Note: Descriptions are shown in the official language in which they were submitted.
Beh.ringwerke Aktiengesellschaft HOE 90/B 026 - Ma 824
and Dr. LP
The General Hospital Corporation 20458649
Description
Fusion proteins with immnunoglobulin portions, the
preparation and use thereof
The invention relates to genetically engineered soluble
fusion proteins composed of human proteins not belonging
to the immunoglobulin family, or of parts thereof, and of
various portions of the constant region of immunoglobulin
molecules. The functional properties of the two fusion
partners are, surprisingly, retained in the fusion
protein.
EP-A 0 325 262 and EP-A 0 314 317 disclose corresponding
fusion proteins composed of various domains of the CD4
membrane protein of human T cells and of human IgGl
portions. Some of these fusion proteins bind with the
same affinity to the glycoprotein gp120 of human iaununo-
deficiency virus as the cell-bound CD4 molecule. The CD4
molecule belongs to the immunoglobulin family and,
consequently, has a very similar tertiary structure to
that of immunoglobulin molecules. This also applies to
the a chain of the T-cell antigen receptor, for which
such fusions have also been described (Gascoigne et al.,
Proc. Natl. Acad. Sci. USA, vol. 84 (1987), 2937-2940).
Hence, on the basis of the very similar domain structure,
in this case retention of the biological activity of the
two fusion partners in the fusion protein was to be
expected.
The human proteins which are, according to the invention,
preferably coupled to the amino terminus of the constant
region of immunoglobulin do not belong to the immuno-
globulin family and are to be assigned to the following
classes: (i) membrane-bound proteins whose extracellular
domain is wholly or partly incorporated in the fusion.
These are, in particular, thromboplastin and cytokine
_ 2 - 2045869
receptors and growth factor receptors, such as the
cellular receptors for interleukin-4, interleukin-7,
tumor necrosis factor, GM-CSF, G-CSF, erythropoietin;
(ii) non-membrane-bound soluble proteins which are wholly
or partly incorporated in the fusion. These are, in
particularly, proteins of therapeutic interest such as,
for example, erythropoietin and other cytokines and
growth factors.
The fusion proteins can be prepared in known pro- and
eukaryotic expression systems, but preferably in mammal-
ian cells (for example CHO, COS and BHK cells).
The fusion proteins according to the invention are, by
reason of their immunoglobulin portion, easy to purify by
affinity chromatography and have improved pharmacokinetic
properties in vivo.
In many cases, the Fc part in fusion protein is
thoroughly advantageous for use in therapy and diagnosis
and thus results, for example, in improved pharma-
cokinetic properties (EP-A 0232 262). On the other hand,
for some uses it would be desirable to be able to delete
the Fc part after the fusion protein has been expressed,
detected and purified in the advantageous manner
described. This is the case when the Fc portion proves to
be a hindrance to use in therapy and diagnosis, for
example when the fusion protein is to be used as antigen
for immunizations.
There are in existence various proteases whose use for
this purpose appears conceivable. Papain and pepsin are
employed, for example, to generate F(ab) fragments from
immunoglobulins (Immunology, ed. Roitt, I. et al., Gower
Medical Publishing, London (1989)), but they do not
cleave in a particularly specific manner. Blood coagula-
tion factor Xa by contrast recognises in a protein the
relatively rare tetrapeptide sequence Ile-Glu-Gly-Arg and
performs a hydrolytic cleavage of the protein after the
3 - 2045869
arginine residue . Sequences which contain the
described tetrapeptide were introduced first by Nagai and
Thogersen in a hybrid protein by genetic engineering
means (Nagai, R. and Thogersen, H.C., Nature, vol. 309
(1984), 810-812). These authors were able to show that
the proteins expressed in E. coli actually are specifi-
cally cleaved by factor Xa. However, there is as yet no
published example of the possibility of such proteins
also being expressed in eukaryotic and, especially, in
animal cells and, after their purification, being cleaved
by factor Xa. However, expression of the proteins
according to the invention in animal cells is preferable
because only in a cell system of this type is there
expected to be secretion of, for example, normally
membrane-bound receptors as fusion partners with
retention of their natural structure and thus of their
biological activity. Secretion into the cell culture
supernatant facilitates the subsequent straightforward
purification of the fusion protein.
