Bioconjugate Chem. 1993, 4,230-235
230
Biological Activity and in Vivo Clearance of Antitumor Antibody/ Cytokine Fusion Proteins Stephen D. Gillies,’,t Delano Young,t*sKin-Ming LO,~Y~I and Stanley Roberts1 Abbott Biotech, Inc., 119 Fourth Avenue, Needham Heights, Massachusetts 02194, and Pharmaceutical Products Division, Abbott Laboratories, Abbott Park, Illinois 60064. Received December 30, 1992 Several human cytokines including IL-2, GM-CSF, and tumor necrosis factors a and B were engineered as fusion proteins to the carboxyl terminus of a chimeric anti-gangliosideantibody, ch14.18, and expressed in transfected hybridoma cells. All of the fusion proteins were expressed at high levels and were easily purified by affinity or ion-exchange chromatography from culture supernatants. The effect of fusion on antigen binding activity was tested and found to vary with the particular cytokine. No significant decreases in antigen binding were observed, and fusion of IL-2 had the greatest positive effect in a direct antigen binding assay. All fusion proteins maintained normal levels of biological activity except for GM-CSF, which was approximately 20% active, compared to recombinant GM-CSF produced in bacteria. The clearance of the fusion proteins was examined in normal Balb/c mice after intraperitoneal injection or in athymic (nu/nu) mice after intravenous injection and was generally quite rapid, relative to ch14.18. This was mainly due to a very rapid initial clearance rate (aphase) since the half-lives of the phase of the fusion proteins (about 30 h) were comparable to that of the free antibody (about 58 h). These results demonstrate that biologically active antibody/cytokine fusion proteins can be constructed by genetic engineering. Their relatively rapid clearance may require constant infusion rather than bolus injection in order to achieve clinical efficacy.
INTRODUCTION
The ability of many cytokines to elicit cellular immune responses to tumor cells has stimulated a renewed interest in cancer vaccines. Many groups have shown that tumor cells transfected with genes encoding various cytokines are rejected upon transplantation in syngeneic animal systems and that tumor-specific cellular immunity can be demonstrated (1-4). The clinical application of such an approach would involve the isolation, transfection, and readministration of a patient’s tumor cells-a process that would be both time consuming and costly. An alternative approach would be to target cytokines to tumor sites using the binding specificity of a tumorspecificantibody. Such an approach would not be patientspecific and would involve a simple injection or infusion of the Ig-cytokine fusion protein. Successful therapy might be achieved with only a small percentage of the tumor mass being targeted by the fusion protein since this would serve as the immunogen that would generate a specific, and hopefully long-lasting, cellular response. We have already demonstrated ( 5 , 6 )that the human cytokines lymphotoxin (TNFB)and IL-2 can be engineered as fusion proteins to chimeric mouse/human antibody 14.18 (ch14.18), which is reactive with the ganglioside GD2 expressed on most neuroblastoma and many melanoma tumors (7,8). In both cases the antigen binding and cytokine functions were maintained. Such proteins do not have direct cytotoxic effects on GD2-positive tumor cells in vitro but may serve to elicit cellular responses *Address correspondence to Dr. Stephen D. Gillies at his present address: Fuji ImmunoPharmaceuticals, Corp. 125Hartwell Avenue, Lexington, MA 02173. + Abbott Biotech. Inc. Abbott Laboratories. 8 Present address: Transkaryotic Therapies Inc., 195 Albany St., Cambridge, MA 02139. 11 Present address: Fuji ImmunoPharmaceuticals, Corp. 125 Hartwell Avenue, Lexington, MA 02173.
