaking for 10 minutes.

[0228] (8) Substrate solution having the following composition was portioned into a fresh 96-well plate at 100 μl/well, and the pins of PepSet were immersed into the wells to allow the reaction for 20 minutes at room temperature.

[0229] Composition of the substrate solution:

[0230] 0.4 mg/ml OPD, 0.4 μl/ml 30% H₂O₂, 0.1M citrate phosphate buffer (pH5.1).

[0231] (9) After removal of the pins of PepSet from the plate, 50 μl of 2N H₂SO₄was added to the wells to quench the reaction.

[0232] (10) Absorbance in the well liquids on the plate was determined by a plate reader (produced by BIO RAD).

[0233] (11) As a result, color changes by the enzymatic reaction of HRP conjugated to the peptide immobilized on the pins were observed in the wells immersed with the pins immobilized with the peptides, LYFVYSKVYFRGQSC (SEQ ID NO:133), SKVYFRGQSCNNLPL (SEQ ID NO:134), and RGQSCNNLPLSHKVY (SEQ ID NO:137), (i.e., 188-202 positions, 193-207 positions, 198-212 positions from the N-terminus of Fas ligand) for the cultured supernatants of NOK1 and NOK4 hybridomas, and in the wells with YPQDLVMMEGKMMSY (SEQ ID NO:136) VMMEGKMMSYCTTGQ (SEQ ID NO:137), and KMMSYCTTGQMWARS (SEQ ID NO:138), for the NOK2 hybridoma (218-232 positions, 223-237 positions, 228-242 positions from the N-terminus of Fas ligand). Thus, it has been shown that the anti-Fas ligand antibody produced by NOK1 and NOK4 hybridomas recognize the region, LYFVYSKVYFRGQSCNNLPLSHKVY (SEQ ID NO:139), of Fas ligand, whereas the anti-Fas ligand antibody of NOK2 hybridoma recognizes the region, YPQDLVMMEGKMMSYCTTGQMWARS (SEQ ID NO:140), of Fas ligand.

[0234] Analysis of the site recognized by anti-Fas ligand antibodies using the Fas ligand variants as described in Example 6 shows that most of the sequences of the synthetic peptides which were observed to be reactive in this Example encompasses the aforementioned recognition site against anti-Fas ligand antibody as identified in the analysis. This shows again that the above site contains amino acids which play an important role for antigen-recognition/binding.

[0235] In order to examine the position at which the site recognized by the anti-Fas ligand antibodies resides on Fas ligand as identified in Example 6, modelling for Fas ligand was performed. Fas ligand is understood to belong to a TNF family, which includes TNF-α and TNF-β, and to form a trimer similar to TNF (Manuel C. P., et al., Molecular Immunology, vol. 32 (10), p.761-772, 1996). Thus, the model for the Fas ligand trimer was constructed by a molecular modelling using a computer with TNF as a template, and the examination was performed to seek for the position at which the site recognized by the anti-Fas ligand antibodies resides on Fas ligand which plays an important role to the apoptosis-inhibitory activity which had been confirmed in the binding experiment between the anti-Fas ligand antibody and the Fas ligand variants

[0236] First, in the modelling for the Fas ligand trimer, the modelling was performed on its monomer. Specifically, Modeler (Molecular Simulations Inc.), a software for modelling which runs on Silicongraphics was utilized using as templates the three-dimensional structures of TNF-α and TNF-β, which are known to show the high homology to Fas ligand. The amino acid sequence of the extracellular domain of Fas ligand was aligned with those of TNF-α and TNF-β as shown in FIG. 21, and the results were inputted into Modeler. As templates, the data for the coordinates of TNF-α (PDB ID:1TNF) and TNF-β (PDB ID:1TNR) from Brookhaven Protein Data Bank (PDB) was inputted. Since PDB ID:1TNF represents a data for the crystal structure of the TNF-α trimer whereas PDB ID:1TNR represents a data for the crystal structure of the complex comprising TNF-β and TNF receptor 55, only data for the TNF-α monomer and TNF-β monomer was extracted from them. According to the Modeler's instructions, the modelling was performed modifying the conditions to obtain 10 models, such that Model No. 3 was selected as a Fas ligand model, which is lower in the energy following energy-minimizing calculation, the probability density function (i.e., PDF), and the value of Root Mean Square of all atoms (i.e., RMS).

