CA2074081A1 - Cytotoxic cell-specific protease-related molecules and methods - Google Patents

Abstract

A peptide capable of inhibiting, in a mammal to which the molecule is administered, the biological activity of a cytotoxic cell protease.

Description

.:
i WO91~1~85 PCT/US91/00~0 CYTOTOXIC CELL-SPECIFIC PROTEASE-REL~TED
MOLECULES AND METHODS
Backqround of the Invention This is a continuation-in-part of copending application serial number 002,960 filed on January 13, 1987.
This invention relates to protease inhibitors.
Thymus derived (T) lymphocytes play a major role in the immune system. Maturation of the T cell lineage involves three distinct stages: (a) generation of a T cell precursor from a pluripotent stem cell, (b) differentiation in the thymus, and (c) migration of mature cells to the peripheral tissues. Maturation of T cells within the thymus is antigen independent. ~owever, once they have left the thymus, upon interaction with an antigen they are driven through the final steps of differentiation to become mature cells. These final steps are complex and involve interactions with other cells and soluble effector molecules.
Several subsets of T cells have been recognized among activated peripheral T cells. There are three main classes: helper, suppressor, and cytotoxic. Helper T
lymphocytes potentiate immune responses (both humoral and cell-mediated) either by cell-cell contact or by synthesis and secretion of factors. These factors, although synthesized in response to an antigen-specific signal, can be either antigen-specific or antigen-nonspecific.
Suppressor T lymphocytes, inhibit the functions of other lymphocytes, again either directly or via soluble factors.
cytotoxic T lymphocytes are the effector cells in cell mediated immune reactions. They specifically recognize foreign antigens on the surface of cells, bind to them, and cause the target cell to lyse. Cytotoxic T lymphocytes are .: .
, 2Q~8~
WO91/1~85 PCT/US91/00 known to cause or exacerbate autoimmune diseases such as rheumatoid arthritis, and are also involved in allograft rejection and graft-versus-host disease.
The various steps in the process of cytotoxic T
lymphocyte induced lysis have been analyzed in some detail, e.g., Berke, (1983) Immunol. Rev. 72:5; Nabholz & MacDonald, (1983) Ann. Rev. Immunol. 1:273. Recent studies by Padack &
Konigsberg, (1984) J. Exp. Med. 160:695 and Henkart et al., (1984) J. Exp. Med. l~Q:75 have suggested that the dense cytoplasmic granules seen in CTL and natural killér cells -~ are directly involved in target cell lysis by a mechanism involving transmembrane channels.
A general description of cytotoxic T lymphocytes, natural killer cells, and killer (K) cells is contained in ;; 15 Stites et al., Basic & Clinical Immunology 227-31 (Lange Medical Publications, Los Altos, Ca., 1984).
Su~marv of the Invention In general, the invention features a vector containing a DNA sequence encoding the CCP1 protein.
In another aspect the invention features a vector containing a DNA sequence encoding the CCP2 protein.
In another aspect the invention features a vector containing a DNA sequence encoding the hCCP1 protein.
In another aspect the invention features a vector containing a DNA seguence encoding the hCCPX protein.
In another aspect the invention features substantially pure CCPl protein expressed from a vector containing a D~A sequence encoding the CCPl protein.
Substantially pure means a preparation with a purity of 95 or greater by weight, and free of the proteins, lipids, and carbohydrates with which the protein is-naturally associated.

' .:

;: ' - - . . . . , ~
: . . : -- ~ - .

.. . . . .
:: , , ~ . . :~, , ~ WO91/10685 2 b 7 ~ ~ 8 ~ PCT/US91/00~0 In another aspect the invention features substantially pure CCP2 protein expressed from a vector containing a DNA sequence encoding the CCP2 protein.
In another aspect the inve~tion features S substantially pure hCCP1 protein expressed from a vector containing a DNA sequence encoding the hCCP1 protein.
In another aspect the invention features substantially pure hCCPX protein expressed from a vector containing a DNA sequence encoding the hCCPX protein.
In another aspect, the invention features a peptide of the formula: Asp-Val-Asp-Ala; Ala-Pro-Asp-Ala; Ala-Asn-Pro-Ala; Phe-Pro-Arg-Phe; Ala-Pro-Arg-Phe; Phe-Pro-Asp-Phe;
Phe-Pro-Asn-Phe; Phe-Asn-Pro-Phe; or Phe-Asp-Pro-Phe.
The term competitive inhibition, as used herein, refers to inhibition in which the inhibitor combines with the free protease such that it competes with the normal substrate of the protease. Competitive inhibition is described, e.g., in Lehninger, Biochemistry 197-200 ~Worth, 2d ed. 1975).
The term protease, as used herein, refers to an enzyme that hydrolyzes, and thus cleaves, peptide bonds.
Cytotoxic lymphocytes, e.g. cytotoxic T lymphocytes (sometimes called T killer cells) and natural killer cells are described in Jandl, Blood: Textbook of Hematology (Little, Brown and Co., Boston, 1987) hereby incorporated by reference.
The term serine protease, as used herein, refers to a protease which has a serine residue at the active site of the enzyDe.
The term peptide, as used herein, includes proteins as well as peptides too short to be characterized as proteins. Generally those peptides having a molecular weight of greater than 5,000 are characteriZed as proteins.

.

WO91/1~5 2 0 7 4~0 ~-~ P~T/USgl/~O ~

The term cytotoxic cell protease, as used herein, refers to any protease, preferably a serine protease, that has 30% or greater homology, more preferably 50% or greater homology, with the protein encoding sequence of the murine C11 gene, and which cleaves at different sites than does ; plasmin. Preferably the cytotoxic cell protease is expressed by cytotoxic lymphocytes, more preferably exclusively by cytotoxic lymphocytes.
Cytotoxic lymphocytes produce, as part of their cytotoxic activity, proteases, some of which, we have discovered, cleave proteins at sites different from the sites cleaved by proteases such as plasmin produced by other cells of the body. These proteases are members of the cytotoxic cell protease family. The inhibitory molecules of the invention, since they mimic the unique cleavage sites recognized by cytotoxic cell proteases, can exclusively inhibit cytotoxic cell proteases e.g., those produced by cytotoxic lymphocytes. Thus a person suffering from an i~mune disorder, or experiencing allograft rejection, c:an be administered a molecule of the invention to inhibit the cytotoxic lymphocytes involved in the disease or rejection process, and the administered molecule will not interfere with, for example, lysis of blood clots, or other normal protease-dependent functions.
Other features and advantages of the invention will be apparent from the following description of the preferred embodiments and from the claims.
Description of the Preferred Embodiment The structure, synthesis, and use of the preferred embodiments are discussed next, after the drawings are briefly described.

' . ' . ' ,, , ~ . : . ' ~ . . . , : .
:,'. " ',, , '.-~. ~ ,' ~ ' . ' ',, ' ' ' "' ,' ` ',: ' ' . ':
... . . ~ . .
,", ' ' ' ' " ' ': '' ' " ;., , ' ' ' ' ' ' '`

~ 1~ 2 ~ 7 ~
WOgl/10685 PCT/US91/~ ~0 Drawinas Fig. 1 is a graph showing the correlation of a protease mRNA expression () with cell activation in a mixed lymphocyte culture.
Fig. 2 is a partial nucleotide sequence comparison of two protease-encoding cDNA's.
Fig. 3 is the nucleotide sequence of one of said cDNA's and the predicted protein structure it encodes.
Fig. 4 is a partial amino acid sequence comparison of five serine proteases.
Fig. 5 is the sequence of CCP2.
Fig. 6 is the sequence of hCll, the human analog of the murine C11 gene.
Fig. 7 is a restriction map of the hCCPX gene.
Fig. 8 is the nucleotide sequence of the hCCPX gene.
Fig. 9 is the predicted cDNA sequence encoded by the hCCPX gene.
Fig. 10 is the amino acid sequence of proteins encoded by hCCPX and the CCP genes.
Fig. 11 is the amino acid sequence of some protease inhibitors of the invention.
Table 1 shows the expression of Cll mRNA in infiltrating cells of tissue grafts.
Table 2 shows the degree of homology between CCPl and various proteins.
Table 3 shows the effect of peptides of the invention on the cytotoxicity of cells from a cyclosporine-A mixed lymphocyte reaction.
Table 4 shows the effect of peptides of the invention on the cytotoxicity of cytotoxic T-cells activated with ConA and interleukin 2.
The Appendix is a copy of Murphy et al. (1988) Proteins: Structure, Function, and Genetics 4:190-204 which i - . ~ - i .
,, '' : : ' ~
.