*
1043-1802/93/290~0230$04.00/0
through their various inflammatory and proliferative activities in vivo. As a first step in evaluating these compounds as potential antitumor agents,we have examined their pharmacokinetic properties in normal Balb/c and athymic (nu/nu) mice. Two additional Ig-cytokine fusion proteins have also been constructed and tested for both function and blood clearance. These include fusion of the ch14.18 Ig to either human GM-CSF1 or TNFa. Results indicate that most of these fusion proteins are cleared quite rapidly from the circulation with the exception of the GM-CSF construct. The potential role of altered antibody structure in determining the rate of clearance is discussed. EXPERIMENTAL PROCEDURES
Plasmid Construction. The Ig-TNFB and Ig-IL-2 fusion protein expression plasmids have been described (56).The TNFa sequence was cloned from IFy-treated U937 cells using PCR. The 5’ sense primer contained a SmaI site for fusion of the mature TNFa sequence to the SmaI site at the end of the human Cy1 gene. The carboxyl terminal Lys residue of the H chain (which follows the Pro, Gly encoded by the SmaI site) was also encoded in the primer and immediately preceded the first residue of TNFa. The 3’ antisense primer places an XhoI site just after the translation stop codon. The PCR product was subcloned, sequenced, and ligated as a SmaI-XhoI fragment to a HindIII-SmaI fragment of the Cy1 gene and cloned as a HindIII-XhoI insert. The GM-CSF sequence was chemically synthesized utilizing a coding bias for high GC content. Again the mature sequence was fused directly to the terminal Lys residue of the Cy1 gene using the SmaI site and subcloned with the Cy1 gene fragment as Abbreviations used: ADCC, antibody-dependent cellular cytotoxicity; CDC, complement-dependent cytotoxicity; GMCSF, granulocyte-monocyte colony stimulating factor; ip, intraperitoneal; iv, intravenous; MTX, methotrexate; PCR, polymerase chain reaction. 0 1993 American Chemical Society
Ig-Cytokine Fusion Proteins
a HindIII-XhoI insert. These inserts were subsequently excised and inserted into the antibody expression vector pdHL2 containing the V regions of the anti-GD2 antibody 14.18 (9). Cell Culture and Transfection. Sp2/0 Ag14 mouse hybridoma cells were maintained and transfected as described (IO). Drug selection in methotrexate (MTX) was at an initial concentration of 0.1 pM. Transfectants secreting fusion proteins were detected by ELISA (IO) and adapted to growth in medium containing higher levels of MTX (1pM). Purification and Analysis of Fusion Proteins. Fusion proteins were purified from spent culture medium using either protein A Sepharose (Repligen, Cambridge, MA) or ABx Bakerbond (J.T. Baker, Phillipsburg, NJ) ion-exchangechromatographyas described (5). The latter method was used for the LT fusion protein in order to prevent loss of biological activity during elution from protein A at low pH. Purified proteins were analyzed by SDS gel electrophoresis in the absence or presence of 5% 2-mercaptoethanolas indicated. In some cases the proteins (50 pg/mL) were first incubated with plasmin (0.1 casein units/mL) at 37 "C for 1h in PBS containing 50 mM Tris, pH 8. The digestion products were concentrated by adsorption to protein A Sepharose and elution into gel sample buffer. The ch14.18, as well as the IL-2 and GMCSF fusion proteins, were examined by FPLC size exclusion chromatography and no significant aggregation was detected (data not shown). Antigen Binding Assay. A direct antigen binding assay was performed on GD2-coated microtiter plates as described (9),except that the GD2 was prepared from M21 human melanoma cells, which express high levels of the antigen. As a result, the assay was more sensitive than previously reported (6). Biological Activity Assays. The cytostatic assay for TNFp and TNFa has been described (5). IL-2 activity was measured in a mouse T cell proliferation assay (11). GM-CSF activity was measured in a proliferation assay using the GM-CSF-dependent acute myelogenous leukemia cell line, AML-193 (obtained from D. Santoli, Wistar Institute). After 2 days of growth in serum-free medium containing