[0237] Then, on the basis of the model for Fas ligand thus obtained, Fas ligand trimer was constructed. Specifically, the model for the above Fas ligand monomer was superimposed on the atomic coordinate of each segment of TNF monomer of the above TNF-α trimer, PDB ID:1TNF using QUANTA/CHARMm (produced by Molecular Simulations Inc.), a software running on Silicongraphics, and after correcting the bad contact, the energy-minimizing calculation was performed, so as to construct the model for the Fas ligand trimer. The monomers which form the trimer are each referred to as Fas ligand A molecule, B molecule, and C molecule, or A segment, B segment, and C segment. Data for the atomic coordinates (PDB format) is shown in Table 1.

[0238] In FIGS. 22-24, in order to clarify the relationship of the positions between the recognition sites of NOK and humanized NOK2 antibodies on the Fas ligand model constructed as shown above, the Fas ligand trimer is depicted with the lines produced by connecting only the carbon atoms at the α-position of the amino acids on the basis of the data for the atomic coordinates described in Table 1 using QUANTA/CHARMm, as well as, among the amino acid residues in the sites recognized by NOK1, NOK2, humanized NOK2 (RNOK201-203), and NOK3 antibodies as identified in Example 6, the amino acid residues resided on the same side as the Fas ligand trimer are indicated by shaded circles, whereas any other residues are indicated by open circles. In the case of NOK2 and humanized NOK2 antibodies in FIG. 23, the amino acid residues especially recognized by humanized NOK2 antibody are indicated by circles with dotted line. To be clearly understood, only two of the three molecules of Fas ligand, which reside on this side (segments B and C in Table 1) are depicted, and the numerical order of amino acids in the recognition site is in accordance with the order of the Fas ligand model as shown in FIG. 21 as alignment data. Van der Waals contact model of the atoms of Fas ligand was displayed on Ray Trace command of QUANTA/CHARMm, and the amino acids in the recognition site of NOK and humanized NOK2 antibodies as identified in Example 6 were superimposed and displayed. On the van der Waals contact model, the recognition sites of NOK1, NOK2, humanized NOK2, and NOK3 antibodies are those indicated in hatched area in FIGS. 29, 30, and 31, respectively. FIGS. 22-24 and FIGS. 29-31 are substantially the same as each other, and thus both represent the same model from the same view point, which expressions are merely different.

[0239] The analysis results noteworthily show that the recognition site which is bound by NOK and humanized NOK2 antibodies should span two molecules of Fas ligand in the light of its stereochemistry. Thus, it has been shown that NOK and humanized NOK2 antibodies recognize the site which appears once the Fas ligand trimer is formed. The finding that the antibody having a strong apoptosis-inhibitory activity such as NOK antibody recognizes the recognition site spanning two molecules of Fas ligand has not been previously reported and thus novel. Accordingly, it is understood that the recognition of and the binding to the site formed by spanning the two molecules of the Fas ligand trimer should be important to provide a strong apoptosis-inhibitory activity.

[0240] NOK and humanized NOK2 antibodies recognize the common amino acids as shown in FIGS. 22-24 and 29-31, such as Gly at position 199, Leu at position 205, Gln at position 220, Asp at position 221, Leu at position 222, Gln at position 237 in common between NOK1-3 and humanized NOK2 antibodies; Asn at position 203 in common between NOK1 and 3 antibodies; Lys at position 228 in common between NOK1 and 3, and humanized NOK2 antibodies; Met at position 230, Met at position 238 in common between NOK2 and 3, and humanized NOK2 antibodies, which numerical order of the amino acids is in accordance with Nagata, et al. Int. Immunology, vol. 6, p.1567-1574, 1994, and therefore, it is predicted that these antibodies may share a common recognized region.

[0241] On the other hand, the recognition site of antibodies on protein molecule antigens which are as large as the Fas ligand trimer would define a kind of plane considering the size of the antibodies, and when the antibodies recognize the antigens, the plane of the antigens composed of these recognized amino acids and the plane of the recognizing region on the antibodies come into contact each other face-to-face. The recognition site or the common site of NOK1-3 and humanized NOK2 antibodies as shown above should define a plane. In other words, the recognition site on the Fas ligand trimer molecule of the antibodies can be defined as a plane on the protein molecule, and any antibodies which strongly inhibit apoptosis-activity should enter or approach the plane defined by the recognized amino acids on the Fas ligand trimer molecule.

