Rational immunotherapy with ribonuclease chimeras

Members of the pancreatic ribonuclease (RNase) family have diverse activities toward RNA that could cause them to function during host defense and physiological cell death pathways. This activity could be harnessed by coupling RNases to cell binding ligands for the purpose of engineering them into cell-type specific cytotoxins. Therefore, the cytotoxic potential of RNase was explored by linking bovine pancreatic ribonuclease A via a disulfide bond to human transferrin or antibodies to the transferrin receptor. The RNase hybrid proteins were cytotoxic to K562 human erythroleukemia cells in vitro with an IC₅₀ around 10⁻⁷ M, whereas>10⁻⁴ M of native RNase was required to inhibit protein synthesis. Cytotoxicity required both components of the conjugate since excess transferrin or ribonuclease inhibitors added to the medium protected the cells from the transferrin-RNase toxicity. Importantly, the RNase conjugates were found to have potent antitumor effects in vivo. Chimeric RNase fusion proteins were also developed. F(ab′)₂-like antibody-enzyme fusions were prepared by linking the gene for human RNase to a chimeric antitransferrin receptor heavy chain gene. The antibody enzyme fusion gene was introduced into a transfectoma that secreted the chimeric light chain of the same antibody, and cell lines were cloned that synthesized and secreted the antibody-enzyme fusion protein of the expected size at a concentration of 1–5 ng/mL. Culture supernatants from clones secreting the fusion protein caused inhibition of growth and protein synthesis toward K562 cells that express the human transferrin receptor but not toward a nonhuman derived cell line. Since human ribonucleases coupled to antibodies also exhibited receptor mediated toxicities, a new approach to selective cell killing is provided. This may allow the development of new therapeutics for cancer treatment that exhibit less systemic toxicity and, importantly, less immunogenicity than the currently employed ligand-toxin conjugates.     

Cytotoxic potential of ribonuclease and ribonuclease hybrid proteins

Pancreatic RNase injected into Xenopus oocytes abolishes protein synthesis at concentrations comparable to the toxin ricin yet has no effect on oocyte protein synthesis when added to the extracellular medium. Therefore RNase behaves like a potent toxin when directed into a cell. To explore the cytotoxic potential of RNase toward mammalian cells, bovine pancreatic ribonuclease A was coupled via a disulfide bond to human transferrin or antibodies to the transferrin receptor. The RNase hybrid proteins were cytotoxic to K562 human erythroleukemia cells in vitro with an IC50 around 10(-7) M whereas greater than 10(-5) M native RNase was required to inhibit protein synthesis. Cytotoxicity requires both components of the conjugate since excess transferrin or ribonuclease inhibitors added to the medium protected the cells from the transferrin-RNase toxicity. Compounds that interfere with transferrin receptor cycling and compartmentalization such as ammonium chloride decreased the cytotoxicity of transferrin-RNase. After a dose-dependent lag period inactivation of protein synthesis by transferrin-RNase followed a first-order decay constant. In a clonogenic assay that measures the extent of cell death 1 x 10(-6) M transferrin-RNase killed at least 4 logs or 99.99% of the cells whereas 70 x 10(-6) M RNase was nontoxic. These results show that RNase coupled to a ligand can be cytotoxic. Human ribonucleases coupled to antibodies also may exhibit receptor-mediated toxicities providing a new approach to selective cell killing possibly with less systemic toxicity and importantly less immunogenicity than the currently employed ligand-toxin conjugates.     

Cytotoxic ribonuclease chimeras. Targeted tumoricidal activity in vitro and in vivo.