The invention thus relates to genetically engineered
soluble fusion proteins composed of human proteins not
belonging to the immunoglobulin family, or of parts
thereof, and of various portions of the constant regions
of heavy or light chains of immunoglobulins'of various
subclasses (IgG, IgM, IgA, IgE). Preferred as immuno-
globulin is the constant part of the heavy chain of human
IgG, particularly preferably of human IgGl, where fusion
takes place at the hinge region. In a particular embodi-
ment, the Fc part can be removed in a simple way by a
cleavage sequence which is also incorporated and can be
cleaved with factor Xa.
Furthermore, the invention relates to processes for the
preparation of these fusion proteins by genetic engineer-
ing, and to the use thereof for diagnosis and therapy.
- 3A - 2045869
The invention will now be described in relation to the
drawings, in which:
Figure 1 shows two oligonucleotide probe molecules
used in cloning of thromboplastin cDNA;
Figure 2 shows the nucleotide sequence of clone 2b-
Apr5 with the thromboplastin amino acid sequence deduced
therefrom;
Figure 3 shows two oligonucleotide sequences which
are partially homologous with the sequence of the coding
strand (A), and with the non-coding strand (B) of
thromboplastin cDNA;
Figure 4 shows the restriction map of plasmid
pTF1Fc;
Figure 5 shows two oligonucleotide sequences which
are partially homologous with the sequence of the coding
strand (A), and with the non-coding strand (B) of the IL-4
receptor cDNA cloned in the vector pDC302/T22-8;
Figure 6 shows the restriction map of plasmid
pIL4RFc;
Figure 7 shows two oligonucleotide sequences A and
B which are partially homologous with the sequence of the
coding strand (A), and with the non-coding strand (B) of the
EPO cDNA cloned in the vector pCES; and
Figure 8 shows the restriction map of plasmid
pEPOFc.
Finally, the invention is explained in further examples.
4 - 2045869
Example 1: Thromboplastin fusion proteins
Blood coagulation is a process of central importance in
the human body. There is appropriately delicate regula-
tion of the coagulation cascade, in which a large number
of cellular factors and plasma proteins cooperate. These
proteins (and their cofactors) in their entirety are
called coagulation factors. The final products of the
coagulation cascade are thrombin, which induces the
aggregation of blood platelets, and fibrin which stabil-
izes the platelet thrombus. Thrombin catalyzes the
formation of fibrin from fibrinogen and itself is formed
by limited proteolysis of prothrombin. Activated factor
X (factor Xa) is responsible for this step and, in the
presence of factor Va and calcium ions, binds to platelet
membranes and cleaves prothrombin.
Two ways exist for factor X to be activated, the extrin-
sic and the intrinsic pathway. In the intrinsic pathway
a series of factors is activated by proteolysis in order
for each of them to form active proteases. In the extrin-
sic pathway, there is increased synthesis of thrombo-
plastin (tissue factor) by damaged cells, and it acti-
vates factor X, together with factor VIIa and calcium
ions. It was formerly assumed that the activity of
thromboplastin is confined to this reaction. However, the
thromboplastin/VIIa complex also intervenes to activate
the intrinsic pathway at the level of factor IX. Thus, a
thromboplastin/VIIa complex is one of the most important
physiological activators of blood coagulation.
It is therefore conceivable that thromboplastin, apart
from its use as diagnostic aid (see below), can also be
employed as constituent of therapeutic agents for treat-
ing inborn or acquired blood coagulation deficiencies.
Examples of this are chronic hemophilias caused by a
deficiency of factors VIII, IX or XI or else acute
disturbances of blood coagulation as a consequence of,
for example, liver or kidney disease. Use of such a
- 5 -
2045869
therapeutic agent after surgicial intervention would also
be conceivable.