[0242] Consequently, the recognized amino acids on the Fas ligand trimer molecule can be examined as a member of plane. In general, plane may be depicted as a flat plane defined by three points A (XA, YA, ZA), B (XB, YB, ZB), and C (XC, YC, ZC) which are not on a straight line. Plane can be estimated by assigning the data for the atomic coordinates of the amino acids recognized by the antibodies to the expression. Thus, plane which is recognized by the antibodies can be depicted by selecting any three amino acids from those recognized by the antibodies and by estimating as such. For example, if Gln at position 57 of the Fas ligand B molecule, Asn at position 60 of the Fas ligand C molecule, and Gln at position 77 of the Fas ligand C molecule as shown in Table 1 from the amino acids recognized by NOK3 antibody, the data for the coordinates of their Cα atoms is estimated to be (20.209, 32.098, 0.872), (11.551, 31.704, 15.937), and (−8.765, 35.501, −9.840), respectively, and thereby, plane defined by these coordinates of the Cα atoms of the amino acids can be estimated. The plane defined by the combination of Gln at position 57 of the Fas ligand B molecule, Asn at position 60 of the Fas ligand C molecule, and Gln at position 77 of the Fas ligand C molecule can cover the plane of the region constituted by the amino acids recognized by NOK3 antibody. The plane which comes into contact with the antibody effectively, that is, the plane through which the antibody atoms approach to the Fas ligand trimer is believed to be a plane which resides opposite to the Fas ligand molecule being apart from the atomic coordinate of each Cα atom by the length of the side chain of the amino acid residues plus about 4.1 Å of the cut off value of the common van der Waals contact.

[0243] As discussed in Example 8, it is predicted that any antibodies which strongly inhibit apoptosis-activity share a common recognition site since NOK and humanized NOK2 antibodies recognize a certain common amino acids. On the other hand, it is believed that NOK and humanized NOK2 antibodies have the variable region which size is the nearly equal to a usual antibody, in the light of the amino acid length of the variable region of each antibody. Accordingly, in order to identify a common recognition site which is on the Fas ligand trimer of NOK and humanized NOK2 antibodies, modelling was also performed on NOK2 antibody selected as a representative, similarly to Fas ligand, thereby examining if the variable region of the antibody, especially the CDR, can cover the region composed of the recognized amino acids on the Fas ligand trimer, and if the answer is affirmative, what amino acids other than the aforementioned recognized amino acids may be a candidate for the recognized amino acids.

[0244] As a model for the stereochemistry of NOK2 antibody, NOK2 antibody model, which had been newly modeled using Modeler was used. First, the variable region of the H chain (VH) of PDB ID:1FOR (SEQ ID No: 131) and the variable region of the L chain (VL) of PDB ID:1TET (SEQ ID No: 132), which show the high homology to the VH and the VL of NOK2 antibody were utilized as templates for the three-dimensional structures. The amino acid sequences of the VH and VL of NOK2 antibody were aligned with those of 1FOR as shown in FIGS. 25 and 26, respectively, and the results were inputted into Modeler. According to the Modeler's instructions, the modelling was performed modifying the conditions to obtain five models, such that Model No. 5 was selected as a model for the variable region of NOK2 antibody, which is lower in the energy following energy-minimizing calculation, the probability density function (i.e. PDF), and the value of Root Mean Square of all atoms (i.e. RMS).

[0245] Size of the antigen-recognizing site within the variable region of NOK2 antibody (six CDRs: complementary-determining region) was examined using Graphical Cylinder command of QUANTA/CHARMm. As a result, the antigen-recognizing site of the NOK2 antibody model has the size nearly equal to the circle of about 17 Å (angstrom) radius, as shown in FIGS. 32 and 33. FIG. 33 shows that the amino acids of the CDR of the NOK2 antibody model are just encompassed within the circle of about 17 Å radius. Then, the circle of about 17 Å radius was positioned on QUANTA/CHARMm to cover the plane defined by the amino acids in the recognition site on the Fas ligand trimer and the individual recognized amino acids as shown in FIGS. 22-24, thereby revealing that the region corresponding to the antigen-recognizing site within the variable region of NOK2 antibody really cover all of the amino acids in the site of the Fas ligand trimer recognized by NOK and humanized NOK2 antibodies (FIG. 27). Consequently, it is believed that NOK and humanized NOK2 antibodies should cover the region nearly equal to the site on the Fas ligand trimer although the reactivities are different. This means that any antibodies having a high apoptosis-inhibitory activity in common recognize the site shown herein.