insulin and transferrin, dilutions of the test sample or recombinant GM-CSF standard (Collaborative Research, Lexington, MA) were added to the wells. After an additional 5 days, 5 pCi of [3H]thymidine was added, and after 16 h, the radioactivity precipitated in 10% trichloroacetic acid was collected on GF/C filters and counted by liquid scintillation. In Vivo Clearance Studies. Chimeric 14.18 antibody and cytokine fusion proteins were diluted to 50 pg/mL in sterile PBS and injected (0.5 mL/animal, two animals per sample) ip into normal Balb/C mice (Taconic Farms, Germantown, NY). Samples were collected from the tail vein at 2 and 24 h and refrigerated overnight. After spinning for 10min in a microcentrifugeto remove clotted material, sera were assayed by ELISA for human Ig determinants (IO). The IL-2 and GM-CSF fusion proteins (and control ch14.18) were studied further by iv injection (1mg/kg in the tail vein) of athymic (nu/nu) mice using three mice per group. Samples were collected from the orbital sinus using microcapillary tubes and were stored at -20 "C prior to analysis as described above. RESULTS
Expression of Cytokine Fusion Proteins. The fusion proteins between the anti-GD2 chimeric antibody ch14.18 and human TNFp and IL-2 have already been described
Bioconjugate Chem., Vol. 4, No. 3, 1993 231 1
2
M
-
1
R
94 69 43
NR
Figure 1. Analysis of chl4.18-GM-CSF fusion protein chain assembly. Proteins were diluted in gel sample buffer and boiled without (NR) or with (R) reduction (5% 2-mercaptoethanol) and analyzed on an 8%polyacrylamide gel containing SDS. Purified ch14.18 antibody (lanes 2) was compared to chl4.18-GM-CSF fusion protein (lanes 1). (5,6). The mature TNFP or IL-2 protein sequences were directly fused to the carboxyl terminus of the CH3 exon of the human C r l gene. The proteins secreted by transfectants, obtained with these constructs, were shown to be assembled into complete antibody molecules with the cytokines peptide-linked to the Ig H chains. In both cases the cytokines retained their biological activities. Two more antibody/cytokine fusion proteins were constructed for potential use in targeting cytokines to solid tumors. In the first case, a TNFa fusion protein was made so that its effect in vivo could be compared to that of the TNFp construct. TNFa has similar biological activities to TNFp but has a more potent inflammatory effect (12). The second construct was a fusion with GM-CSF, a protein that is very different, structurally, from either IL-2 or the TNFs and has been shown to enhance the effector activity (ADCC) mediated by antibodies including ch14.18 (13). Both H chain gene fusions were expressed together with the 14.18 V regions and human CKusing plasmid pdHL2 (9). Transfectants of the hybridoma line Sp2/0 Ag14 were selected and tested for secretion of human antibody determinants by ELISA. The expression levels of the transfectants were comparable to those obtained for the ch14.18 antibody at the initial concentration of MTX used for selection. These levels were increased further by growing cells in increasing concentrations of MTX. Cell cultures growing in media containing intermediate levels of MTX (1pM) were used for fusion protein purification by adsorption to and elution from protein A Sepharose. Characterizationof Fusion Proteins. Purified fusion proteins were analyzed for chain compositionand assembly by SDS-PAGE in the presence or absence of reducing agent. As seen in Figure 1,the unreduced GM-CSF fusion protein migrated as a singlemolecular specieswith a higher apparent molecular weight than the ch14.18 antibody. Upon reduction, this structure dissociatedinto the normal L chain and a fusion H chain with an apparent molecular weight consistent with the fusion of GM-CSF ( 75 kDa). A similar result was obtained with the TNFa construct (not shown). The assembly of IL-2 and TNFp fusion proteins was reported earlier (5, 6). Next we tested the effect that the fusion of cytokines might have on the antigen binding activity of the antibody. Direct binding assays using GD2-coated plates were used to compare the fusion proteins to ch14.18 (Figure 2). As N
Giliies et ai.