Thromboplastin is an integral membrane protein which does
not belong to the immunoglobulin family. Thromboplastin
cDNA sequences have been published by a total of four
groups (Fisher et al., Thromb. Res., vol. 48 (1987),
89-99; Morrisey et al., Cell, vol. 50 (1987), 129-135;
Scarpati et al., Biochemistry, vol. 26 (1987), 5234-5238;
Spicer et al., Proc. Natl. Acad. Sci. USA, vol. 84
(1987), 5148-5152). Thromboplastin cDNA contains an open
reading frame which codes for a polypeptide of 295 amino-
acid residues, of which the 32 N-terminal amino acids act
as signal peptide. Mature thromboplastin comprises
263 amino-acid residues and has a three-domain structure:
i) amino-terminal extracellular domain (219 amino-acid
residues); ii) transmembrane region (23 amino-acid
residues); iii) cytoplasmic domain (carboxyl terminus;
21 amino-acid residues). In the extracellular domain
there are three potential sites for N-glycosylation
(Asn-X-Thr). Thromboplastin is normally glycosylated but
glycosylation does not appear essential for the activity
of the protein (Paborsky et al., Biochemistry, vol. 29
(1989), 8072-8077).
Thromboplastin is required as additive to plasma samples
in diagnostic tests of coagulation. The coagulation
status of the tested person can be found by the one-stage
prothrombin clotting time determination (for example
Quick's test). The thromboplastin required for diagnostic
tests is currently obtained from human tissue, and the
preparation process is difficult to standardize, the
yield is low and considerable amounts of human starting
material (placentae) must be supplied. On the other hand,
it is to be expected that preparation of native,
membrane-bound thromboplastin by genetic engineering will
also be difficult owing to complex purification proces-
ses. These difficulties can be avoided by the fusion
according to the invention to immunoglobulin portions.
CA 02045869 2002-02-04
- 6 -
The thromboplastin fusion proteins according to the
invention are secreted by mammalian cells (for example
CHO, BHR, COS cells) into the culture medium, purified by
TM
affinity chromatography on protein A-Sepharose and have
surprisingly high activity in the one-stage prothrombin
clotting time determination.
Cloning of thromboplastin cDNA
The sequence published by Scarpati et al., Biochemistry,
vol. 26 (1987), 5234-5238, was used for cloning the
thromboplastin cDNA. Two oligonucleotide probe molecules
(see Fig. 1) were derived from this. These two probe
molecules were used to screen a cDNA bank from human
placenta (Grundmann et al., Proc. Natl. Acad. Sci. USA,
vol. 83 (1986), 8024-8028).
cDNA clones of various lengths were obtained. One clone,
2b-Apr5, which is used for the subsequent procedure,
codes for the same amino-acid sequence as the cDNA
described in Scarpati et al. Fig. 2 depicts the total
sequence of the clone 2b-AprS with the thromboplastin
amino-acid sequence deduced therefrom.
Construction of a hybrid plasm.id pTF1Fc coding for
thromboplastin fusion protein.
The plasmid pCD4E gamma 1(EP 0 325 262 A2; deposited at
the ATCC under the number No. 67610) is used for
expression of a fusion protein composed of human CD4
receptor and human IgGl. The DNA sequence coding for the
extracellular domain of CD4 is deleted from this plasmid
using the restriction enzymes HindIiI and BamHI. Only
partial cleavage must be carried out with the enzyme
HindIII in this case, in order to cut at only one of the
two HindIII sites contained in pCD4E gamma 1 (position
2198). The result is an opened vector in which a eukary-
otic transcription regulation sequence (promoter) is
followed by the open HindIII site. The open BamHI site is
- 7 2045869
located at the start of the coding regions for a penta-
peptide linker, followed by the hinge and the CH2 and CH3
domains of human IgGl. The reading frame in the BamHI
recognition sequence GGATCC is such that GAT is trans-
lated as aspartic acid. DNA amplification with
thermostable DNA polymerase makes it possible to modify
a given sequence in such a way that any desired sequences
are attached at one or both ends. Two oligonucleotides
able to hybridize with sequences in the 5'-untranslated
region (A; 5' GATCGATTAAGCTTCGGAACCCGCTCGATCTCGCCGCC 3')
or
coding region
(B: 5' GCATATCTGGATCCCCGTAGAATATTTCTCTGAATTCCCC 3') of
thromboplastin cDNA were synthesized. Of these, oligo-
nucleotide A is partially homologous with the sequence of
the coding strand, and oligonucleotide B is partially
homologous with the non-coding strand; cf. Fig. 3.