[0246] Then, the surface amino acids of the Fas ligand trimer within the circle, having an exposed area with 15 angstrom square or above were selected as a amino acid to be able to interact with the CDR amino acid of the antibody, so as to identify Ser at position 10, Tyr at position 23, Ile at position 25, Arg at position 55, Gly at position 56, Gln at position 57, Gln at position 94, Met at position 95, Arg at position 98, Ser at position 99, Phe at position 126, Glu at position 127, Glu at position 128, and Ser at position 129 of one of adjacent two Fas ligand molecules, as well as Ser at position 14, Met at position 15, Pro at position 16, Glu at position 18, Lys at position 35, Gly at position 36, Asn at position 60, Asn at position 61, Leu at position 62, Pro at position 63, Ser at position 65, Lys at position 67, Tyr at position 69, Arg at position 71, Tyr at position 75, Pro at position 76, Gln at position 77, Asp at position 78, Leu at position 79, Val at position 80, Lys at position 85, Met at position 87, Tyr at position 89, His at position 113, Tyr at position 115, Asn at position 117, Ser at position 119, Glu at position 120, Leu at position 121, and Ser at position 122 of the other Fas ligand molecule, wherein the numerical order of amino acids with respect to any molecules is in accordance with that of the Fas ligand monomer as shown in the alignment data of FIG. 21. These amino acids are shown in FIGS. 28 and 34. In the figures, the large circle encompassing the amino acids is the circle of about 17 Å radius, and among the selected amino acids, those which also correspond to the amino acids in the site recognized by NOK and humanized NOK2 antibodies are indicated by the shaded circles whereas the amino acids newly selected as those in the recognition site of the antibodies indicated by the open circles.

[0247] These amino acid numbers are rearranged in accordance with Nagata, et al., Int. Immunology, vol. 6, p1567-1574, 1994, to provide Ser at position 153, Tyr at position 166, Ile at position 168, Arg at position 198, Gly at position 199, Gln at position 200, Gln at position 237, Met at position 238, Arg at position 241, Ser at position 242, Phe at position 269, Glu at position 270, Glu at position 271, and Ser at position 272 of one of adjacent two Fas ligand molecules, as well as Ser at position 157, Met at position 158, Pro at position 159, Glu at position 161, Lys at position 178, Gly at position 179, Asn at position 203, Asn at position 204, Leu at position 205, Pro at position 206, Ser at position 208, Lys at position 210, Tyr at position 212, Arg at position 214, Tyr at position 218, Pro at position 219, Gln at position 220, Asp at position 221, Leu at position 222, Val at position 223, Lys at position 228, Met at position 230, Tyr at position 232, His at position 256, Tyr at position 258, Asn at position 260, Ser at position 262, Glu at position 263, Leu at position 264, and Ser at position 265 of the other Fas ligand molecule.

[0248] It is believed that these amino acids include not only the amino acids recognized by NOK3 antibody, but also the common amino acids in the site which is recognized by any antibodies strongly inhibiting apoptosis-activity such as NOK antibody and humanized NOK2 antibody. Thus, it is shown that the interaction with the amino acids encompassed in the site on the Fas ligand trimer is important for proteins, peptides, desirably antibodies or analogues thereof to show a high apoptosis-inhibitory activity.

[0249] The present invention is worth not only in firstly finding the site on the Fas ligand trimer recognized by anti-Fas ligand antibodies having a high apoptosis-inhibitory activity, but also in providing important information to design other various molecules for inhibiting apoptosis-activity. Thus, the invention can be utilized to develop new therapeutic or diagnostic agents by creating proteins, peptides, or desirably antibodies or analogues thereof, which are specifically reactive to the aforementioned site and the amino acids within the site (the site and amino acids described in Examples 6-9) so as to inhibit apoptosis.

[0250] The present invention provides a humanized immunoglobulins which are specifically reactive to Fas ligand to inhibit the physiological reactions of Fas-Fas ligand, said immunoglobulins being capable of using as a therapeutic agent for diseases such as AIDS, graft-versus-host diseases in bone marrow transplantation, autoimmune diseases (SLE, RA), diabetes mellitus, and possessing the characteristics as follows:

[0251] (1) The humanized immunoglobulin is compatible with other aspects of the immune system in human since the effector part thereof is from human, and for example it can lead to more effective dissolution of the targeted cells by complement-dependent cytolysis (CDC) or antibody-dependent cell-mediated cytotoxicity (ADCC);

[0252] (2) Human immune system does not recognize the framework or the C region of the humanized immunoglobulin, and therefore, the level of undesirable immune responses to the humanized immunoglobulin is lower than the mouse antibody in which the whole is xenogenic and the chimeric antibody in which the part is xenogenic; and

[0253] (3) The humanized immunoglobulin can be expected to provide the lower dose and the lower frequency of the administration compared to the mouse antibody since the former has the half life nearer to naturally-occurring human antibodies.

[0254] Further, the present invention which has found the site which is on Fas ligand and required to induce apoptosis may be utilized to find a new therapeutic or diagnostic agent by creating a recombinant protein or peptide which is specifically reactive to the amino acids within the site.