, . .
-207~33~
WO91/10685 PCTIUS~1/00~0 provides a detailed example ~f computer aided analysis of enzyme and substrate structure.
Structure The inhibitory molecules of the invention competitively inhibit the activity of cytotoxic cell proteases produced e.g. by cytotoxic lymphocytes, while not inhibiting the activity of proteases produced by other cell types or any other proteases produced by the cells producing the cyto`toxic cell proteases. Preferably the inhibitory molecules are peptides.
Cytotoxic lymphocytes synthesize a characteristic set of cytotoxicity-related proteases which are expressed at much reduced levels, if at all, in other subsets of lymphocytes. The cytotoxicity- related proteases can be divided into two groups, effector proteases and non-effector proteases. Effector proteases are released by a cytotoxic lymphocyte when it comes in contact with a target cell, and break down proteins in the membrane of the target cell or enter the target cell and hydrolyze intracellular proteins, leading to the cell's destruction. Non-effector proteases are involved in the enzymatic processes that lead to the production and/or release of the effector proteases (or other effector molecules) from the lymphocyte. Inhibiting the action of either an effector protease or a non-effector protease inhibits the ability of cytotoxic lymphocytes to destroy a target cell.
The preferred peptides contain the two amino acids that constitute the cleavage site recognized by the protease, and have between 3 and 20 (more preferably between -3 and S) amino acids residues. Shorter peptides are preferred because they are, in general, readily taken up cells. The peptides should not contain a cleavage site recognized ~y other proteases, for example, those sites .. . . . . . .

'"'-'`` 20~b8l';
~^ WO91t10685 PCT/US91/00~0 described by Zreighton, Proteins: Structure and Molecular Properties 336-37, 427-38 (W.H. Friedman, N.Y., 1983).
Described in Example l below is the isolation, cloning, and characterization of two genes expressed 5 exclusively in the cytotoxic T lymphocytes of mice.
(Exclusively means that either the genes are not expressed, or are only expressed in very low (less than 5 molecules of mRNA per cell) levels, in other types of cells in the organism). Example 2 describes the sequencing of the two lO genes, the determination of the amino acid sequencé of the protease which one of the genes encodes, and the characterization of the protease. Example 3 describes the identification and isolation of a human gene (hCII) encoding a cytotoxic cell protease (hCCPl) produced 15 exclusively by human cytotoxic T lymphocytes. Example 4 describes the isolation, cloning, and characterization of a gene encoding another human cytotoxic cell protease, human cytotoxic cell protease X (hCCPX). Example 5 describes the sequencing of the hCCPX gene, the determination of the amino 20 acid sequence of the hCCPX protease, and the characterization of the protease. Example 6 describes the determination of three dimensional structure of a cytotoxic ; cell protease and the structure of a peptide that can act as a competitive inhibitor of that protease. Example 7 25 describes several inhibitors of the invention. Example 8 ~ describes the production of substantially pure proteases and f their use in the design of inhibitors.
Example l Cells - The cytotoxic T-cell lines MTL2.8.2 and 30 MTLll.l were generated from CBA/J mice as described by ; Bleackley et al., (1982) J. Immunol. 128:758. EL4.El is an interleukin 2 (IL-2)-producing variant of the EL4 cell line described by Farr et al., (1980) J. Immunol. 125:2555. CHl .

': , ~ ' ' ' . /
2i~7~81 WO91/1~85 PCT/US91/00~0 is a CBA/J ~ CBA/J X BALB/c antigen- specific helper T-cell line. It was produced from a 2-day mixed lymphocyte ~ulture by continuous restimulation with irradiated Fl spleen cells in RPMI 1640 medium supplemented with 10% fetal bovine serum and 100 ~M 2-mercaptoethanol (RHFM). To generate human cytotoxic T lymphocytes (CTL), peripheral blood lymphocytes were incubated in RHFM and stimulated with irradiated allogeneic cells at days 0 and 7 and harvested at day 10.
The fetal-derived cells used are described by Teh et al., 10 (1985) J. Immunol. 135:1582. For the time course of cell activation, spleen cells from CBA/J mice were incubated in RHFM (106 cells per ml) and purified IL-2 (described by Riendeau et al., (1~8~) J. Biol. Chem. 25~:12114), either with an equal number of mitomycin C-treated EL4.El cells or Con A (2 ~qtml). Samples were removed at day 1 through day 6, assayed for cytotoxic activity by the procedure described in Shaw et al., (1978) J. Immunol. l~Q:1974, and analyzed by cytodot hybridization.
cDNA Library Construction - Double-stranded cDNA was synthesized from 4 ~g of MTL.2.8.2 mRNA as described by Gubler and Hoffman, (1983) Gene 25:263. Following repair with the Klenow fragment of DNA polymerase and T4 DNA
polymerase to maximize flush ends, phosphorylated ~_RI
linkers (P-L Biochemicals) were ligated to the cDNA in 70 mM
Tris-HCl, pH 7.6/10 mM MgC12/5 m~ dithiothreitol/l mM ATP/1 unit of ~4 DNA ligase at 14C overnight (Goodman &
MacDonald, (1979) Methods Enzymol. 68:75). After digestion with EcoRI, the product was run on a 5-ml Sepharose 4B
column, and the excluded fractions were pooled and ethanol-precipitated. The cDNA was ligated to FcoRI/bacterialalkaline phosphatase-treated pUC13 (P-L 8iochemicals) in 66 mM Tris-HCl, pH 7.6/6.6 mM MgC12tlO mM dithiothreitol/1 mM
ATP. Reactions were heated to 37C for 5 min, quick-chilled , , : : .

20~.~0:g~
`~ WO91/1~85 PCT/US91/~ ~0 _ g _ before the addition of 1 unit of T4 DNA ligase, and incubated at 14C for 2 hr. Escherichia ~li JM83 cells were made competent by using the CaC12/RbCl procedure described by Maniatis et al. in Molecular Cloning: A Laboratory S Manual (Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y., 1982) and were transformed with the ligated cDNA.
White colonies (those containing inserts) were ordered in 96-well microtiter plates and stored in LB medium containing 20% glycerol at -70C.
Differential Screening - Colonies were replicated in triplicate onto nitrocellulose filters, grown for 6 hr, and then amplified on chloramphenicol (100 ~g/ml) for 12 hr.
Bacteria were lysed, and the filters were prewashed to remove bacterial debris, as described by Maniatis, supra.
Prehybridization at 42C for 12-20 hr was done in 50%
(vol/vol) formamide containing 2x Denhardt's solution (lx Denhardt's solution = 0.02% polyvinylpyrrolidone/0.02%
Ficoll/0.02% bovine serum albumin), 4x SET buffer (lx SET
buffer = 0.6 M NaCl/0.12 M Tris-HCl, pH 8/1 mM EDTA), 0.1%
NaDodS04, 100 ~g of yeast tRNA per ml, and 125 ~g of poly(A) per ml (Sigma). Hybridization in the same buffer included 1-5 x 105 cpm of cDNA probe per m} synthesized from mRNA
with 20 ~g of T-primers per ml (Collaborative Research, Waltham, MA); 50 mM Tris-HCl (pH 8.3); 10 mM MgC12; 5 mM
dithiothreitol, 500 ~M each of dGTP, dATP, and dTTP; 70 mM
KCl; 30 ~Ci (1 Ci = 37 GBq) of ~-32P~dCTP (New England Nuclear, 300 Ci/mmol); and 15 units of avian myeloblastosis virus reverse transcriptase at 42C for 60 min. Template RNA was hydrolyzed by the addition of NaOH to 1.5 M.
Samples were boiled for 3 min and fractionated by Sephadex G-50 column chromatography. Filters were washed in 5x SET
buffer for 15 min at 22C and then in 2x SET buffer/50%
formamide for 20 min at 42C and were exposed to film (Kodak . ., ",: .

.
..

. - .

