Biotherapy 3: 43-53, 1991. © 1991 Kluwer Academic Publishers. Printed in the Netherlands.
Chimeric antibody: Potential applications for drug delivery and immunotherapy Seung-Uon Shin UCLA, Department of Microbiology, The Molecular Biology Institute, 405 Hilgard Avenue, Los Angeles, CA 90024, USA Received 6 March 1990
Key words: cellular targeting, chimeric antibody, drug delivery, growth factors, growth factor receptors, immunotherapy Abstract
Antibodies, because of their inherent specificity, seem ideal agents for recognizing and destroying malignant cells. When monoclonal antibodies became available, they appeared ideal candidates for use as anti-cancer drugs. However, monoclonal antibodies as currently constituted still have certain inherent limitations. Transfectomas provide an approach to overcoming some of these limitations. Genetically engineered antibodies can be expressed following gene transfection into lymphoid cells. One of the major advantages of expressing genetically engineered antibodies, is that one is not limited to using antibodies as they occur in nature. In particular, non-immunoglobulin sequences can be joined to antibody sequences creating multi-functional chimeric antibodies. Creation of a family of multi-functional chimeric antibodies with a growth factor joined to a combining specificity may be useful in targeting therapy to malignant cells and delivering drugs into specific locales in the human body. Presence of the growth factor may facilitate transcytosis of chimeric antibody across the blood-brain barrier using growth factor receptors. These novel chimeric antibodies constitute a new family of immunotherapeutic molecules for cancer therapy.
Abbreviations: ADCC: Antibody-Dependent Cell-mediated Cytotoxicity; D: diversity region; DNS: dansyl; H: heavy chain; J: joint region; L: light chain; Mab: monoclonal antibody; PCR: Polymerase-catalyzed Chain Reaction; V: variable region.
Introduction
The challenge in cancer therapy has been to find a means of selectively killing the malignant cells while leaving the normal cells intact. Traditional chemotherapy has been directed against actively dividing cells and as
such kills normal cells, such as those of the bone marrow, which are actively proliferating. Antibodies, because of their remarkable specificity, have long had appeal as the "magic bullet" which will selectively identify and eliminate malignant cells. The challenge is to generate antibodies which specifically
44 distinguish between tumor and normal cells and to produce them in a form which will kill the tumor cells. The ability to produce monoclonal antibodies (Mabs) [1-3] made it easier to produce antibodies which are effective anti-cancer drugs. Although Mabs have successfully treated drug toxicity [4, 5], kidney transplant rejection [6], autoimmune desease [7], and have facilitated bone marrow transplantation [8], Mabs as currently constituted still have certain limitations. Firstly, while it is easy to produce rodent monoclonal antibodies, it still remains difficult to produce human monoclonals; in addition, the majority of the human monoclonals produced are of the IgM isotype, which makes them difficult to purify and limits their in vivo application. Rodent antibodies, because of their inherent immunogenicity, have only limited application in vivo. A second difficulty has been producing antibodies which are truly tumor specific and which are effectively targeted to the maligant cells. In addition, Mabs frequently induce the disappearance or modulation of their target antigen on the surface of target cells. A third limitation has been that rodent antibodies are not effective in recruiting the human immune system and effector functions. Transfectomas which express genetically engineered antibody genes provide one approach to overcoming some to the limitations inherent in Mabs. Human constant regions can be expressed joined to mouse variable regions [9-11]; these molecules make accessible all of the available mouse combining specificities and have reduced immunogenicity [12] and an increased ability to interact with the human effector systems [13]. An approach to make the genetically engineered antibodies even more human is to substitute the mouse complementarity determining regions into human framework [14, 15]; however, whether this will in fact be necessary to reduce immunogenicity has not yet been determined. The ability to express genetically
engineered antibodies also provides one with the ability to isotype switch the few available human antibodies so that classes in addition to IgM can be produced. One of the advantages of expressing genetically engineered antibodies, is that one is not limited to using antibodies as they occur in nature. Constant regions can be mutated to enhance or eliminate certain effector functions. In addition, antibody combining specificities can be joined with other properties not usually found in antibodies. These include the products of oncogenes [16]; tissue plasminogen activator [17], and growth factors [ 18].