Thus, amplification results in a DNA fragment (827 bp)
which contains (based on the coding strand) at the 5' end
before the start of the coding sequence a HindiIl site,
and at the 3' end after the codon for the first three
amino-acid residues of the transmembrane region a BamHI
site. The reading frame in the BamHI cleavage site is
such that ligation with the BamHI site in pCD4E gamma 1
results in a gene fusion with a reading frame continuous
from the initiation codon of the thromboplastin cDNA to
the stop codon of the heavy chain of IgGl. The desired
fragment was obtained and, after treatment with HindilI
and BamHi, ligated into the vector pCD4E gamma 1, as
described above, which had been cut with HindIiI
(partially) and BamHI. The resulting plasmid was called
pTFlFc (Fig. 4).
Transfection of pTF1Fc into manamalian cells
The fusion protein encoded by the plasmid pTFlFc is
called pTF1Fc hereinafter. pTF1Fc was transiently
expressed in COS cells. For this purpose, COS cells were
2045869
- 8 -
transfected with pTF1Fc with the aid of DEAE-dextran
(EP A 0 325 262). Indirect immunofluorescence investiga-
tions revealed that the proportion of transfected cells
was about 25 %. 24 h after transfection, the cells were
transferred into serum-free medium. This cell supernatant
was harvested after a further three days.
Purification of pTF1Fc fusion protein from cell culture
supernatants
170 ml of supernatant from transiently transfected COS
cells were collected overnight in a batch process in a
column containing 0.8 ml of protein A-Sepharose at 4 C,
washed with 10 volumes of washing buffer (50 mM tris
buffer pH 8.6, 150 mM NaCl) and eluted in 0.5 ml frac-
tions with eluting buffer (93:7 100 mM citric acid:
100 mM sodium citrate). The first 9 fractions were
immediately neutralized with 0.1 ml of 2M tris buffer
pH 8.6 in each case and then combined, and the resulting
protein was transferred by three concentration/dilution
cycles in an Amicon microconcentrator (Centricon 30) into
TNE buffer (50 mM tris buffer pH 7.4, 50 mM NaCl, 1 mM
EDTA). The pTF1Fc obtained in this way is pure by
SDS-PAGE electrophoresis (U.K. L'ammli, Nature 227 (1970)
680-685). In the absence of reducing agents it behaves in
the SDS-PAGE like a dimer (about 165 KDa).
Biological activity of purified TF1Fc in the prothrombin
clotting time determination
TF1Fc fusion protein is active in low concentrations
(> 50 ng/ml) in the one-stage prothrombin clotting time
determination (Vinazzer, H. Gerinnungsphysiologie und
Methoden im Blutgerinnungslabor (1979), Fisher Verlag
Stuttgart). The clotting times achieved are comparable
with the clotting times obtained with thromboplastin
isolated from human placenta.
- 9 - 2045869
Example 2: Interleukin-4 receptor fusion proteins
Interleukin-4 (IL-4) is synthesized by T cells and was
originally called B-cell growth factor because it is able
to stimulate B-cell proliferation. It exerts a large
number of effects on these cells. One in particular is
the stimulation of synthesis of molecules of immuno-
globulin subclasses IgGl and IgE in activated B cells
(Coffmann et al., Immunol. Rev., vol. 102 (1988) 5). In
addition, IL-4 also regulates the proliferation and
differentiation of T cells and other hemopoietic cells.
It thus contributes to the regulation of allergic and
other immunological reactions. IL-4 binds with high
affinity to a specific receptor. The cDNA which codes for
the human IL-4 receptor has been isolated (Idzerda et
al., J. Exp. Med., vol. 171 (1990) 861-873). It is
evident from analysis of the amino-acid sequence deduced
from the cDNA sequence that the IL-4 receptor is composed
of a total of 825 amino acids, with the 25 N-terminal
amino acids acting as signal peptide. Mature human IL-4
receptor is composed of 800 amino acids and, like
thromboplastin, has a three-domain structure: i) amino-
terminal extracellular domain (207 amino acids);
ii) transmembrane region (24 amino acids) and iii)
cytoplasmic domain (569 amino acids). In the extra-
cellular domain there are six potential sites for
N-glycosylation (Asn-X-Thr/Ser). IL-4 receptor has
homologies with human IL-6 receptor, with the p-subunit
of human IL-2 receptor, with mouse erythropoietin
receptor and with rat prolactin receptor (Idzerda et al.,
loc. cit.). Thus, like thromboplastin, it is not a member
of the immunoglobulin family but is assigned together
with the homologous proteins mentioned to the new family
of hematopoietin receptors. Members of this family have
four cysteine residues and a conserved sequence
(Trp-Ser-X-Trp-Ser) in the extracellular domain located
near the transmembrane region in common.