3~ ~
WO 91tlO~5 PCT/US91/00~0 -- 10 -- ::
X-Omat AR) with an intensifying screen for 1 to 3 days at -70C. Hybridized probe was removed by boiling the filters for 10 min in distilled water.
Blot Analysis - Cytodots were prepared as described by White and Bancroft (1982) J. Biol. Chem. 257:8569. For blot-hybridization analysis, total cytoplasmic RNA (10 ~g) or poly(A)+ mRNA (2 ~g) was denatured in 6.3%
formaldehyde/50% formamide at 55C and size fractionated on a O.8~ agarose gel containing O.66% formaldehyde. RNA was transferred to nitrocellulose as described by Thomas (1980) Proc. Natl. Acad. Sci. USA 77:5201. Plasmid DNA was digested with EcoRI, run on a 0.7% agarose gel, and transferred to nitrocellulose, as described by Southern (1975) J. Mol. Biol. 26:365. Filters were baked at 80C for 2 hr, then prehybridized at 42C for 6-12 hr in 50%
formamide containing 20 mM phosphate buffer (pH 6.8), 2 mM
pyrophosphate, 100 ~M ATP, 5x Denhardt's solution, 0.75 M
NaCl, 0.075 M sodium citrate (pH 7), 100 ~g of salmon sperm DNA per ml, 0.1% NaDodS04, 50 ~g of poly(A) per ml, and 2.5 mM EDTA. Hybridization was carried out in the same buffer with a nick-translated plasmid of specific activity 1 x 108 cpm/~g (Bethesda Research Laboratories kit) at 1 x 106 cpm/ml.
Results - Triplicate copies of the library were hybridized first with cDNA synthesized from MTL2.8.2 mRNA, then, after autoradiography and washing, with helper T-cell cDNA, and finally with thymocyte cDNA. Colonies that gave a higher hybridization signal with killer cell mRNA in at least two of the three copies of the library were picked.
Upon rescreening, again in triplicate, 36 of these 121 colonies appeared to be clearly CTL-specific. Plasmid DNA
isolated from these colonies was cut with EcoRI, and a series of cross-hybridizations was performed. Two clones .
; ~

~ WO91/10685 2 0 ~ ~ ~ 8 ~ PCT/USg1/~340 - 11 - . -were chosen for more extensive analysis: clone BlO because it appeared to be the most abundant in the library, cross-hybridizing strongly with eight other inserts, and clone Cll because it weakly cross-hybridized with BlO but not with all BlO-related clones (one other Cll-related sequence was found).
Cytodots prepared from a variety of cells and tissues were hybridized with nick-translated BlO and Cll.
The number of cells per dot was 104. The data with probe Cll are similar and are not discussed. The highest signal was detected in MTL2.8.2--i.e., the killer cell line that was used to generate the cDNA library. A weaker but positive signal was observed with MTL-III, a variant of MTL2.8.2 that had a low level of cytotoxicity and had become IL-2 and antigen independent. A similar level of expression was observed in a novel T-cell clone derived from murine fetal thymus of Teh, supra. In all over 20 cytotoxic T cell lines and cultures have been tested and all have been positive for BlO and Cll expression.
Natural killer (NK) and T killer (T~) cells were purified, cultured, and tested for the expression of Cll mRNA by the methods described in Manyak et al. (1989) J. Immunol. 142:3707-3713. Culturing NK cells in IL-2 induced: i) lytic activity, ii) chymase and tryptase enzymatic activities and iii) the total ~RNA levels of the Cll gene in a dose-dependent manner. Cll ~RNA reached peak activity on days 5 to 7 of culture. Similar resu~ts were seen with TK cells.
There was no evidence for expression of BlO or Cll in either mouse thymocytes or a helper T-cell line (CHl) that secretes IL-~ in response to antigen. Mouse brain, mouse liver, and a human CTL line were similarly negative under the high-stringency conditions of this experiment. In .

WO91/10~85 2 $ 7 ~ ~ 8 ~ pcr/us91~oo34o ~

addition, no evidence for expression of BlO or Cll was found in a helper T-cell hybridoma that secretes an antigen-specific factor (Kwong et al. (1984)) J. Immunol. 133:653.
To ensure that the negative samples did contain hybridizable RNA, all of the cytodots were reprobed with either a lymphocyte-specific probe or oligo(dT) or the T-cell antigen receptor ~-chain gene (Hendrick et al. (1984) Nature 308:153). Although the level of signal varied, all samples were positive.
To enrich for the B cells of a spleen cell suspension, lymphocytes were separated from adherent cells on Petri dishes and then treated with anti-Thy-l.2 antiserum. The enriched B cells were then incubated with lipopolysaccharide (LPS) or Con A or RHFM medium. After 24 hr, the cells were harvested, cytodots were prepared and the filter was probed with BlO or Cll. No expression of either sequence could be detected in any sample. However, when the blot was hybridized with an immunoglobulin ~ heavy chain probe (Calame et al. (1980) Nature 284:452) a strong positive signal was seen in the LPS-stimulated cells.
Poly(A) RNA was isolated from a variety of cell sources, run on a denaturing agarose gel, and transferred to nitrocellulose. The same filter was probed first with nick-translated BlO, then with Cll, and finally with probe lO, a cloned gene that detects mRNA in a variety of cell types (Paetkau et al., in Contemporary Topics in Molecular Biology 10:35 (S. Gillis ed., Plenum, N.Y., 1984)). Probe BlO
detected a single band (approximately 900 bases) in two different murine cytotoxic T cell clones, MTL2.8.2 and MTLll.l. No bands were detected in RNA from thymocytes, an antigen-specific helper cell line, or murine thymoma EL4. -When the blot was reprobed with Cll, again only the two cytotoxic T cell clones showed bands. However, in contrast . ; . , :, .

"
. ' ' '.,. .~
' ' '.':.

2 ~ 8 ~
W091/1~85 PCT/US91/00~0 to BlO, this probe hybridized to two bands, one of approximately 900 bases and the other of 1200 bases. Probe lO detected a band in all cell samples. In addition, blot-hybridization analysis was performed on poly(A)~ RNA from a number of murine cells including eleven CTL lines, two helper lymphocyte lines, brain cells, liver cells, three helper T-cell lines, unstimulated and LPS-stimulated B
lymphocytes, and one ~-cell myeloma. Of these, only the actively cytotoxic T cells expressed mRNAs that hybridized with BlO and Cll. To ensure that all tracks contained hybridizable RNA, the blot was rehybridized with probe lO.
A band of the expected size was seen in all tracks.
The results from the cytodots and blot-hybridization analysis indicates that both BlO and Cll are murine cytotoxic T lymphocyte specific.
CBA/J (H-2k) spleen cells were stimulated with either mitomycin C-treated EL4 cells (Fig. lA) or Con A
(Fig. lB). On each of the 6 days after stimulation, the level of cytotoxicity was measured in a chromium-release assay against EL4 (H-2b) (~ , S194(H-2d) (~), and RI(~=~k) (O) cell lines. Cytodots were also prepared on each of these days, and the blots were hybridized with nick-translated BlO and Cll. Data are presented only for BlO, as Cll gave indistinguishable results. Relative BlO mRNA
levels () were determined by scanning densitometry on an ELISA plate reader. In the allo-specific response (Fig. lA), the peak of cytotoxicity was observed on day 4, while the peak of BlO or Cll mRNA expression appeared to be on days 3 and 4. The peak of killing activity in the Con A-stimulated cells (Fig. lB) was also at day 4; however, thepeak of mRNA expression was very sharply on day 3. In both experiments, the mRNA expression was reduced to background levels by day 6, while there were still significant levels ... . . . . .. .
.. . . .
- -.:: . .
, ~ ,: :
. ' ' ' :' ~ ' 2~7~08~`:

of cytotoxicity on this day. When the cytodots were hybridized with 32P-end- labelled oligo(dT), the peak of total mRNA was seen on day 2. (data not shown) The experimental results illustrated in Fig. 1 indicates that the maximum expression of B10 and Cll mRNAs precedes the peak of cytotoxicity in'an n vlvo allogenic or mitogen-induced cytotoxic response by 24hrs; thus, they both fulfill the primary prerequisite for genes encoding proteins that are important in the lytic process.
In situ hybridization experi~ents indicate that a high proportion of T lymphocytes that infiltrate incompatible heart allografts in vivo express the Cll gene.
Complete details of the in situ hybridization procedure, and all related techniques, are found in Mueller et al., (1988) 15 J. Exp. Med. 167:1124-1136, Transplantations in these experiments were performed as described in Mueller et al. (1988) J. Exp.
Med. 167:1124-1136 and Billingham et al. (1977) Transplantation 23:171. In short, the myocardium of newborn 20 (12-36 h) BALB/cJ (H-2d) donor mice were diced into 0.1-0.2-cm fragments and subsequently transplanted under the kidney capsule of adult (6-8 wk) sex-matched C57 Bl/Ka recipients (H-2b; experimental animals). As a control, adult BALB/cJ (H-2d) mice received grafts from the same donor animals under the kidney capsule. On days 2, 4, 6, 8, 10, and 12 after transplantation, three experimental and two control animals were killed and 5-~m frozen sections through the graft were prepared. Labelled probe for in situ hybridization was prepared as described in Mueller et al. (1988) J. Exp. Med. 167:1124-1136 and as follows.
A 1.1-kb fragment of the C11 gene was subcloned into the polylinker of the transcription vector pSPT 672 using standard techniques. This vector ha a SP6 and a T7 promotor : . . . .
- : , : . . - . . .
:. . . ..
-' ~..
~ . . . ' ' - , . ~ ' , :' ' ..