Monoclonal antibodies to the transferrin receptor or to the T cell antigen, CD5, were chemically linked to mammalian RNase A and found to specifically inhibit protein synthesis in antigen-positive cells. Antibody-mediated specificity of these cytotoxic ribonuclease chimeras (CRCs) was demonstrated in three ways. 1) Toxicity was due to the chemical linkage of RNase to antibody, as the individual components added separately or in combination did not inhibit protein synthesis; 2) the anti-transferrin receptor CRCs inhibited protein synthesis in those cells expressing the human transferrin receptor (K562, U251, Jurkat cells) but had no detectable toxicity to cells lacking the human transferrin receptor (Vero or NIH 3T3 cells); 3) free antibody to either the human transferrin receptor (454A12 or 5E-9) or to the T cell antigen, CD5 (T101), blocked the cytotoxicity of the respective CRC. Two CRC species, designated P1 and P2, that differed in size and stoichiometry of RNase A to antibody, were purified by size-exclusion high performance liquid chromatography. The higher molecular weight P1 conjugate had an IC50 of 20-30 nM, whereas the P2 conjugate had a higher IC50 of 300-500 nM. Bioactivity could be reversibly increased more than 10-fold by freezing. The cytotoxicity of the CRCs was examined in vivo in a solid tumor animal model. Intratumoral injections of an anti-transferrin receptor CRC into established U251 human glioblastoma tumors grown in the flanks of nude mice prevented tumor growth, whereas RNase A mixed with antibody was ineffective. CRCs, therefore, express cytotoxicity in vitro and in vivo. Mammalian nucleases coupled to antibodies may be utilized as cell type-selective cytotoxins and have potential as pharmacologic reagents. The systemic toxicity and immunogenicity observed with mammalian derived cytotoxins may be significantly less than that of the currently employed plant- and bacterial-derived immunotoxins.     

Ribonucleases expressed by human pancreatic adenocarcinoma cell lines.

Human ribonucleases have been considered as a possible tumor marker for pancreatic cancer, and elevated serum levels of ribonuclease activity in patients with pancreatic cancer have been reported by many authors. The reason for this elevation is unknown. In this study, we demonstrate that human pancreatic adenocarcinoma cell lines synthesize and secrete different ribonucleases. We isolated and characterized human pancreatic, or secretory, ribonuclease (RNase 1) from the conditioned media of the human pancreatic adenocarcinoma cell lines Capan-1, MDAPanc-3, IBF-CP3 and Panc-1, and the ampullary adenocarcinoma cell line MDAAmp-7, which represent a wide range of differentiation stages. Only one of these cell lines, Panc-1, produces significant amounts of nonsecretory ribonuclease. We then established a purification procedure for both secretory and nonsecretory ribonucleases, consisting of concentration of the supernatant by tangential filtration, anion-exchange and cation-exchange liquid chromatography and C4 RP-HPLC. Ribonuclease activity fractions were monitored using both the spectrophotometric and negative-staining zymogram techniques. The results of N-terminal sequence analysis, kinetic analysis and endoglycosidase digestion studies indicate that the main ribonuclease secreted by all the cell lines is the secretory-type ribonuclease and that it is composed of several differently N-glycosylated forms. Northern blot analyses confirm that some of the cell lines express secretory ribonuclease mRNA. The mRNA levels produced by Panc-1 and MDAPanc-28 are too low to be detected. Similar levels of expression of nonsecretory ribonuclease are found by Northern blot analysis in all the cell lines except Panc-1, which expresses higher levels. Here, we describe, for the first time, that several human pancreatic cancer cell lines with different degrees of differentiation express and secrete ribonucleases. This fact indicates that one origin of the elevated serum RNase levels in patients with pancreatic cancer are tumor cells. Analysis of the oligosaccharide moiety of the RNase 1 secreted by Capan-1 shows that it is highly glycosylated and its N-glycan chains are significantly different from that of the RNase 1 produced by normal pancreas. These results renew the possibility of using human serum RNase 1 determination as a tumor marker.     

Insertional-fusion of basic fibroblast growth factor endowed ribonuclease 1 with enhanced cytotoxicity by steric blockade of inhibitor interaction

Basic fibroblast growth factor (bFGF) was inserted in the middle of human ribonuclease 1 (RNase1) sequence at an RNase inhibitor (RI)-binding site (Gly89) by a new gene fusion technique, insertional-fusion. The resultant insertional-fusion protein (CL-RFN89) was active both as bFGF and as RNase. Furthermore, it acquired an additional ability of evading RI through steric blockade of RI-binding caused by fused bFGF domain. As a result, CL-RFN89 showed stronger growth inhibition on B16/BL6 melanoma cells than an RI-sensitive tandem fusion protein. Thus, the insertional-fusion technique increases accessible positions for gene fusion on RNase, resulting in construction of a potent cytotoxic RNase.     