On the basis of the described function of the IL-4/IL-4
- 10 - 2`0045869
receptor system, there is a possible therapeutic use of
a recombinant form of the IL-4 receptor for suppressing
IL-4-mediated immune reactions (for example transplant
rejection reaction, autoimmune diseases, allergic reac-
tions).
The amount of substance required for therapy makes it
necessary to prepare such molecules by genetic
engineering. Because of the straightforward purification
by affinity chromatography and improved pharmacokinetic
properties, according to the invention the synthesis of
soluble forms of the IL-4 receptor as immunoglobulin
fusion protein is particularly advantageous.
The IL-4 receptor fusion proteins are secreted by mammal-
ian cells (for example CHO, BHK, COS cells) into the
culture medium, purified by affinity chromatography on
protein A-Sepharose and have, surprisingly, identical
functional properties to the extracellular domain of the
intact membrane-bound IL-4 receptor molecule.
Construction of a hybrid plasmid pIL-4RFc coding for IL-4
receptor fusion protein.
Cutting of the plasmid pCD4E gammal with XhoI and BamHI
results in an opened vector in which the open XhoI site
is located downstream from the promoter sequence. The
open BamHI site is located at the start of the coding
regions for a pentapeptide linker, followed by the hinge
and the CH2 and CH3 domains of human IgGl. The reading
frame in the BamHI recognition sequence GGATCC is such
that GAT is translated as aspartic acid. DNA amplifica-
tion with thermostable DNA polymerase makes it possible
to modify a given sequence in such a way that any desired
sequences can be attached at one or both ends. Two
oligonucleotides able to hybridize with sequences in the
5'-untranslated region
(A: 5' GATCCAGTACTCGAGAGAGAAGCCGGGCGTGGTGGCTCATGC 3') or
coding region
- 11 - 2045869
(B: 51 CTATGACATGGATCCTGCTCGAAGGGCTCCCTGTAGGAGTTGTG 3')
of the IL-4 receptor cDNA which is cloned in the vector
pDC302/T22-8 (Idzerda et al., loc. cit.) were
synthesized. Of these, oligonucleotide A is partially
homologous with the sequence of the coding strand, and
oligonucleotide B is partially homologous with the non-
coding strand; cf. Fig. 5. Amplification using thermo-
stable DNA polymerase results in a DNA fragment (836 bp)
which, based on the coding strand, contains at the 5' end
before the start of the coding sequence an XhoI site, and
at the 3' end before the last codon of the extracellular
domain a BamHI site. The reading frame in the BamHI
cleavage site is such that ligation with the BamHI site
in pCD4E gamma 1 results in a gene fusion with a reading
frame continuous from the initiation codon of the IL-4
receptor cDNA to the stop codon of the heavy chain of
IgGl. The desired fragment was obtained and, after
treatment with XhoI and BamHI, ligated into the vector
pCD4E gamma 1, described above, which had been cut with
XhoI/Ba.mIiI. The resulting plasmid was called pIL4RFc
(Fig. 6).
Transfection of pIL4RFc into mammalian cells
The fusion protein encoded by the plasmid pIL4RFc is
called pIL4RFc hereinafter. pIL4RFc was transiently
expressed in COS cells. For this purpose, COS cells were
transfected with pIL4RFc with the aid of DEAE-dextran
(EP A 0 325 262). Indirect immunofluorescence investiga-
tions revealed that the proportion of transfected cells
was about 25 %. 24 h after transfection, the cells were
transferred into serum-free medium. This cell supernatant
was harvested after a further three days.