~ WO9l/1~5 2 ~ 7 ~ 0 8 ~ PCT/US91/oo~o at the 5' and 3' end of the multicloning site, respectively. After linearization of the vector with an appropriate restriction enzyme, sense and antisense probes were prepared using SP6-polymerase and T7-polymerase (both from New England Biolab, Beverly, MA) reactions and (S-35) UTP No. SJ 1303, Amersham Corp., Arlington Heights, IL) at a final concentration of 12 ~M. The labelled nucleotide was dried down before adding the other reagents of the reaction mixture. A typical reaction (35 ~l) contained 7 ~l 5X SP6 buffer (final concentration; 40mM Tris-HCl, pH 7.9, 5mM
MgCl2; 2 mM spermidine); 3.5 ~l 100 mM dithiothreitol (DTT);

3.5 ~l ribonucleotides (CTP, ATP, and GTP; 10 mM each, in lo mM Hepes, pH 7.4); 3.5 ~l bovine serum albumin (BSA), 5 mg/ml; 1 ~l Rnasin, 40 U/~l (New England Biolab); 1 ~l linearized DNA template, 1 ~g/~l; 13.5 ~l H20. SP6 and T7 reactions were incubated for 90 min at 40C and 37C, respectively. DNA template was digested with DNase I (2U/~g DNA; Worthington) for 15 min at 37C. The RNA probe was subsequently extracted with phenol/chloroform, separated on a Bio-Gel P-60 spin column, and ethanol precipitated after adding 7.5 ~g of yeast tRNA per lo6 cpm-labelled probe. The probe was subsequently resuspended at 2 X 109 cpm/~l in Tris-EDTA (TE), boiled for 2 min. and stored frozen at -70C. For the hybridization, this probe was mixed with formamide (final concentration 50%), dextran sulfate (10%), DTT (lOOmM), NaCl (300mM), Tris-HCl, pH 7.5 (20mM), EDTA
(5mM) Denhardt's solution (lX) at a concentration of 2 X lo6 cpm/~l hybridization solution.
In situ hybridizations were performed according to Angerer et al. (1987) In In Situ hybridization:
Applications to the CNS, K. Valentino, J. Eberwine, and J. 8archus, eds. Oxford University Press, New York pp. 42-; 70 as modified in Mueller et al. (1988) J. Exp. Med.

.

. ,.=~., . - - -- . .
.
., : . , ,. . . . , - .,- :.: . . -.. ~. . . , , . ~ . . .. .. ..

WOsl/1~5 PCT/US91/00~0 ~
2 0~

167:1124-1136. 5-~m-thick cryostat sections were placed on poly-L-lysine (Sigma Chemical Co.)-coated glass slides and fixed in 4% paraformaldehyde dissolved in lX phosphate buffered saline (PBS) for 20 min., rinsed in PBS, and dehydrated through graded ethanol. Slides were stored at this stage at 4C before being used for in situ hybridization. In situ hybridizations on different cell populations were done on sorted cells that were spun onto poly-L-lysine-coated glass slides with a Shandon cytocentrifuge. These cytospin preparations were fixed and hybridized as described for cryostat sections. The fixed sections or cytospin preparations were treated with proteinase K (Boehringer Mannheim, Federal Republic of Germany), 1 ~g/ml in lOo mM Tris-HCl, pH 8.0, and S0 mM EDTA
15 at 37C for 30 min. The slides were postfixed again with 4%
paraformaldehyde for 20 min. Free amino groups on tissue sections were acetylated by treatment with 0.25% acetic anhydride in 0.1 M triethanolamine for 10 min. Eor the hybridization step, 10 ~l of the hybridization solution 20 (described above) containing 106 cpm S-35 UTP-labelled RNA
probe were placed on each section, covered with a siliconized coverslip (18 x 18 mm), and sealed with rubber cement. The sections were hybridized at 46C for 16-18 h.
Thereafter, the slides were washed in a solution containing 25 50~ formamide, 2x SSC ~SSC = 0.15M sodium chloride, 0.3M
sodium citrate at pH 7), 20 mM Tris at pH 7.5, and 5 mM EDTA
in four changes for a total of 2 h at 56C. After the first wash a digestion step with RNase A (20 ~g/ml) and RNase (1 U/ml) (both obtained from Sigma Chemical Co.) for 30 min at 37C was included. The slides were dipped into NTB-2 nuclear track emulsion ~Eastman Kodak, Rochester, NY), 1:2 diluted with 600 mM ammonium acetate, and exposed at 4C for 8 days. The slides were developed with Kodak developer D-.~,., . - : .

'', . :~ , . , '.
,.. ,. ~ ' , . .
': ' .. ' ' ' ' ' ~: . , ' - . , ' ' ' . . .
, - ' '.. ' ' : ' ' ' , ' , 2~08~
WO91/t~5 PCT/US91/~ ~0 19 for 2.5 min and fixed with Kodak fixer for 5 min.
Counterstaining was done with 4% Giemsa stain (Fisher Scientific Co., Orangeburg, NY) for 10-15 min. From each animal, two sections were each hybridized with a labelled S Cll antisense probe (complementary sequence to the cytoplasmic Cll mRNA) and one section was each hybridized with a labelled C11 sense probe.
The results of in situ hybridizations with Cll-specific probes demonstrated that the cellular infiltrate in rejecting allografts contains a high proportion of cells expressing Cll transcripts. See Table 1 which shows the frequency of infiltrating cells with detectable levels of Cll mRNA. Cryostat sections of the graft were hybridized with radiolabelled RNA antisense probe of the Cll qene. The results in Table 1 are expressed as the number of positive cells per unit area (1 mm2) of infiltration area. Three animals with an allograft and two animals with a syngeneic graft were examined and two sections of each animal and each probe were used for evaluation.
The first cells with detectable levels of C11 mRNA
were found on day 2 after transplantation both in animals with an allogeneic and those with a syngeneic graft. These positive cells, however, were extremely rare at this timepoint and were normally not found on every section of 2S the same animal. On day 4 after transplantation, the experimental animals showed a 5-10-fold higher frequency of ... . ., : , ., . ~, ., :-.. . . ~: . .... ., , - ... . .

2~ LO~ ~
WO91/10685 PCT/US91/00~0 Frequency of Infiltrating Cells with Detectable Levels of Cll mRNA

5 Days after Allogeneic graft Syngeneic graft transplan- -tation C11 C11 2 3 + 4 4 + 1 -4 44 + 69 3 + 4 10 6 205 + 84 7 + 4 8 313 + 56 21 + 12 323 + 112 15 + 1 12 350 + 189 3 + 1 C11 cells than the control group with a syngeneic graft.
The frequency of inflammatory cells expressing the gene increased dramatically between day 4 and 12 after allotransplantation and was at least eight times higher than in the control animals during this period.
In one of the control animals, the syngeneic graft became necrotiG and no viable syngeneic graft cells could be detected 8 d after transplantation. This animal, which was ; not included in Table 1, had 5-10 times more C11 mRNA+ cells than other control animals at the sa~e timepoint. However, 2~ compared to the experimental animals 8 days after transplantation, the frequency of positive cells was still -50% lower. In the ~irst 4 days after the mice received the allograft, about equal numbers of Cll+ cells were found among the infiltrating cells.
The amount of Cll specific ~RNA per cell, as measured as the number of silver grains over a single cell, increased steadily during the entire observation period in - . - . . . . . . ~ .
.. . . .:

: ~
: , . ~ ' ' ~ " ' ' ~ WO91/1~85 2 ~ 7 ~ ~ $ 1 PCT/US91/00~0 experimental animals, indicating that the gene was expressed over long periods, perhaps after local induction by alloantigen and/or mediators. In control animals, the expression level increased only slightly after more than 4 days after transplantation.
The phenotype of the Cll transcript positive cells was determined as detailed in Mueller et al. (1988) J. Exp. Med. 167:1124-1136. Briefly, small pieces of the allograft were digested with collagenase, and the resulting suspension of the isolated infiltrating T cells wére sorted on a fluorescence activated cell sorter for subsequent in situ hybridization. The infiltrating cells of the allograft and the splenocytes of six animals that received a heart muscle graft 6 days before were pooled and sorted according to their phenotype. The frequency of C11-positive cells in the CD8+ subset was generally 10-20 times higher in the infiltrate of the allograft than in the spleen of the same animals. ~he recovery of CD4+ cells from the infiltrate was always very low and the frequency of positive cells in this subpopulation was at least 10-fold lower than in the CD8 subset of infiltrating cells; of 84 C11 mRNA + cells analyzed, 82 were CD8+ (98%) and 2 CD4+ (2%). On cytospin preparations from sorts cells, C11-transcript positive cells were mainly found among the blast-like CD8+ cells. In double stainings of cell suspensions and tissue sections, no evidence for a significant contribution of CD4 , CD8 or CD4+, CD8+ T cells among the allograft infiltrating cells and the Cll transcript positive cells were found~
Example 2 Clones B10 and Cll were sequenced according to the dideoxy method of Stanger et al. (1980) J. ~ol.
Biol. 143:161. Sequence analysis of B10 and C11 (Fig. 2) reveals that they are related to each other and that the - - , . - .: . . ~ : . ~ ,. . .

:.. . - . : . : ; . . . : . .
'- . - ' : . ~ ......................... , ................... : ' ' ' .

æ~
W091/1~5 PCT/US91/00~0 hypothetical proteins they encode contain a short region characteristic of serine proteases, Asp-Ser-Gly-Gly (a sequence homologous to that surrounding Ser195 of chymotrypsin).
With B10 and Cll as probes, another CTL
complementary DNA (cDNA) library was screened, in which inserts greater than 1000 base pairs were cloned in AgtlO.
Forty thousand recombinants were screened and 39 plaques corresponding to C11 were isolated.
A cDNA insert of 1400 base pairs, which hybridized with C11, was selected for sequence analysis. The predicted protein sequence encoded, of molecular weight 25,319, is shown in Fig. 3. The putative start codon is preceded by a potential ribosome binding site CCW CCG (Hagenbuchle et al. (1978) Cell 13:551) and a polyadenylation signal sequence AAUAAA (Proudfoot & Brownlee, (1966) Nature 263:211) occurs just upstream from the poly(A) tract. Of the first 12 amino acids predicted, ten are hydrophobicl and the amino acid in position 2 (Lys) is basic, suggesting that this sequence may act as a signal to direct secretion or intracellular organelle location. A search of the National Biomedical Research Foundation (NBRF) protein sequence data bank revealed that the protein encoded by Cll resembles a number of serine proteases (Table 2).
Table 2 _ __ _ _ _ p~ ;" EC Sp~ Rc~u~ a~ed p~
~ Ca~l hn~ p~on ho~noloQ
7S no~e ~h ~or 3.4.21 Munne 29 22 ~ 26-229 ~o amr~ ,~3.',21.1 Ib~ine 1 200 16 216 35 a-_ B3.~21.1 Ib~inc 1200 16-216 36 Coi ~,d~ Clr 3.~,.21.~1 H~nun 52-22~. 56 238 3s _i-e 3A21.11 Po~inc 3-220 3-233 33 3.~.2~ -225 192- 2~ 33 I~MCr~ 3.~21 R-t 121~. 1213 51 ~in 3.~21~ R-c 26-225 S1 262 36 1~ 3.~21.~ n 3 22~ 563 ~7 3 CD~ 3.~2131 Hum-n 72 22~ 339 560 3s ~pin3.~ 21.'. S.~niw 19 220 22-al~, 33 S 3.'.. 21.' R~29 226 31 223 39 , :
':

.
.

~ WO91~1~85 2 ~ ~ I 0 81 PCT/US91/00340 When the sequences were optimally aligned according to the Dayhoff algorithm (Dayhoff, in Atlas of Protein Sequencing and Structure 5:l (Supp. 3) (National Biomedical Res. Found., Washington, D.C., 1979)), the homologies generally varied between 30 and 40 percent. The greatest homology was found with rat mast cell protease type II
(RMCPII), which had amino acids identical to lO9 of 215 amino acids encoded by Cll, giving a match per length of 51 percent. The amino acid residues known to form the catalytic triad of the active site in serine proteases ~His57, Aspl02, and Serl95) were all found in the protein encoded by Cll (Fig. 3, ~). The sequences around these residues, which are highly conserved among serine proteases, are also conserved in the Cll gene product. Indeed, largely because of conservation around this region, the protein encoded by Cll appears to be somewhat homologous (about 30 percent of 209 residues) even to the prokaryotic proteases trypsin and type B from Stre~tomvces ariseus.
The cytotoxic T lymphocyte-specific proteins (CCP"3) encoded by Cll and BlO will be referred to as CCPl and CCP2, respectively. In Fig. 4 the optimal protein alignment with CCPl is presented for RMCPII, bovine chymotrypsin, bovine trypsin, and CCP2 (not numbered, as the full sequence is not presented). The full sequence of CCP2 can be obtained by application of the procedures applied to Cll and CCPl. The full sequence of CCP2 is presented in Fig. 5.
RMCPII is an intracellular serine protease found in the granules of atypical mast cells. The high level of homology of CCPl with RMCPII is particularly intriguing as RMCPII has a num~er of structural features that make it exceptional in the serine protease superfamily. Protein CCPl contains cysteines in precisely the same positions as RMCPII which, by analogy with RMCPII, form three disulfide ~- . .: . ~ . : , .. - :, . : .,; , " , . , - : .
: . . . . . . .

WO91/1~85 PCT/US91/00~0 bonds. These occur in the same positions in chymotrypsin, trypsin, and elastase. Both CCP1 and RMCPII lack a disulfide bond that is present in all other known serine proteases, including several from prokaryotes, and that links Cysl91 with Cys220 in chymotrypsin. In both CCPl and RMCPII the first of these two half-cysteines is replaced by a phenylalanine, while the second half-cysteine has been deleted along with other residues. Linkage of Cysl91 to Cys220 is thought to b~ important in stabilizing the conformation of the substrate binding site (Woodbury et al., (1978) Biochem. 17:811). Its absence in CCP1 and RMCPII may lead to significant changes in that site and, hence, in substrate specificity.
Two other primary structure changes previously seen only in RMCPII and thought to alter substrate binding are also present in the predicted CCP1 protein. In RMCPII and CCP1 the amino aci~ six residues before the active-site serine is alanine. In chymotrypsin-like proteases it is serine and in trypsin-like proteases, aspartic acid. The residue in this position lies at the bottom of the S1 binding site, so the change to a less polar residue would indicate a preference for a hydrophobic amino acid at the Pl position in the substrate. Furthermore, the sequence Ser-Trp-Gly216 in chymotrypsin, which forms hydrogen bonds with the Pl and P3 residues of the substrate, is replaced by Ser-Tyr-Gly in CCP1 and RMCPII, again suggesting altered substrate specificity. Both of these changes are also seen with CCP2.
One of the few RMCPII-specific differences that is not present in CCP1 is the substitution of isoleucine at position 99 in chymotrypsin for asparagine. In most mammalian serine proteases this residue is hydrophobic, and indeed in CCP1 it appears to be phenylalanine. However, .-. ~ ' : ' ~ . . .
:.: , . ' . ': : ' ,. ' .
. :.