Conformational stability is a determinant of ribonuclease A cytotoxicity

Onconasetrade mark, a homolog of bovine pancreatic ribonuclease A (RNase A) with high conformational stability, is cytotoxic and has efficacy as a cancer chemotherapeutic agent. Unlike wild-type RNase A, the G88R variant is toxic to cancer cells. Here, variants in which disulfide bonds were removed from or added to G88R RNase A were used to probe the relationship between conformational stability and cytotoxicity in a methodical manner. The conformational stability of the C40A/G88R/C95A and C65A/C72A/G88R variants is less than that of G88R RNase A. In contrast, a new disulfide bond that links the N and C termini (residues 4 and 118) increases the conformational stability of G88R RNase A and C65A/C72A/G88R RNase A. These changes have little effect on the ribonucleolytic activity of the enzyme or on its ability to evade the cytosolic ribonuclease inhibitor protein. The changes do, however, have a substantial effect on toxicity toward human erythroleukemia cells. Specifically, conformational stability correlates directly with cytotoxicity as well as with resistance to proteolysis. These data indicate that conformational stability is a key determinant of RNase A cytotoxicity and suggest that cytotoxicity relies on avoiding proteolysis. This finding suggests a means to produce new cancer chemotherapeutic agents based on mammalian ribonucleases.     

Cytotoxicity of bovine seminal ribonuclease: monomer versus dimer

Bovine seminal ribonuclease (BS-RNase) is a homologue of bovine pancreatic ribonuclease (RNase A). Unlike RNase A, BS-RNase has notable toxicity for human tumor cells. Wild-type BS-RNase is a homodimer linked by two intermolecular disulfide bonds. This quaternary structure endows BS-RNase with resistance to inhibition by the cytosolic ribonuclease inhibitor protein (RI), which binds tightly to RNase A and monomeric BS-RNase. Here, we report on the creation and analysis of monomeric variants of BS-RNase that evade RI but retain full enzymatic activity. The cytotoxic activity of these monomeric variants exceeds that of the wild-type dimer by up to 30-fold, indicating that the dimeric structure of BS-RNase is not required for cytotoxicity. Dimers of these monomeric variants are more cytotoxic than wild-type BS-RNase, suggesting that the cytotoxicity of the wild-type enzyme is limited by RI inhibition following dissociation of the dimer in the reducing environment of the cytosol. Finally, the cytotoxic activity of these dimers is less than that of the constituent monomers, indicating that their quaternary structure is a liability. These data provide new insight into structure-function relationships of BS-RNase. Moreover, BS-RNase monomers described herein are more toxic to human tumor cells than is any known variant or homologue of RNase A including Onconase, an amphibian homologue in phase III clinical trials for the treatment of unresectable malignant mesothelioma.     

A new RNase-based immunoconjugate selectively cytotoxic for ErbB2-overexpressing cells

We report a new tumor-directed immunoRNase, a chimeric protein made up of an antibody fragment (single-chain Fv fragment) directed to ErbB2, a cell surface receptor, and a non-toxic, human ribonuclease, which upon cell internalization becomes cytotoxic. The immunoRNase is active as a ribonuclease, specifically binds and selectively kills ErbB2-positive cells. ErbB2 is one of the most specific tumor-associated antigens identified so far, overexpressed on tumor cells of different origin. Its choice as target antigen and that of a non-toxic, human RNase as the killer moiety makes this immunoRNase a new, potentially attractive anticancer agent.     

Separation of Albumin–Ribonuclease Conjugates

A method for separation of albumin–ribonuclease (RNase) conjugates has been proposed based on the use of macroporous silicates. It was established that about 76% of ligand-free human serum albumin (LFHSA) formed complexes with enzymes. It was shown that most of the conjugates of albumin and pancreatic RNase contained up to 2 mol enzyme per 1 mol LFHSA. The conjugates of albumin and bacterial RNase isolated from cells of the strain Bacillus intermedius 7P displayed higher specific activities, containing, on average, 2.3 mol RNase per 1 mol LFHSA (for the conjugates with molecular weights below 92 kDa) or 3.3 mol RNase per 1 mol protein carrier (for the conjugates with higher molecular weight).     