Biocon/ugate Chem., Vol. 4, No. 3, 1993
232
0.8
'gb -8
0.6
D
.-
5
0.4
0.2
0 0
100
200
300
Antibody Conc. (ng/ml)
Figure 2. Antigen binding activities of Ig-cytokine fusion proteins. Serial dilutions of purified Ig-cytokine fusion proteins were compared to ch14.18 in a direct antigen binding ELISA using GD2-coated plates for capture and HRP-conjugated antihuman K antiserum for detection. 12
-
10
--
TNF-a Conc. (pM)
h
e
TNF-a
$
........0........
ch14.18-TNF-a
m X
Figure 4. Activity of chl4.18-TNFa in a mouse L cell cytostatic assay. Dilutions of purified chl4.18-TNFa and recombinant TNFa were tested in a 2-day cytostatic assay using a TNFsensitive mouse L 929 cell line. Monolayers were stained with crystal violet and the amount of stain retained by each well was measured and compared to cells not receiving any sample (zerodose control).
6 v
-zg
4 -
2
-.
0 4
1
I
10
100
1000
10000
GMCSF Conc. (pg/ml)
Figure 3. Proliferation activity assay of GM-CSF and ch14.18GM-CSF fusion protein. Dilutions of purified Ig-GM-CSF fusion protein and recombinant GM-CSF were compared in a proliferation assay using a human GM-CSF-dependent acute myelogenous leukemia cell line. Values were normalized to the content of GM-CSF in the fusion protein.
reported earlier, the fusion of IL-2 resulted in a marked increase in antigen binding, and the fusion of TNFB caused a small increase, as measured by both direct and competitive binding (5,6). The fusion of TNFa caused, as in the case of TNFP, a slight but reproducible increase in binding while the fusion of GM-CSF caused a slight reduction. Biological Activities of Fusion Proteins. The GMCSF activity of the Ig-GM-CSF construct was tested in a proliferation assay using a GM-CSF-dependent acute myelogenous leukemic cell line (Figure 3). When compared to recombinant GM-CSF (produced in bacteria) on a molar basis, the fusion protein required approximately 5 times the concentration to achieve the same level of proliferation when assayed in the range of 100-800 pg/ mL. At around 1ng/mL, however, the activity of the IgGM-CSF approached that of the recombinant GM-CSF. Very little difference was seen between the activity of the GM-CSF fusion protein in the transfected cell culture media and that purified on protein A Sepharose, indicating that the low pH used for elution does not significantly reduce activity (not shown). The TNFa activity of the Ig-TNFa construct was tested in a standard cytostatic assay using a mouse L cell line. The fusion protein was found to be as active as the TNFa standard and to be somewhat more active at the higher
concentrations (Figure 4). Unlike the Ig-TNFB fusion protein (5), the activity of the Ig-TNFa fusion protein was not reduced by the acidic pH used for elution from protein A. In Vivo Clearanceof CytokineFusion Proteins. We initially used ip injection to estimate the rate of clearance in vivo. Purified fusion proteins were injected ip into Balblc mice. No adverse effects were seen for any of the constructs. Most of the fusion proteins were rapidly cleared to approximately 10% of the serum concentration of the control 14.18 antibody by 2 h. The one exception was the GM-CSF fusion protein, which appeared to be intermediate in its rate of clearance and even at 24 h was found at concentrations that were 3-5 times those of the other constructs (data not shown). We then examined the clearance of the IL-2 and GMCSF fusion proteins more rigorously using iv injection of athymic nu/nu mice. This eliminates the time required for the proteins to enter the bloodstream from the peritoneum and, since nude mice were used, likely reduces the number of IL-2 receptor-bearing cells in the circulation. In this case three mice were used per group and each serum was assayed separately so that a range between different mice could be established. The results (Figure 5) show that the IL-2 fusion protein, despite its very large size (200 m a ) , was rapidly removed from serum. In contrast, the GM-CSF fusion protein had an intermediate clearance rate (Figure 5). A comparison of the half-life ( t l $ of the a and B phases (Table I) showed that the B phase (elimination phase) t l p values of the free antibody and the two fusion proteins are very similar;the large difference in their catabolic rates are mainly due to the a phase (distribution phase) t l j z . This suggests that an intravascular event, which removed the bulk of the fusion proteins
Ig-Cytoklne Fusion Proteins
Bloconlugate Chem., Vol. 4, No. 3, 1993 233
GMLT TNF ---IL-2
CSF
c - + - + - + - + M *9
w
34 39
13
1
V. I
0
10
20
30
40
50
60
Hr post-injection --I- ch14.18
14
+ 14.18-
14.18-112