Strategy for the expression of chimeric antibodies One strategy for expression of transfected immunoglobulin genes is to use the rearranged immunoglobulin genes with their own promotor and regulatory sequences for expression in plasma cells. A functional immunoglobulin gene is generated through somatic rearrangement of multiple DNA gene segments: V and J segments for light chain and V, D and J segments for heavy chain. The sequences encoding the variable region can be obtained from genomic [19], eDNA [20, 21], or PCR amplified clones [22]. Genomic clones have the advantage that an immunoglobulin promotor is adjacent to the variable region and can be used for its transcription. However, it is generally easier to obtain eDNA or PCR clones than genomic clones because primers specific for the J region or the constant region joined to the variable region and for the leader sequence region can be used to enrich for the desired variable region. However, eDNA or PCR cloned variable regions must be joined to a suitable promotor before they can be expressed. Once the desired variable region genes are cloned, they must be joined to the appropri-
45
A.
s Human
CHI H CH2
CH3
Human
Variable region cassettes R
B
L
V
DJ
R
IgG1I
s H u m a n IgG2 I.... S Human
B.
Constant region cassettes
B
Mouse V Region
B
IgG31 s
R
R
B
Mouse r V V RegionL L
s
R
/
IgG4 I
\ R
R
e
A m p ~ c H
I Human
pSV2AHgpt ~ Ampr
C Region E.colJ-gpt SV40
pBR322 ori
S
V
4 ~ E.coli-gpt
CH3
Fig. 1. Cloning cassettes for the human constant region genes (A) and the murine variable region genes (B). A. The human IgG constant region genes are flanked by unique Sal I(S) and Barn HI(B) restriction enzyme sites which facilitate isotype switching between chimeric genes. B. By using the EcoRI sites flanking cloned variable region genes, the specificity of the chimeric antibody can be changed, Thus these cloning cassettes facilitate the isotype switching of the human lgG and the diversification of antigen binding specificity in the pSV2AHgpt transfection vectors.
ate constant region or non-immunoglobulin sequence. The general strategy for constructing mouse/human chimeric genes has been to create cloning cassettes (Fig. 1). These facilitate the interchange of variable and constant region gene segments and make it relatively easy to produce a wide spectrum of antibodies with different combining specificities and effector functions. The genetically engineered immunoglobulin genes can be expressed in an immunoglobulin non-producing myeloma by gene transfection. Since gene transfection is an inetiicient process, only a small number of the myeloma cells receiving foreign DNA stably express the exogenous DNA ( < 10-3). Including dominant selectable markers in the
DNA sequences used for transfection makes it feasible to isolate the rare stable transfectomas from among the many non-transfected cells. The most commonly used transfection vector is pSV2 (Fig. 1) which contains a plasmid origin of replication (pBR322 origin of replication), a prokaryotic selectable marker (B" lactamase gene), and a marker selectable in eukaryotic cells [23-25]. One of these eukaryotic selectable markers is the gpt gene encoding xanthine-guanine phosphoribosyltransferase. This enzyme allows cells supplied with xanthine to survive in the presence of mycophenolic acid which blocks purine biosynthesis. A second selectable gene is the neo gene derived from Tn5 which encodes a
46
HEAVY CHAIN GENE !
LIGHT CHAIN GENE
Mouse
L
|
VH
House[
pBRo3r22~pSV2AHgpt ~cHH2|Human
~
.~--~L
Human
pSVI84AHneo
E.Icoli-gpt
ELECTROPORATION
TRANSFORMATION y
BACTERIUM
TRANSFECTOMA CHIMERICANTIBODY
I
II
MOUSE V REGION
HUMANC REGION
I
Fig. 2. Schematic diagram of gene transfection and selection. The mouse/human chimeric heavy chain gene was cloned into pSV2AHgpt vector and the mouse/human chimeric light chain gene into the pSV184AHneo vector. Bacteria transformed with both transfection vectors were selected with Ampieillin (Am) and Chlorampenicol (Cm) and used to transfect both vectors into immunoglobulin non-producing myeloma cells by protoplast fusion. Both vectors could also be introduced into the non-producing myeloma by electroporation. Stable transfectomas are selected with either G418 or hypoxanthine/mycophenolic acid/xanthine and the desired transfectomas which produce functional chimeric antibodies identified by ELISA.