Bovine Brain Ribonuclease Is the Functional Homolog of Human Ribonuclease 1

Mounting evidence suggests that human pancreatic ribonuclease (RNase 1) plays important roles in vivo, ranging from regulating blood clotting and inflammation to directly counteracting tumorigenic cells. Understanding these putative roles has been pursued with continual comparisons of human RNase 1 to bovine RNase A, an enzyme that appears to function primarily in the ruminant gut. Our results imply a different physiology for human RNase 1. We demonstrate distinct functional differences between human RNase 1 and bovine RNase A. Moreover, we characterize another RNase 1 homolog, bovine brain ribonuclease, and find pronounced similarities between that enzyme and human RNase 1. We report that human RNase 1 and bovine brain ribonuclease share high catalytic activity against double-stranded RNA substrates, a rare quality among ribonucleases. Both human RNase 1 and bovine brain RNase are readily endocytosed by mammalian cells, aided by tight interactions with cell surface glycans. Finally, we show that both human RNase 1 and bovine brain RNase are secreted from endothelial cells in a regulated manner, implying a potential role in vascular homeostasis. Our results suggest that brain ribonuclease, not RNase A, is the true bovine homolog of human RNase 1, and provide fundamental insight into the ancestral roles and functional adaptations of RNase 1 in mammals.     

Tandemization endows bovine pancreatic ribonuclease with cytotoxic activity

Due to their ability to degrade RNA, selected members of the bovine pancreatic ribonuclease A (RNase A) superfamily are potent cytotoxins. These cytotoxic ribonucleases enter the cytosol of target cells, where they degrade cellular RNA and cause cell death. The cytotoxic activity of most RNases, however, is abolished by the cytosolic ribonuclease inhibitor (RI). Consequently, the development of RNase derivatives with the ability to evade RI binding is a desirable goal. In this study, tandem enzymes consisting of two RNase A units that are bound covalently via a peptide linker were generated by gene duplication. As deduced from the crystal structure of the RNase A.RI complex, one RNase A unit of the tandem enzyme can still be bound by RI. The other unit, however, should remain unbound because of steric hindrance. This free RNase A unit is expected to maintain its activity and to act as a cytotoxic agent. The study of the influence of the linker sequence on the conformation and stability of these constructs revealed that tandemization has only minor effects on the activity and stability of the constructs in comparison to monomeric RNase A. Relative activity was decreased by 10-50% and the melting temperature was decreased by less than 2.5 K. Furthermore, the cytotoxic potency of the RNase A tandem enzymes was investigated. Despite an in vitro inhibition by RI, tandemization was found to endow RNase A with remarkable cytotoxic activity. While monomeric RNase A is not cytotoxic, IC(50) values of the RNase A tandem variants decreased to 70.3-12.9 microM. These findings might establish the development of a new class of chemotherapeutic agents based on pancreatic ribonucleases.     

Three-dimensional structure of a human pancreatic ribonuclease variant, a step forward in the design of cytotoxic ribonucleases

We have determined the crystal structure of a human pancreatic ribonuclease or RNase 1 variant at 1.65 A resolution. Five residues in the N-terminal region were substituted by the corresponding amino acids of the bovine seminal RNase. In addition, a Pro to Ser mutation was present at position 50. The substitution of part of the N terminus has been critical both in improving the expression of this enzyme as a recombinant protein and in achieving its crystallisation. The determination of the crystal structure revealed the characteristic RNase fold including a V-shaped beta-sheet and three alpha-helices. It differs from its bovine RNase orthologue mainly in the loop regions. The active-site cleft shows a similar architecture to that of its bovine counterpart, with the essential residues occupying equivalent positions. In the present structure, however, His119 is displaced as it is in the structure of RNase A at high pH. An interaction model of human ribonuclease with the ribonuclease inhibitor, together with inhibition assays, indicate that, in contrast to RNase A, the modification of the loop beta4beta5 is not enough to avoid inhibition. This study represents the first crystallographic approach to the human enzyme, and should constitute an invaluable tool for the design of ribonuclease variants with acquired cytotoxic properties.     

Endowing human pancreatic ribonuclease with toxicity for cancer cells

Onconase is an amphibian protein that is now in Phase III clinical trials as a cancer chemotherapeutic. Human pancreatic ribonuclease (RNase 1) is homologous to Onconase but is not cytotoxic. Here, ERDD RNase 1, which is the L86E/N88R/G89D/R91D variant of RNase 1, is shown to have conformational stability and ribonucleolytic activity similar to that of the wild-type enzyme but > 10(3)-fold less affinity for the endogenous cytosolic ribonuclease inhibitor protein. Most significantly, ERDD RNase 1 is toxic to human leukemia cells. The addition of a non-native disulfide bond to ERDD RNase 1 not only increases the conformational stability of the enzyme but also increases its cytotoxicity such that its IC(50) value is only 8-fold greater than that of Onconase. Thus, only a few amino acid substitutions are necessary to make a human protein toxic to human cancer cells. This finding has significant implications for human cancer chemotherapy.     