Figure 5. Clearance of ch14.18-IL2 and chl4.18-GM-CSF following iv injectionof athymic (nu/nu) mice. Mice were injected in the tail vein (1mg/kg) with the indicated protein (three mice per group) and blood samples were taken a t 15 and 30 min and 1, 2,4,8,24, and 48 h and assayed by ELISA. Table I. Half-Lives of the a and Phases of the Free Antibody and the Cytokine Fusion Proteins
ch14.18 ch14.18-IL2 chl4.18-GM-CSF
!O
CY
B
7.5 0.3
59 30
1.4
28
into the extravascular space, may be responsible for the rapid clearance of the fusion proteins. As mentioned in Experimental Procedures, we found that these fusion proteins did not aggregate even after long-term storage. Thus we do not expect aggregation to be the cause of this rapid clearance. Protease Sensitivity of the Antibody-Cytokine Junctions. One possible explanation for the rapid clearance of the cytokine fusion proteins (with the exception of the Ig-GM-CSF construct) is that the added domains alter the overall structure of the antibody resultingin recognition by some type of clearancereceptor. The fact that fusion of the same cytokines (IL2, TNFP, and TNFa) led to both enhanced antigen binding and rapid clearance suggests that both altered properties are related to structural differences that are not present in the GM-CSF construct. We demonstrated earlier that the peptide bond between the carboxyl terminus of the ch14.18 antibody and IL-2 is susceptible to cleavage with plasmin and that this cleavage restores normal antigen binding (6). This result suggeststhat IL-2 actively changes the antibody conformation, resulting in the accessibility to proteolysis of what was originally the carboxyl terminal tail of the antibody. When we probed the plasminsensitivity of the GM-CSF fusion protein (Figure 6) we found that the junction was resistant to cleavage. All of the other fusion proteins were cleaved near the junction to generate a normal H chain. This finding suggests that the fusion of GM-CSF to the ch14.18 antibody has a different effect on the protein structure than the other cytokines. DISCUSSION
The recent widespread application of recombinant DNA technology to the study of protein structure and function has often involved the fusion of unrelated proteins or the
Figure 6. Protease sensitivity of Ig-cytokine fusion proteins. Samples (50 pg/mL, 0.5 mL) were incubated with or without plasmin, concentrated into gel sample buffer by adsorption to and elution from protein A Sepharose, and analyzed by SDSPAGE. Proteins were visualized by Coomassie Blue staining.
shufflingof functional domains. Antibody fusion proteins usually are designed for the purpose of combining the targeting ability of the antigen binding domains with functional domains of other proteins such as enzymes, toxins,growth factors, etc. (14-17).In another application, the binding ability of non-antibody-binding proteins are joined to the Fc portion of an antibody H chain, replacing the antigen binding function of Fab. Such proteins, or “immunoadhesins”rely on the antibody portion either for effector function (ADCC, CDC, placental transfer in the case of IgG) or for conferring a longer in vivo half-life on the ligand binding domain (18). Very few antibody fusion proteins have been examined in animal systems in terms of their pharmacokinetic properties. In the case of the CD4 immunoadhesins, the addition of antibody H chain domains results in an extended circulating half-life compared to CD4 alone but it is still shorter than that of a normal IgG (18). These proteins contain an intact H chain CH2 exon which has been shown to play a major role in the catabolism of antibodies (19).In the present study we have made fusion proteins in which the cytokine is peptide linked to the carboxyl terminus of an intact antibody. In this way we hoped to achieve complete assembly of the L and fusion H chains and to obtain a molecule with a long circulating half-life. Only the fusion protein between the ch14.18 antibody and GM-CSF had a long half-life, although most of the others had half-lives longer than the native cytokines. For example, chl4.18-GM-CSF had a @ half-life in nude mice of about 28 h (Figure 6) whereas that of native GM-CSF, following bolus injection in humans, has been reported to be from 50 to 85 min. (20,21). It is important to point out, however, that human GM-CSF has low affinity (and activity) for the mouse GM-CSF receptor (22). This lack of binding to mouse cells may explain the long circulating half-life of ch14.18-GM-CSF. The ch14.18-IL2 construct, on the other hand, had a very rapid initial clearance rate followed by a /3 half-life of about 30 h. Thus the initial clearance may be due, in part, to the binding of the fusion protein to IL2 receptor-bearing cells. These cells would be lost when the blood was clotted, prior to preparation of the serum samples, together with the immunoreactive material being measured by ELISA. However, experi-
234
Glllles et al.