47
phosphotransferase that inactivates G418, an inhibitor of protein synthesis in eukaryotes. An additional transfection vector (pSV184) was developed [26], which uses the origin of replication from the plasmid pACYC184 and the chloramphenicol-resistance gene [Fig. 2]. The pBR and the pACYC origins of replication are compatable enabling the pBR and pACYC vectors to replicate simultaneously in Escherichia coli host cells; selection with chloramphenicol and ampicillin guarantees that both plasmids are present. Both the heavy chain gene in the pSV2 vector with a gpt eukaryotic selectable marker and the light chain gene in the pSV184 vector with a neo eukaryotic selectable marker [Fig. 2] can be introduced into a bacterium and then simultaneously transferred into the same immunoglobulin non-producing myeloma cell by protoplast fusion [26,27]. The overall strategy is illustrated in Fig. 2. Protoplast fusion and electroporation now routinely yield transfection frequencies of more than 10 -3 [27]. The stable transfectomas can be isolated and characterized by using an enzyme-linked immunosorbent assay (ELISA) and Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis (SDS-PAGE). The ELISA can be used to easily identify transfectomas producing and secreting the desired protein and to quantitate the level of production. SDSPAGE of proteins immunoprecipitated from cytoplasmic lysates of transfectomas will show which proteins are produced and their assembly pattern. Analysis of immunoprecipitates from secretions will show what the size of the secreted product is. SDS-PAGE of immunoprecipitates treated with disulfide breaking reagents will give the molecular weight of the individual polypeptides. For example, the cytoplasmic mouse/human IgG3 chimeric antibody assembles by the pathways H+H~H2~H2L~H2L2 and H+L--* HL-*H2L2; only fully assembled H2L2 molecules are secreted into the culture supernatant (Fig. 3).
t~ cD m
U m
IS
tv3 Ci~
03
-C3
.I
t--
. U
e3 CD r~ ill FLad rr" U
ill cO
"03'
KDa -92.5
H2L2-
H_
-69
H2L-46
H2HL-
H
L_
~
-30
......... - 2 1 . 5 Reduced
Nonreduced Fig. 3. SDS-PAGE analysis of the synthesis, assembly, and secretion of chimeric antibodies. The cytoplasmic and secreted chimeric antibodies (mouse/human IgG3) were biosynthetically labeled with 35S-methionine. The synthesis, assembly, and secretion patterns were analyzed on a 5% Phosphate gel; after reduction 10% Tris-glyein gels were used to determine the size of the chimeric heavy and light chains.
The levels of production of chimeric immunoglobulins by transfectomas is less than that of certain mouse myelomas and hybridomas which secrete up to 200/~g of antibody/ ml of culture supernatant. However, transfectomas usually secrete 1-30 vg of chimeric antibody/ml [28, 29], similar to the quantities produced by human hybridomas and low producing murine hybridomas. It remains unclear why the production level is less than that seen in the hybridomas or myelomas from which the variable region genes were cloned.