Ribonuclease-neutralized pancreatic microsomal membranes from livestock for in vitro co-translational protein translocation

Here, we demonstrate that pancreatic microsomal membranes from pigs, sheep, or cattle destined for human consumption can be used as a valuable and ethically correct alternative to dog microsomes for cell-free protein translocation. By adding adequate ribonuclease (RNase) inhibitors to the membrane fraction, successful in vitro co-translational translocation of wild-type and chimeric pre-prolactin into the lumen of rough microsomes was obtained. In addition, the human type I integral membrane proteins CD4 and VCAM-1 were efficiently glycosylated in RNase-treated microsomes. Thus, RNase-neutralized pancreatic membrane fractions from pig, cow, or sheep are a cheap, easily accessible, and fulfilling alternative to canine microsomes.     

Evaluation of human pancreatic RNase as effector molecule in a therapeutic antibody platform

Human antibody-ribonuclease (RNase) fusion proteins, referred to as immunoRNases, have been proposed as an alternative to heterologous immunotoxins, without their immunogenicity and unspecific toxicity issues. In this study, we investigated if human pancreatic RNase will be suitable as effector component in a therapeutic antibody development platform. We generated several fusion proteins consisting of tumor-specific human immunoglobulins (IgGs) and human pancreatic RNase. Transient mammalian cell production was efficient and IgG-RNases were purified to homogeneity. Antigen binding was comparable to the parental antibodies and RNase catalytic activity was retained even in the presence of 50-fold molar excess of human cytosolic RNase inhibitor (RI). Serum stability, cell binding and internalization of IgG-RNases were comparable to the parental IgGs. Despite these promising properties, none of the IgG-RNases revealed significant inhibition of tumor cell growth in vitro even when targeting different antigens putatively employing different endocytotic pathways. The introduction of different linkers containing endosomal protease cleavage sites into the IgG-RNase did not enhance cytotoxicity. Similarly, RI evasive human pancreatic RNase variants mediated only small inhibiting effects on tumor cell growth at high concentrations, potentially reflecting inefficient cytosolic translocation. Taken together, human pancreatic RNase and variants did not prove to be generally suitable as effector component for a therapeutic antibody drug development platform.     

Evaluation of human pancreatic RNase as effector molecule in a therapeutic antibody platform

Human antibody-ribonuclease (RNase) fusion proteins, referred to as immunoRNases, have been proposed as an alternative to heterologous immunotoxins, without their immunogenicity and unspecific toxicity issues. In this study, we investigated if human pancreatic RNase will be suitable as effector component in a therapeutic antibody development platform. We generated several fusion proteins consisting of tumor-specific human immunoglobulins (IgGs) and human pancreatic RNase. Transient mammalian cell production was efficient and IgG-RNases were purified to homogeneity. Antigen binding was comparable to the parental antibodies and RNase catalytic activity was retained even in the presence of 50-fold molar excess of human cytosolic RNase inhibitor (RI). Serum stability, cell binding and internalization of IgG-RNases were comparable to the parental IgGs. Despite these promising properties, none of the IgG-RNases revealed significant inhibition of tumor cell growth in vitro even when targeting different antigens putatively employing different endocytotic pathways. The introduction of different linkers containing endosomal protease cleavage sites into the IgG-RNase did not enhance cytotoxicity. Similarly, RI evasive human pancreatic RNase variants mediated only small inhibiting effects on tumor cell growth at high concentrations, potentially reflecting inefficient cytosolic translocation. Taken together, human pancreatic RNase and variants did not prove to be generally suitable as effector component for a therapeutic antibody drug development platform.     