Bloconlugate Chem., Vol. 4, No. 3, 1993
ments with iodinated ch14.18-IL2, in which total labeled fusion protein was measured directly, showed the same rapid clearance in nude mice (B. Mueller and R. Reisfeld, personal communication). This finding would argue against the idea that rapid clearance is due to receptor binding on circulating cells, especially since nude mice (which have greatly reduced numbers of mature T cells) were used. An alternative explanation for the rapid initial (a) clearance of many of the cytokine fusion proteins is that structural changes in the antibody itself lead to recognition by some clearance mechanism and removal from the circulation. We have shown that the fusion of TNFP, TNFa, and especially IL2 to the ch14.18 antibody significantly enhances its antigen binding activity (5,6) and that removal of IL2, for example, by proteolytic cleavage restores normal binding. These results strongly suggest that the fusion of the cytokines actively alters the structure of the antibody domains and results in changes in antigen/ antibody interactions. The same process that alters binding may also increase the rate of clearance. In this regard it is noteworthy that the GM-CSF fusion protein had the least effect on antigen binding and had the longest half-life. The common link between the two activities may be the CH2 domain. As discussed earlier (23))the CH2 domain interacts with Fab and may reduce antigen binding in some systems (e.g. the 14.18anti-GD2 system) by reducing the flexibility of the Fab arm. Removal of the CH2 domain in the ch14.18 antibody was found to greatly enhance GD2 binding, presumably by increasing Fab flexibility. How the fusion of cytokines to the carboxyl terminus of an intact Ig molecule can achieve the same effect is more speculative, but it is possible that cytokine/CH2 interactions are strong enough to disrupt CH2/Fab interactions and free the Fab arm for antigen binding. This same CH2/ cytokine interaction would then be responsible for the distortion of the Ig structure that results in rapid clearance, since this domain is critical in determining Ig half-life (19).