48 Characterization of chimeric antibodies and applications in immunotherapy
mediated hemolysis [31] (Table 1). It is n o t clear why there is no exact correlation between C lq binding and complement-mediated hemolysis and why the IgG3 molecules used in the two studies show variability in their effectivenss. Only IgG1 and IgG3 exhibit ADCC with IgG1 being more effective than IgG3. The molecular structure mediating biological effector functions of chimeric antibody molecules has been systematically studied with a family of anti-DNS chimeric antibodies [31, 33]. The ability to activate complement can be correlated with segmental flexibility [31]; that is, more flexible molecules (IgGl and IgG3) activate complement more effectively than rigid molecules (IgG2 and IgG4) (Table 1). Using chimeric anti-DNS antibodies, it has been shown that IgG1 and IgG3 bind with high affinity ( ' ~ 1 0 9 M - l ) t o human FcyRI. IgG4 binds with reduced affinity ( -~ 10s M -1) while binding by IgG2 is not detected (
Chimeric antibody: Potential applications for drug delivery and immunotherapy Seung-Uon Shin UCLA, Department of Microbiology, The Molecular Biology Institute, 405 Hilgard Avenue, Los Angeles, CA 90024, USA Received 6 March 1990
Key words: cellular targeting, chimeric antibody, drug delivery, growth factors, growth factor receptors, immunotherapy Abstract
Antibodies, because of their inherent specificity, seem ideal agents for recognizing and destroying malignant cells. When monoclonal antibodies became available, they appeared ideal candidates for use as anti-cancer drugs. However, monoclonal antibodies as currently constituted still have certain inherent limitations. Transfectomas provide an approach to overcoming some of these limitations. Genetically engineered antibodies can be expressed following gene transfection into lymphoid cells. One of the major advantages of expressing genetically engineered antibodies, is that one is not limited to using antibodies as they occur in nature. In particular, non-immunoglobulin sequences can be joined to antibody sequences creating multi-functional chimeric antibodies. Creation of a family of multi-functional chimeric antibodies with a growth factor joined to a combining specificity may be useful in targeting therapy to malignant cells and delivering drugs into specific locales in the human body. Presence of the growth factor may facilitate transcytosis of chimeric antibody across the blood-brain barrier using growth factor receptors. These novel chimeric antibodies constitute a new family of immunotherapeutic molecules for cancer therapy.
Abbreviations: ADCC: Antibody-Dependent Cell-mediated Cytotoxicity; D: diversity region; DNS: dansyl; H: heavy chain; J: joint region; L: light chain; Mab: monoclonal antibody; PCR: Polymerase-catalyzed Chain Reaction; V: variable region.
Introduction
The challenge in cancer therapy has been to find a means of selectively killing the malignant cells while leaving the normal cells intact. Traditional chemotherapy has been directed against actively dividing cells and as
such kills normal cells, such as those of the bone marrow, which are actively proliferating. Antibodies, because of their remarkable specificity, have long had appeal as the "magic bullet" which will selectively identify and eliminate malignant cells. The challenge is to generate antibodies which specifically
44 distinguish between tumor and normal cells and to produce them in a form which will kill the tumor cells. The ability to produce monoclonal antibodies (Mabs) [1-3] made it easier to produce antibodies which are effective anti-cancer drugs. Although Mabs have successfully treated drug toxicity [4, 5], kidney transplant rejection [6], autoimmune desease [7], and have facilitated bone marrow transplantation [8], Mabs as currently constituted still have certain limitations. Firstly, while it is easy to produce rodent monoclonal antibodies, it still remains difficult to produce human monoclonals; in addition, the majority of the human monoclonals produced are of the IgM isotype, which makes them difficult to purify and limits their in vivo application. Rodent antibodies, because of their inherent immunogenicity, have only limited application in vivo. A second difficulty has been producing antibodies which are truly tumor specific and which are effectively targeted to the maligant cells. In addition, Mabs frequently induce the disappearance or modulation of their target antigen on the surface of target cells. A third limitation has been that rodent antibodies are not effective in recruiting the human immune system and effector functions. Transfectomas which express genetically engineered antibody genes provide one approach to overcoming some to the limitations inherent in Mabs. Human constant regions can be expressed joined to mouse variable regions [9-11]; these molecules make accessible all of the available mouse combining specificities and have reduced immunogenicity [12] and an increased ability to interact with the human effector systems [13]. An approach to make the genetically engineered antibodies even more human is to substitute the mouse complementarity determining regions into human framework [14, 15]; however, whether this will in fact be necessary to reduce immunogenicity has not yet been determined. The ability to express genetically
engineered antibodies also provides one with the ability to isotype switch the few available human antibodies so that classes in addition to IgM can be produced. One of the advantages of expressing genetically engineered antibodies, is that one is not limited to using antibodies as they occur in nature. Constant regions can be mutated to enhance or eliminate certain effector functions. In addition, antibody combining specificities can be joined with other properties not usually found in antibodies. These include the products of oncogenes [16]; tissue plasminogen activator [17], and growth factors [ 18].