~WO91/1~85 2 0 7 ~ PCT/USg1/oo~o most of the RMCPII-specific changes are present in CCPl protein, suggesting that the substrate binding site of CCPl resembles that of RMCPII and is significantly different from those of other mammalian serine proteases.
Example 3 The initial step in determining the structure of a protease expressed exclusively by human cytotoxic T
lymphocytes and recognizing a unique protein cleavage site is to clone human cytotoxic T lymphocyte specific cDNAs.
lOPolyA RNA from a human cytotoxic T lymphocyte cell line, e.g., one of the lines on deposit at the Coriel Institute for Medical Research, Copewood and Davis Street, Camden, NJ, is used as a template for the synthesis, by standard procedures, of double stranded complementary DNA.
EcoRI recognition sequences are then ligated onto the ends of the dscDNA by standard methods, and the resultant molecules are size selected on low melt agarose and then inserted into the EÇ_RI site of ~gtll, all by conventional procedures. These recombinant molecules are then packaged -into A phage heads (Gigapack plus, Stragene) and used to infect E. ~1i Yl088. DNA from plaques harboring recombinant molecules are hybridized with radioactive probes generated from BlO and Cll by standard procedures to identify corresponding human genes. The screening is conducted in duplicate to minimize the possibility of false positives. hCll, a human counterpart of Cll, was found using the above procedures.
The phage DNA from any positive plaques are isolated and immediately recloned, using conventional procedures, in the plasmid vector pUC13. Large amounts of these recombinant plasmid DNAs are then isolated for further analysis. The human cytotoxic T lymphocyte specific clones can be characterized by restriction enzyme digestions and, ~ .

- - . .. .... .. ... ,. .. : .... . . . .

. - : ' : , . . . ' ~ , . . . : . . ' . . . . .

' . . ' ' ' '.' : .'' `'. ~ : ' .

5 PCT/US91/~

ultimately, sequence analysis. In addition, their relationships to one another can be investigated by standard cross-hybridization and heteroduplex mapping.
Tissue specific expression and transcript sizes of isolated genes can be established using the same methods as described for B10 and C11. Using Northern ~lot analysis, as described above, a number of different cell lines (all obtained from ATCC) were tested for expression of hC11.
CEM-CM3 (acute lymphoblastic leukemia), CCRF-CEM (acute lymphoblastic leukemia), CCRF-SB (acute lymphoblastic leukemia), RPMI 7666 (B lymphoblast), DLD-1 (colon adenocarcinoma), and CRL-7123 (spleen line) all failed to express hC11. Human thymocytes and peripheral blood lymphocytes were also negative. Cytolytic T cells, activated by mitogen, interleukin 2, anti-T cell-receptor antibody, or fucose, were all positive for hC11 expression.
A human cytotoxic T cell line was also positive. Thus, expression of hC11 appears to be specific to cytotoxic T
cells.
The correlation between the level of cytotoxicity and the expression of the human genes also can be examined using the above-described methods. The expression of hC11 was found to correlate with the cytolytic activity of the cells in which it was expressed. Expression of hCll was detected in lymphokine activated killer (LAX) cells. ~he procedure for generating LAX cells is essentially that of Rosenberg et al. (1985) N.E.d. Med ~ 1485.
When expression of a human cytotoxic T lymphocyte-specific gene correlates with toxicity, the gene is sequenced by standard methods (as was done with the B10 and Cll genes). From the gene seguence, the structure of the protease can be determined, and a computer analysis of the structure of protease performed, as with the C11 gene.

~..... . . :
-' ~

. . ~

, 2~7~81 ~i WO91/1~85 PCTtUS91/00~0 Further computer analysis can show the location of the active site of the enzyme, and the appropriate sequence of a peptide that can act as a competitive inhibitor can be determined.
hC11 was sequenced, as described above, and found to be very similar to murine gene C11 (Fig. 6). The active site of hCCPl, the protein encoded by hCll, resembles the active site of the murine protein, CCP1, very closely. Most importantly, like CCPl, hCCPl appears to have an Arg at Sl, imparting the unusual specificity of Asp at Pl. The only other difference is the substitution of an aromatic amino -acid two residues downstream from the Arg. Due to the similarity of the proteins encoded by hCll and Cll inhibitors synthesized to inhibit one should inhibit the -other.
A partially purified preparation of hCCPl does not cleave at sites recognized by trypsin and chy~otrypsin.
Analysis of hCll gene expression, by in situ hybridization to biopsy sample, indicates that hCll is expressed in cardiac tissue of a patient that rejected a transplanted heart. In situ hybridization and related procedures were performed as described above.
Example 4 A human placental genomic library, in ~ charon 4A, was screened by hybridization in 20% formamide and 6 x SSC
(l x SSC is 0.15 M sodium chloride, 0.3 M sodium citrate, pH7) at 41C with a mixture of radioactivity labelled cDNAs corresponding to the murine cytotoxic cell proteases CCPl-4, Bleackley et al., (1988) FEBS Letters 234: 153-159 and Lobe et al. (1976) Science 232: 858-861.
Phage DNA from one of the positive plaques gave a 6.3 kb EcoRI fragment (and ultimately a l.5 kb Bam fragment) (Fig. 7) that hybridized with the murine probes but failed, . .

WO91/1~85 PCT/US91/00~0 under conditions of high stringency, to hybridize with hCCPl. Preliminary sequence analysis revealed that the 1.5 kb fragment encoded a protein which was highly homologous to the murine cytotoxic proteases. Thus this gene is a new member of the human CCP family but is different from hCCP1.
hCCPX is expressed in cytotoxic cells. Poly A+ RNA
was purified from resting and activated peripheral blood lymphocytes and subjected to Northern blot analysis using the l.S kb genomic fragment as a probe. A transcript is clearly present in the activated cells that is absent in RNA
from the unstimulated control. Sometimes a small amount of transcrlpt ls seen in the unstimulated cells, perhaps due to cell~ar contamlnation, however, the transcript is always induced upon sti~ulation.
Beaause of the hl~h le-~el of homologs~ between the var~ous ~P ~amlly mem~ers cross-hy~rldlzatlon can occur.
In the ca~e of the murine genes, C~P1 can be disting~shed from the others because of a difference in transcript size.
; However, the transcripts detected by hCCP1 and HCCPX are very similar in mobility. Therefore, high stringency was~ing conditions were used to minimize cross-hybridization. With washing at 41C the 1.5 kb probe detects transcripts in both human and mouse cytotoxic cells. However at 550 the signal due to the cross-hybridization with the mouse transcripts is markedly lessthan that seen for the human RNA, even though this ~ouse cell line expresses extremely high levels of the protease transcripts. The human-human and human-mouse identities are both approximately 70%, thus we believe that the signal seen under high stringency washing conditions in the RNA from activated human cells is due to specific hybridization With hCCPX transcripts.

.'.~."': .
.

': ' . ; ,. ~'.. .. .

~ 2~7`~31 ~ WO91/10685 PCT/US91/00~0 In addition, no detectable signal was detected using this probe on RNA samples from a number of human cell lines obtained from the ATCC including CEM-CM3, CCRF-CAM, CCRF-SB
(acute lymphoblastic leukemias), RPMI 7666 (EBV-transformed B lymphoblast), DLD-l (colon, adenocarcinoma), CRL-7020 (thymus), CRL-7123 (spleen) and freshly isolated human splenocytes and thymocytes.
Example 5 The nucleotide sequence of the region indicated by the heavy line in Fig. 7 is presented in Fig. 8. A
comparison of this sequence with those of the murine ccP
genes revealed high levels of homology (-70% identity) in regions which correspond to exons and dissimilarity in reqions which correspond to introns. By placing the introns (the underlined regions in Fig. 8) in exactly the same places that they occur in the murine sequences (all four murine genes have introns in precisely the same positions, Lobe et al. (1988) Biochemistry 27: 6941-6946), the sequence of a cDNA could be determined (Fig. 9). A cDNA
corresponding to exons 3, 4 and 5 has been isolated and confirms the positioning of the introns. The predicted protein which would be encoded by this gene is 246 amino acids in length (molecular weight = 27,318). The amino acid sequence is shown below the nucleotide sequence in Fig. 9.
This protein was not found in the GenBank data base. It is however, homologous to a wide variety of serine proteases.
The highest level of identity was with the cytotoxic cells proteases (human 70%), murine (61%), cathepsin G (human 57%), and mast cell proteases (40-50%). In addition, a significant level of identity (-30%) was found with many other trypsin and chymotrypsin like enzymes. This protein is a serine protease and is related to the cytotoxic cell - - ,. , . . ,- .

.