The interaction between the carboxyl-terminal cytokine and the CH2 domain would vary with differencesin surface charge interactions, but when such an interaction takes place, it would likely involve a looping back around the CH3 domain. Such a structure might be susceptible to proteolytic cleavage, whereas a fusion protein in which the cytokine does not interact with CH2 would be more resistant. This explanation would account for the differences seen between chl4.18-GM-CSF and the other fusion proteins (Figure 6). We have also demonstrated that it is possible to make biologically active antibodyJcytokine fusion proteins with little or no loss of activity of either function. All of the constructs had full biological activity except for ch14.18GM-CSF, which was approximately 20% active. It should be noted, however, that the recombinant GM-CSF standard that was used for comparison was produced in bacteria while the fusion protein was expressed in mammalian cells. Such preparations of recombinant GM-CSF, as well as those that are produced in mammalian cells and enzymatically treated to remove carbohydrate, have much higher (- 20-fold) specific activities than the fully glycosylated forms (24). Thus, the fusion protein may be as active as native glycosylated GM-CSF. The more important question is whether or not these fusion proteins are useful in targeting the various cytokines to tumors in vivo. One potential problem, in addition to rapid clearance, is that the fusion proteins will bind to the
effector cells prior to reaching the tumor. It is also hard to predict what effect cytokine bivalency (since each H chain has a cytokine attached) will have on receptor internalization and subsequent signal transduction. We have already shown that GD2-positive melanoma cells coated with ch14.18-IL2 are much more readily killed by their autologous CTLs than are uncoated cells in the presence of free IL2 (6). This suggests that the fusion protein may serve a bridging function between the effector and target as an additional adhesion molecule. Experiments to test the efficacy of these fusion proteins in tumor-bearing mice are currently underway and should determine whether of not their relatively short half-life is a serious problem. The construction of a ch14.18 fusion protein using mouse GM-CSF would also make the analysis in a syngeneic mouse tumor model possible and at the same time make it possible to test whether the long halflife of the 14.18-GM-CSF fusion protein is due to a lack of GM-CSF receptor binding. ACKNOWLEDGMENT
We thank Ed Reilly for in vivo studies and Sue Foley, Joyce Coll, and Jing Sun for expert technical assistance. LITERATURE CITED (1) Tepper, R. I., Pattengale, P. K., and Leder, P. (1989) Murine
interleukin-4 displays potent anti-tumor activity in uiuo. Cell 57, 503. (2) Fearon, E. R., Pardoll, D. M., Itaya, T., Golumbek, P., Levitsky, H. I., Simons, J. W., Karasuyama, H., Volgelstein, B., and Frost, P. (1990) Interleukin-2 production by tumor cells bypasses T helper function in the generation of an antitumor response. Cell 60, 397. (3) Colombo, M. P., Ferrari, G., Stoppacciaro, A., Parenza, M., Rodolfo, M., Mavillo,F., and Parmiani, G. (1991) Granulocyte colony-stimulating factor gene transfer suppresses tumorigenicity of a murine adenocarcinoma in uiuo. J . Exp. Med. 173, 889. (4) Asher, A. L., Mule, J. J., Kasid, A., Restifo, N. P., Salo, J. C., Reichert, C. M., Jaffe, G., Fendly, B., Kriegler, M., and Rosenberg, S. (1991) Murine tumor cells transduced with the gene for tumor necrosis factor-a: evidence for paracrine immune effects of tumor necrosis factor against tumors. J . Immunol. 146, 3227. (5) Gillies, S. D., Young, D., Lo, K.-M., Foley, S. F., and Reisfeld, R. (1991) Expression of genetically engineered immunoconjugates of lymphotoxin and a chimeric anti-ganglioside GD2 antibody. Hybridoma 10,347. (6) Gillies, S. D., Reilly, E. B., Lo, K.-M., and Reisfeld, R. (1992) Antibody-targeted interleukin 2 stimulates T-cell killing of autologoustumor cells. Proc. Natl. Acad. Sei. U.S.A.89,1428. (7) Mueller, B. M., Romerdahl, C. A., Gillies, S. D., and Reisfeld, R. (1990) Enhancement of antibody-dependent cytotoxicity with a chimeric anti-GD2 antibody. J. Immunol. 