Strategy for the expression of chimeric antibodies One strategy for expression of transfected immunoglobulin genes is to use the rearranged immunoglobulin genes with their own promotor and regulatory sequences for expression in plasma cells. A functional immunoglobulin gene is generated through somatic rearrangement of multiple DNA gene segments: V and J segments for light chain and V, D and J segments for heavy chain. The sequences encoding the variable region can be obtained from genomic [19], eDNA [20, 21], or PCR amplified clones [22]. Genomic clones have the advantage that an immunoglobulin promotor is adjacent to the variable region and can be used for its transcription. However, it is generally easier to obtain eDNA or PCR clones than genomic clones because primers specific for the J region or the constant region joined to the variable region and for the leader sequence region can be used to enrich for the desired variable region. However, eDNA or PCR cloned variable regions must be joined to a suitable promotor before they can be expressed. Once the desired variable region genes are cloned, they must be joined to the appropri-
45
A.
s Human
CHI H CH2
CH3
Human
Variable region cassettes R
B
L
V
DJ
R
IgG1I
s H u m a n IgG2 I.... S Human
B.
Constant region cassettes
B
Mouse V Region
B
IgG31 s
R
R
B
Mouse r V V RegionL L
s
R
/
IgG4 I
\ R
R
e
A m p ~ c H
I Human
pSV2AHgpt ~ Ampr
C Region E.colJ-gpt SV40
pBR322 ori
S
V
4 ~ E.coli-gpt
CH3
Fig. 1. Cloning cassettes for the human constant region genes (A) and the murine variable region genes (B). A. The human IgG constant region genes are flanked by unique Sal I(S) and Barn HI(B) restriction enzyme sites which facilitate isotype switching between chimeric genes. B. By using the EcoRI sites flanking cloned variable region genes, the specificity of the chimeric antibody can be changed, Thus these cloning cassettes facilitate the isotype switching of the human lgG and the diversification of antigen binding specificity in the pSV2AHgpt transfection vectors.
ate constant region or non-immunoglobulin sequence. The general strategy for constructing mouse/human chimeric genes has been to create cloning cassettes (Fig. 1). These facilitate the interchange of variable and constant region gene segments and make it relatively easy to produce a wide spectrum of antibodies with different combining specificities and effector functions. The genetically engineered immunoglobulin genes can be expressed in an immunoglobulin non-producing myeloma by gene transfection. Since gene transfection is an inetiicient process, only a small number of the myeloma cells receiving foreign DNA stably express the exogenous DNA ( < 10-3). Including dominant selectable markers in the
DNA sequences used for transfection makes it feasible to isolate the rare stable transfectomas from among the many non-transfected cells. The most commonly used transfection vector is pSV2 (Fig. 1) which contains a plasmid origin of replication (pBR322 origin of replication), a prokaryotic selectable marker (B" lactamase gene), and a marker selectable in eukaryotic cells [23-25]. One of these eukaryotic selectable markers is the gpt gene encoding xanthine-guanine phosphoribosyltransferase. This enzyme allows cells supplied with xanthine to survive in the presence of mycophenolic acid which blocks purine biosynthesis. A second selectable gene is the neo gene derived from Tn5 which encodes a
46
HEAVY CHAIN GENE !
LIGHT CHAIN GENE
Mouse
L
|
VH
House[
pBRo3r22~pSV2AHgpt ~cHH2|Human
~
.~--~L
Human
pSVI84AHneo
E.Icoli-gpt
ELECTROPORATION
TRANSFORMATION y
BACTERIUM
TRANSFECTOMA CHIMERICANTIBODY
I
II
MOUSE V REGION
HUMANC REGION
I
Fig. 2. Schematic diagram of gene transfection and selection. The mouse/human chimeric heavy chain gene was cloned into pSV2AHgpt vector and the mouse/human chimeric light chain gene into the pSV184AHneo vector. Bacteria transformed with both transfection vectors were selected with Ampieillin (Am) and Chlorampenicol (Cm) and used to transfect both vectors into immunoglobulin non-producing myeloma cells by protoplast fusion. Both vectors could also be introduced into the non-producing myeloma by electroporation. Stable transfectomas are selected with either G418 or hypoxanthine/mycophenolic acid/xanthine and the desired transfectomas which produce functional chimeric antibodies identified by ELISA.