2~7~08~ ~
WO91/1~ PCT/US91/00~0 proteins, it will be referred to as human cytotoxic cell protease-X (hCCPX).
An alignment of the hCCPX sequence with those predicted from the murine genes (Fig. 10) illustrates the high degree of primary se~lence similarity and also reveals that hCCPX shares many features in common with the CCP
genes, Bleackley et al. (1988) FEBS Letters ~ 153-159.
hCCPX is very basic (14% basic, 6% acidic amino acids) and contains a hydrophobic leader sequence of 18 residues followed by a putative zymogen dipeptide which precedes the mature protease amino terminal Ile residue. It is believed that the basic nature of the proteins may play a role in sequestering them within granules bound to proteoglycans, Stevens et al. (1988) Current Topics in Microbiology and Immunology 140: 93-108. The two sequences +21 to l24 (Ile Ile Gly Gly) and +29 to +36 (Pro His Ser Arg Pro ~yr Met Ala) which are fGund in all the CCPs, granzymes, RMCPI and II, and cathepsin G are also conserved in hCCPX as are the six cysteine residues which form disulfide bonds, Jenne et al., (1988) Current Topics in Microbiology and Immunology 140:33-48. The catalytic triad residues (marked with an "*"
in Fig. 10) which form the active site of the serine proteases are all present in the correct positions, Neurath (1984) Science ~ 350. The sequences surrounding these, which are highly conserved in serine proteases, are also conserved.
CCP1 and 2 both contain unusual residues in regions that are believed to be important in defining substrate specificity, Lobe et al. (1986) Science 232: 858-861 and Murphy et al. (1988) Proteins 4:190-204. In addition, they lack a disulfide bond which in other serine proteases is important in restricting the size of the substrate binding pocket. Si~ilar results were subsequently found for the ' .
.~ '' ~ WO91/1~85 2 0 7 ~ O ~ 1 PCT/US91t~ ~0 other CCPs and granzymes, Bleackley et al. (1988) FE~S
Letters 234:153-159 and Masson et al. (1987) Cell 49:679-685. The protease described here also has unusual residues in these same sites and lacks the disulfide bond. However, the pattern of amino acids seen in thi6 protein, namely Thr, Ser-Tyr-Gly, and Gly at positions -6, +15 to +17, and +25 relative to the active site Ser, does not correspond to any of the murine proteases characterized to date. It would appear then that hCCPX would also have an unusual substrate specificity.
Purified insert from the cDNA containing plas~id was labelled by rando~ priming and used as a probe for in situ -hybridization on human metaphase spreads. The gene is present at a single locus on chromosome 14 at qll.2. The human gene encoding hCCP1 maps to the same region. In mice the genes encoding CCPl, CCP2, CCP3, and CCP4 are all located on chromosome 14 close to the ~-chain of the T cell antigen receptor locus Brunet et al. (1986) Nature 322:268-271.
Exa~Fple 6 The three-dimensional structure of CCPl, the protease encoded by C11, was predicted by computer analysis. The use of comparative molecular modeling to predict the structure of a protease and its characteristic substrate is particularly reliable when the protein of unknown structure is relatively homologous with a protein of known three dimensional structure. The existence of a large database of known three dimensional structures of related proteins and their substrate is also very helpful. In the case of the cytotoxic cell proteases both of these criteria are met.
The model building procedure (as applied to CCP1 and -another unrelated serine protease) is described in detail in .

, , , ' ' ~

WO91/1~85 ` PCT/US91/00~0 Murphy et al. (1988) Proteins: Structure, Function, and Genetics 4:190-204 which is included herein as an appendix.
(The computer program MUTATE referred to in the appendix is available from Dr. R. Read, Department of Medical Microbiology, University of Alberta, Canada). Briefly, the process begins with aligning the sequence of the protein of unknown structure with the sequence of a template protein, a protein of known three-dimensional structure. In the case of highly homologous proteins the alignment is lo straightforward: the sequences are ali~ned and a computer generated model of the template protein is modified to yield a model of the structure of the unknown protein. The side chain of each amino acid of the template is then replaced with the side chain of the corresponding amino acid of the protein of unknown structure. The replacement side chain conformations are adjusted to follow the conformation of the replaced, i.e., template, side chain conformations when possible. When this is not possible preferred side chain angles are selected from a dictionary of preferred side chain conformations.
Subsequent refinements include adjusting the model to remove unacceptably close non-bonded intramolecular contacts and adjusting the placement of deletion and insertion loops. In the final step, the deduced structure is adjusted to relieve any remaining unacceptably close non-bonded contents.
The prediction of substrate structure is drawn from several types of information. This procedure begins with an examination of the deduced three dimensional structure of the protease and an analysis of the identity of amino acid residues in key positions on the catalytic site of the protease. This information is compared to the reactive site on the substrate of a closely related protease. The . :

, ~ .

~ WO91/10685 2 ~ 7 ~ ~ 8 i PCT/US91/00~0 sequence of that substrate can then be altered to achieve a sequence complementary to the catalytic site of the modeled protein.
Analysis of CCP1 indicates that the active 6ite has histidine, aspartic acid, and serine residues at the back, meaning that it is a serine protease. Computer analysis further indicated that this active site cleaves proteins at a cleavage site (between the C-linkage of an Asp residue and an N-linkage of an adjacent amino acid, Phe) different from the cleavage sites recognized by any other known eukaryotic serine proteases. This deduced cleavage site permits the synthesis of synthetic peptides which, by mimicking all or a portion of the natural cleavage site, can bind to the active site of the protease and competitively inhibit it.
Exam~le 7 The amino acid residues of a substrate are 4P3P2P1Pl P2 P3'P4' with cleavage by the protease occurring between P1 and Pl'. The corresponding interacting amino acids of the binding pocket of an enzyme are designated S4S3S2slsl's2's3~s4~ with S1 for e-g-~example interacting with Pl.
The computer generated three-dimensional structure of CCPl indicates that the residues of the binding pocket which might interact with a substrate are: Pro 28-Cys 42;
His 57-Asn 65; Leu 32; Ile 41; Ile 73; Tyr 151; Gly 153; Phe 99; Ser 214-Asp 219; Phe l91-Ser 195; Arg 226; and Asn 174-Arg 175. (See pages 198-200 of Murphy et al., Appendix).
The most important prediction is that Sl equals Arg 226.
This predicts an acid substrate specificity (probably Asp) at Pl, the site of cleavage. ~his specificity is unique among eukaryotic serine proteases. S2 appears from the computer analysis to be Phe 99, indicating a small amino acid e.g., Val, at P2. The presence of basic residues in S3 - .: .:; . :
:,' . . :.

WO91/1068S 2~7~ PCT/USgl/oo~o ~

and S4 predict acidic residues at P3 and P4. Guided by these considerations the inhibiting peptides, corresponding to residues P3-P1' of the substrate, were synthesized.
These peptides are shown in Fig. 11. The effect of the inhibitors on the cytotoxic properties of cytotoxic T
lymphocytes is shown in Tables 3 and 4.

% Lysis Control cytotoxicity 15% 15% 15% 15%
+ 100 ~g/ml peptide8% 13% 12% 9%
Control cytotoxicity 30~ 30% 30% 30%
+ 50 ~g/ml peptide 19% 17% 17% 23%

Cytotoxicity was measured with cells from a cyclosporine-A
induced mixed lymphocyte reaction mixed lymphocyte reaction (CsA-MIR). Spleen cells were obtained aseptically by pressing the spleen through a wire mesh into a medium of RPMI 1640 (GIBCO Laboratories, Grand Island, NY), 10~ (v/v) fetal bovine serum (GIBCO Laboratories~, 10-4M 2-mercaptoethanol, and 10 mM HEPES buf~er (Sig~a, St. Louis, MO) (RHFM). Responder cells (1-2x10 /ml) were cocultured with e ~ ~ numbers of allogenic stimulator cells (1500 rad from a Cs source) in RHFM plus 300mg/ml CsA and 200 units/ml interleukin 2 in a final vo~ume of 4 ml (Costar 6 well cluster) or 25 ml (Costar 75 cm tissue culture flask). The cultures were incubated at 37C in 5% CO and 90% relative humidity. Cells from the primary MLR cu~tures were harvested, washed in RHFM and then recultured with cytokines at a cell density of 2-5 x 105 cells/ml for 24 or 48 hours. In some experiments, viable cells were isolated by gradient density centrifugation. 4For cytotoxicity assays,5~ells were incubated with 10 target cells labelled with Na CrOI (New England Nuclear, Boston, MA) in a round-bottom microt~ter plate (final volume of 200 ~L). After 4hours at 37, 100 ~L of supernatant was removed from each well for counting. Specific lysis was calculated as:

.