Purification of IL4RFc fusion protein from cell culture
supernatants
500 ml of supernatant from transiently transfected COS
- 12 - 2045869
cells were collected overnight in a batch process in a
column containing 1.6 ml of protein A-Sepharose at 4 C,
washed with 10 volumes of washing buffer (50 mM tris
buffer pH 8.6, 150 mM NaCl) and eluted in 0.5 ml frac-
tions with eluting buffer (93:7 100 mM citric acid:
100 mM sodium citrate). The first 9 fractions were
immediately neutralized with 0.1 ml of 2M tris buffer
pH 8.6 in each case and then combined, and the resulting
protein was transferred by three concentration/dilution
cycles in an Amicon microconcentrator (Centricon 30) into
TNE buffer (50 mM tris buffer pH 7.4, 50 mM NaCl, 1 mM
EDTA). The IL4RFc obtained in this way is pure by
SDS-PAGE electrophoresis (U.K. Lammli, Nature 227 (1970)
680-685). In the absence of reducing agents it behaves in
the SDS-PAGE like a dimer (about 150 KDa).
Biological activity of purified IL4RFc
IL4RFc proteins binds 125I-radiolabeled IL-4 with the same
affinity (Kp=0.5 nM) as membrane-bound intact IL-4 recep-
tor. It inhibits the proliferation of IL-4-dependent cell
line CTLLHuIL-4RI clone D (Idzerda et al., loc. cit.) in
concentrations of 10-1000 ng/ml. In addition, it is
outstandingly suitable for developing IL-4 binding assays
because it can be bound via its Fc part to microtiter
plates previously coated with, for example, rabbit anti-
human IgG, and in this form likewise binds its ligands
with high affinity.
Example 3: Erythropoieti.ra fusion proteins
Mature erythropoietin (EPO) is a glycoprotein which is
composed of 166 amino acids and is essential for the
development of erythrocytes. It stimulates the maturation
and the terminal differentiation of erythroid precursor
cells. The cDNA for human EPO has been cloned
(EP-A-0 267 678) and codes for the 166 amino acids of
mature EPO and a signal peptide of 22 amino acids which
is essential for secretion. The cDNA can be used to
- 13 - 2045869
prepare recombinant functional EPO in genetically
manipulated mammalian cells and the EPO can be employed
clinically for the therapy of anemic manifestations of
various etiologies (for example associated with acute
renal failure).
Because of the straightforward purification and the
improved pharmacokinetic properties, according to the
invention synthesis of EPO as immunoglobulin fusion
protein is particularly advantageous.
Construction of a hybrid plasmid pEPOFc coding for
erythropoietin fusion protein.
This construction was carried out in analogy to that
described in Example 2 (section: "Construction of a
hybrid plasmid pIL-4RFc coding for IL-4 receptor fusion
protein"). Two oligonucleotides able to hybridize with
sequences in the vicinity of the initiation codon
(A: 5'GATCGATCTCGAGATGGGGGTGCACGAATGTCCTGCCTGGCTGTGG 3')
and of the stop codon
(B: 5' CTGGAATCGGATCCCCTGTCCTGCAGGCCTCCCCTGTGTACAGC 3')
of the EPO cDNA cloned in the vector pCES (EP-A 0 267
678) were synthesized. Of these, oligonucleotide A is
partially homologous with the sequence of the coding
strand, and oligonucleotide B is partially homologous
with the non-coding strand; cf. Fig. 7. Amplification
with thermostable DNA polymerase results in a DNA frag-
ment (598 bp) which, based on the coding strand, contains
at the 5' end in front of the initiation codon an XhoI
site and in which at the 3' end the codon for the
penultimate C-terminal amino acid residue of the EPO
(Asp) is present in a BamHI recognition sequence. The
reading frame in the BamHI cleavage site is such that
ligation with the BamHI site in pCD4E gamma 1 results in
a gene fusion with a reading frame continuous from the
initiation codon of EPO cDNA to the stop codon of the
heavy chain of IgGl. The desired fragment was obtained
and, after treatment with XhoI and BamHI, ligated into
' 14 - 2045869
the vector pCD4E gamrna 1, described above, which had been
cut with XhoI/BamHI. The resulting plasmid was called
pEPOFc (Fig. 8).