144,1382. ( 8 ) Mujoo, K., Cheresh, D. A., Yang, H. M., and Reisfeld, R. (1986) Disialoganglioside GD2 on neuroblastoma: Target antigen for monoclonal antibody-mediated cytolysis and suppression of tumor growth. Cancer Res. 47, 1098. (9) Gillies, S. D., Lo, K.-M., and Wesolowski, J. (1989) Highlevel expression of chimeric antibodies using adapted cDNA variable region cassettes. J . Immunol. Methods 125, 191. (10) Gillies, S. D., Dorai, H., Wesolowski, J., Majeau, G., Young, D., Boyd, J., Gardner, J., and James, K. (1989) Expression of anti-tetanus toxoid antibody in transfected murine myeloma cells. Biotechnology 7, 799. (11) Gillis, S., Ferm, M. M., Ou, W., and Smith, K. A. (1978) T-cell growth factor: Parameters of production and a quantitative microassay for activity. J . Immunol. 120,2027. (12) Goeddel, D. V., Aggarwal, B. B., Gray, P. W., Leung, D. W., Nedwin, G. E., Palladino, M. A., Patton, J. S., Pennica, D., Shepard, H. M., Sugarman, B. J., and Wong, G. H. (1986)
Ig-Cytoklne Fuslon Proteins
Tumor necrosis factors: gene structureand biological activities. Cold Spring Harbor Sym. Quant. Biol. 51, 597. (13) Barker, E., Mueller, B. M., Handgretinger, R., Herter, M., Yu, A. L., and Reisfeld, R. A. (1991) Effect of a chimeric antiganglioside GD2 antibody on cell-mediated lysis of human neuroblastoma cells. Cancer Res. 51, 144. (14) Neuberger, M. S., Williams, G. T., and Fox, R. 0. (1984) Recombinant antibodies possessing novel effector functions. Nature 312, 604. (15) Chaudhary, V. K., Queen, C., Junghans, R. P., Waldman, T. A., Fitzgerald, D. J., and Pastan, I. (1989) A recombinant immunotoxin consisting of two antibody variable domains fused to Pseudomonas exotoxin. Nature 339, 394. (16) Shin, S.-U., and Morrison, S. L. (1990) Expression and characterization of an antibody binding specificity joined to insulin-likegrowth factor 1: Potential applications for cellular targeting. R o c . Natl. Acad. Sci. U.S.A. 87, 5322. (17) Gillies, S. D., Wesolowski, J. S., and Lo, K.-M. (1990) Targeting human cytotoxicT lymphocytes to kill heterologous epidermal growth factor receptor-bearing tumor cells. J. Zmmunol. 146,1067. (18) Capon, D. J., Shaw, S. M., Mordenti, J., Marsters, S. A., Gregory,T., Mitsuya, H., Bym, R. A.,Lucas, C., Wurm, F. M., Groopman,J. E., Broder, S., and Smith, D. H. (1989)Designing CD4 immunoadhesins for AIDS therapy. Nature 337, 525. (19) Mueller, B. M., Reisfeld, R. A., and Gillies, S. D. (1990) Serum half-life and tumor localization of a chimeric antibody
Bioconlugete Chem., Vol. 4, No. 3, 1993 236
deleted of the CH2 domain and directed against the disialoganglioside GD2. R o c . Natl. Acad. Sci. U.S.A. 87, 5702. (20) Cebron,J.,Dempsey, P., Fox,R. R., Kanomakis,G.,Bonnem, E., Burgess, A. W., and Morstyn, G. (1988) Pharmacokinetics of human granulocyte-macrophage colony stimulating factor using a sensitive immunoassay. Blood 72, 1340. (21) Herrmann, F., Schulz, G., Lindemann, A., Meyenburg, W., Oster, W., Krumwieh, D., and Mertelsmann, R. (1989) Hematopoietic responses in patients with advanced malignancy treated with recombinant granulocyte-macrophage colonystimulating factor. J. Clin. Oncol. 7, 159. (22) Kaushansky, K., Shoemaker, S. G., Alfaro, S., and Brown, C. (1989) Hematopoietic activity of granulocyte/macrophage colony-stimulating factor is dependent upon two distinct regions of the molecule: Functional analysis based upon the activities of interspecies hybrid growth factors. Proc. Natl. Acad. Sci. U.S.A. 86, 1213. (23) Gillies, S. D., and Wesolowski, J. (1990) Antigen binding and biological activities of engineered mutant chimeric antibodies with human tumor specificities. Hum. Antibod. Hybridomas 1,47. (24) Moonen, P., Mermod, J.-J., Ernst, J. F., Herschi, M., and DeLamarter, J. F. (1987) Increased biological activity of deglycosylatedrecombinant human granulocyte/macrophage colony-stimulating factor produced by yeast or animal cells. Proc. Natl. Acad. Sci. U.S.A. 84,4428.