47
phosphotransferase that inactivates G418, an inhibitor of protein synthesis in eukaryotes. An additional transfection vector (pSV184) was developed [26], which uses the origin of replication from the plasmid pACYC184 and the chloramphenicol-resistance gene [Fig. 2]. The pBR and the pACYC origins of replication are compatable enabling the pBR and pACYC vectors to replicate simultaneously in Escherichia coli host cells; selection with chloramphenicol and ampicillin guarantees that both plasmids are present. Both the heavy chain gene in the pSV2 vector with a gpt eukaryotic selectable marker and the light chain gene in the pSV184 vector with a neo eukaryotic selectable marker [Fig. 2] can be introduced into a bacterium and then simultaneously transferred into the same immunoglobulin non-producing myeloma cell by protoplast fusion [26,27]. The overall strategy is illustrated in Fig. 2. Protoplast fusion and electroporation now routinely yield transfection frequencies of more than 10 -3 [27]. The stable transfectomas can be isolated and characterized by using an enzyme-linked immunosorbent assay (ELISA) and Sodium Dodecyl Sulfate-Polyacrylamide Gel Electrophoresis (SDS-PAGE). The ELISA can be used to easily identify transfectomas producing and secreting the desired protein and to quantitate the level of production. SDSPAGE of proteins immunoprecipitated from cytoplasmic lysates of transfectomas will show which proteins are produced and their assembly pattern. Analysis of immunoprecipitates from secretions will show what the size of the secreted product is. SDS-PAGE of immunoprecipitates treated with disulfide breaking reagents will give the molecular weight of the individual polypeptides. For example, the cytoplasmic mouse/human IgG3 chimeric antibody assembles by the pathways H+H~H2~H2L~H2L2 and H+L--* HL-*H2L2; only fully assembled H2L2 molecules are secreted into the culture supernatant (Fig. 3).
t~ cD m
U m
IS
tv3 Ci~
03
-C3
.I
t--
. U
e3 CD r~ ill FLad rr" U
ill cO
"03'
KDa -92.5
H2L2-
H_
-69
H2L-46
H2HL-
H
L_
~
-30
......... - 2 1 . 5 Reduced
Nonreduced Fig. 3. SDS-PAGE analysis of the synthesis, assembly, and secretion of chimeric antibodies. The cytoplasmic and secreted chimeric antibodies (mouse/human IgG3) were biosynthetically labeled with 35S-methionine. The synthesis, assembly, and secretion patterns were analyzed on a 5% Phosphate gel; after reduction 10% Tris-glyein gels were used to determine the size of the chimeric heavy and light chains.
The levels of production of chimeric immunoglobulins by transfectomas is less than that of certain mouse myelomas and hybridomas which secrete up to 200/~g of antibody/ ml of culture supernatant. However, transfectomas usually secrete 1-30 vg of chimeric antibody/ml [28, 29], similar to the quantities produced by human hybridomas and low producing murine hybridomas. It remains unclear why the production level is less than that seen in the hybridomas or myelomas from which the variable region genes were cloned.
48 Characterization of chimeric antibodies and applications in immunotherapy
mediated hemolysis [31] (Table 1). It is n o t clear why there is no exact correlation between C lq binding and complement-mediated hemolysis and why the IgG3 molecules used in the two studies show variability in their effectivenss. Only IgG1 and IgG3 exhibit ADCC with IgG1 being more effective than IgG3. The molecular structure mediating biological effector functions of chimeric antibody molecules has been systematically studied with a family of anti-DNS chimeric antibodies [31, 33]. The ability to activate complement can be correlated with segmental flexibility [31]; that is, more flexible molecules (IgGl and IgG3) activate complement more effectively than rigid molecules (IgG2 and IgG4) (Table 1). Using chimeric anti-DNS antibodies, it has been shown that IgG1 and IgG3 bind with high affinity ( ' ~ 1 0 9 M - l ) t o human FcyRI. IgG4 binds with reduced affinity ( -~ 10s M -1) while binding by IgG2 is not detected (