~ . :
' .

'~`.` WO9l/10685 2 ~ 3 1 PCT/US91/00~0 lysis = çx~erimental - s~ontaneous release x 100 total release - spontaneous release Spontaneous release was obtained by incubating 51Cr-labelled targets alone, and total release from target cells incubated with 1% Zap-Isoton lytic agent (Coulter Electronics of Canada, Ltd., Mississauga, Ontario).

Untreated Pretreated PeptideControl effectors effectors 10 EF2394 54% 90% 58%
EF2395 61% 86% 78 EF2396 54% 73% 50%
EF2397 54% 87% 50%
EF2398 54% 107S 89%
15 EF2368 54% 100% 75%
EF2369 54% 100% 28%
EF2372 54% 97% 35%
EF2373 54% 87~ 40%
Cytotoxicity in cytotoxic T-cells activated with ConA and - .
interleukin 2 (I12). Cytotoxic T Cells were actiYated using 10 ~g/ml ConA and 10 U/ml IL2. The cytolytic activity was measured in a standard chromium release assay. Targets were pretreated for 3 hr with 100 ~g/ml peptide and were then mixed with either pretreated lOO ~g/ml peptide and were then mixed with either pretreated 100 ~g/ml peptide) or untreated effectors. All results are at an effector to target ratio of 5:1. Results are calculated as described in the legend of Table 1.

Example 8 The cDNA clones of the invention can be used to generate copious quantities of purified cytotoxic cell proteases by inserting the coding sequence of a cytotoxic cell protease gene into an expression vector and expressing the desired proteins in an expression system. These procedures are well known to those skilled in the art.

'. ' ..

.- ' :' : ' .. - : ~

a~
W091/10685 Pcr/us91/oo34 The possession of purified protease allows for a greatly simplified alternative approach to the design of inhibitor molecules. Rather than the extremely cumbersome and complex immunologically based assays used to produce the 5 results in Tables 3 and 4, the enzymatic action of the purified protease on a given substrate can be followed directly, by cleavage of the substrate, when the purified protease is available. (Sequence specific protease cleavage can be followed with standard thioester-based assays such as that described in Harper et al. (1984~ Biochem. 23:2995-3002). This allows a large number of potential inhibitors to be tested with relative ease. The purified protease based assay can be used alone, or in conjunction with the rational design factors obtained by computer analysis, to 15 screen large numbers of potential inhibitors. Positive compounds could then be tested for their immunosuppressive properties.
Inhibitor Peptide Synthesis The inhibitory peptides of the invention can be 20 prepared by standard solid phase synthesis, for example, a method in which a tert-butyloxycarbonylamino acid is attached to either chloromethyl resin containing 0.75 mM Cl g 1, or the p-methylbenzhydrylamine resin containing 0.35 mM
NH2 g ~ followed by the sequential addition of desired 25 amino acid residues to produce the desired peptide.
Synthetic reactions are performed in 70 ml polypropylene syringes fitted with a polyethylene frit using appar~tus and techniques described in Burton et al., (1975) Biochemistry 14:3892, and Merrifield, (1963) J. Amer. Chem. Soc. 85:2149.
30 Completeness of coupling is determined by the standard ninhydrin test. The C-terminal amino acid is attac:hed using procedures described in Stewart et al., Solid Phase Peptide Synthesis (W.~. Freeman ed. 1970), or Pietta et al., 1970 .

' ' ; .
. ~ . .
, "~ J WO 91/1~8~ 2 ~ PCT/US91/00~0 Chem. Comm. 650. Hplc purifications of the synthetic peptides are carried out using a Beckman ODS colu~n (10 x 250 mm).
Amino acid analyses of the synthetic peptides are, if desired, performed using a Durrum D-500 analyzer.
Cysteinyl residues in the peptides are quantitated as cysteic acid using a modification of the method of Moore (196~) in which loo mM peptide is oxidized with 2.0 ml performic acid (1 ml 30% H202 + 9 ml 88% HCOOH) for 2 -hrs. at 0. Performic acid is removed in a ~eacti-Therm at 40 using N2, and 0.5 ml distilled water is then added to the residue and re-evaporated. The product is then hydrolyzed using 6 N HCl. Free sulfhydryl groups are determined using the method of Ellman et al. ~1959).
Use The inhibitory molecules are effective inhibitors of -cytotoxic cells, e.g., cytotoxic lymphocytes. The inhibition of the target cell destroying activity of such cells can be used to treat patients suffering of autoimmune diseases such as Hashimoto's thyroiditis, primary myxedema, thyrotoxicosis, pernicious anaemia, autoimmune atrophic gastritis, Addison's disease, myasthenia gravis, juvenile diabetes, Goodpasture's syndrome, pemphigus vulgaris, pemphigoid, sympathetic ophthalmia, phacogenic uveitis, autoimmune haemolytic anaemia, idiopathic thrombocytopenic purpura, idiopathic leucopenia, primary biliary cirrhosis, active chronic hepatitis HBs-ve, cryptogenic cirrhosis (some cases), ulcerative colitis, Sjogren's syndrome, systemic lupus erythematosus (SLE), discoid LE, dermatomyositis, 30 scleroderma, rheumatoid arthritis, and possibly multiple ~-sclerosis, and similar diseases in other mammals, for example, various types of livestock such as cows. Such inhibition can also be used to treat allograft (a tissue or .. , . , . . .. . ~ .

, , ~ , : ; ' : ':
,: ! ' " ' ~ ' ' ' ' ' ', , , ~
~' '~., . ' ` ' ' ~

2 ~ 7 ~ ~ 3 ~
wo9l/lo68s PCT/US91/00~0 organ graft from a donor who is a genetically dissimilar member of the same species as the receptor) rejecti~n, and graft v. host disease.
The peptides can be administered to a mammal in a dosage of 25 to 500 mg/kg/day, preferably 50 to 100 mg/kg/day. When administered to mammals (e.g., orally, intravenously, parenterally, nasally, or by suppository), the peptides inhibit the ability of cytotoxic T lymphocytes to destroy cells, thus inhibiting the cell-mediated immune response to provide an effective treatment for thé above listed disorders.
Nucleic acid probes (prepared by standard methods) capable of hybridizing to a gene encoding a protease expressed only by cytotoxic lymphocytes can be used in a variety of useful hybridization assays. For example, such probes can be used to monitor cytotoxic T lymphocytes in transplanted tissue, e.g., by the n situ hybridization methods of Cox et al. (1984) Dev. Biol. 101:485. The presence of the lymphocytes in the transplanted tissue is an indication that the tissue is being rejected by the host organism and that appropriate immunotherapy should be undertaken.
The probes can also be used to assess the potential cytotoxicity of lymphokine activated killer cells. The generation and use of such cells to treat tumor patients is described by Rosenberg et al. (1985) N.E.J. Med. 313:1485.
Rosenberg describe how human peripheral-blood lymphocytes are treated with interleukin-2 (a lymphokine) to generate killer cells that will attack tumor cells when reintroduced into the host. The probes can be used in a hybridization assay with the nucleic acid of the treated lymphocytes by standard methods; the assay monitors the degree to which the activated killer cells have been generated by . . ,, ~ : ' ~ ., i- WO91/1068~ 2 ~ A PCT/VS91/00 determining the level of expression of the protease-encoding gene in the cells.
Other embodiments are within the following claims.
What is claimed is: -.. ...

:. -,-- - ,' ~ ' . - . ' ~ . ', . . : ~: i , . . .

' , '~' ' ': ~

Claims (9)

Claims

1. A vector comprising a DNA sequence encoding the CCP1 protein.

2. A vector comprising a DNA sequence encoding the CCP2 protein.

3. A vector comprising a DNA sequence encoding the hCCP1 protein.

4. A vector comprising a DNA sequence encoding the hCCPX protein.

5. A substantially pure CCP1 protein expressed from the vector of claim 2.

6. A substantially pure CCP2 protein expressed from the vector of claim 3.

7. A substantially pure hCCP1 protein expressed from the vector of claim 4.

8. A substantially pure hCCPX protein expressed from the vector of claim 5.

9. A peptide of the formula:
Asp-Val-Asp-Ala;
Ala-Pro-Asp-Ala;
Ala-Asn-Pro-Ala;
Phe-Pro-Arg-Phe;
Ala-Pro-Arg-Phe;
Phe-Pro-Asp-Phe;

;
; or .