Expression of bovine pancreatic ribonuclease A in Escherichia coli

A synthetic gene for bovine pancreatic ribonuclease A (RNase A) has been expressed in Escherichia coli as a fusion protein with beta-galactosidase linked by the tetrapeptide Ile-Glu-Gly-Arg. RNase A was cleaved from the fusion using factor Xa, and the resulting product purified and reconstituted. The isolated RNase A was chromatographically, catalytically, and immunologically identical with authentic RNase A. This work argues that the method suggested by Nagai and Thogersen [Nagai, K. & Thogersen, H. C. (1984) Nature (Lond.) 309, 810-812] for releasing fusion proteins is quite general, even when applied to particularly complicated expression problem. The procedure here makes RNase A available for the first time as a model for studying structure-function relationships in proteins using site-directed mutagenesis.     

Glycosylation of human pancreatic ribonuclease: differences between normal and tumor states

Characterization of the N-glycans from human pancreatic ribonuclease (RNase 1) isolated from healthy pancreas and from pancreatic adenocarcinoma tumor cells (Capan-1 and MDAPanc-3) revealed completely different glycosylation patterns. RNase 1 from healthy cells contained neutral complex biantennary structures, with smaller amounts of tri- and tetraantennary compounds, and glycans with poly-N-acetyllactosamine extensions, all extensively fucosylated. In contrast, RNase 1 glycans from tumor cells (Capan-1) were fucosylated hybrid and complex biantennary glycans with GalNAc-GlcNAc antennae. RNase 1 glycans from Capan-1 and MDAPanc-3 cells also contained sialylated structures completely absent in the healthy pancreas. Some of these features provide distinct epitopes that were clearly detected using monoclonal antibodies against carbohydrate antigens. Thus monoclonal antibodies to Lewis(y) reacted only with normal pancreatic RNase 1, whereas, in contrast, monoclonal antibodies to sialyl-Lewis(x) and sialyl-Lewis(a) reacted only with RNase 1 secreted from the tumor cells. These glycosylation changes in a tumor-secreted protein, which reflect fundamental changes in the enzymes involved in the glycosylation pathway, open up the possibility of using serum RNase 1 as a tumor marker of pancreatic adenocarcinoma.     

Human ribonucleases. Quantitation of pancreatic-like enzymes in serum, urine, and organ preparations

Antibodies against pure human pancreatic ribonuclease (RNase) were used to study ribonuclease levels in human tissues and body fluids. The antibodies completely inhibit the activity of purified RNase as well as ribonuclease activity in crude pancreatic extracts. RNase activity is inhibited by 70-80% in serum and urine, indicating that a significant proportion of the RNases in these preparations are structurally like the pancreatic enzyme. In contrast, inhibition of RNase activities from spleen (8%) and liver (30%) was inefficient suggesting that most of the RNases in these tissues are structurally unlike the pancreatic enzyme. A competitive binding radioimmunoassay (RIA), sensitive in the range of 1-100 ng of RNase, was developed to quantitate the pancreatic like enzymes. The RIA of crude tissue preparations and samples fractionated by gel filtration was compatible with inhibition results. Enzymes structurally like pancreatic RNase could be quantitated despite the presence of other RNase activities. Immunological quantitation of pancreatic like RNases was also found to be much more simple and precise than enzymatic assays comparing RNA and polycytidylate substrates. We suggest the immunological assays will be useful in the quantitation and definition of tissue of origin of RNases in serum of patients with pancreatic carcinoma.     

Targeting of tumor necrosis factor to tumor cells: secretion by myeloma cells of a genetically engineered antibody-tumor necrosis factor hybrid molecule

The construction, synthesis and secretion of a genetically engineered antibody-cytokine fusion molecule is described. To target tumor necrosis factor (TNF) to tumor cells, recombinant antibody techniques were used to produce a Fab-like antibody-TNF conjugate. At the gene level, the heavy chain gene of an antitransferrin receptor antibody was linked to a synthetic TNF gene encoding human TNF. Transfection of the heavy chain-TNF gene into a myeloma derived cell line which was producing the light chain of the same antibody, allowed the isolation of a cell line secreting a fusion protein of the expected molecular weight and composition. The culture supernatant of the cell line contained TNF cytotoxic activity towards murine L929 cells and human MCF-7 cells. Cytotoxicity towards the human cancer cells was inhibited by an excess of the original antitransferrin receptor antibody, indicating that the antibody-TNF molecules are targeted to the transferrin receptor rich tumor cells. Since the antibody genes used are chimeric (i.e. composed of mouse variable and human constant regions) and since DNA encoding human TNF was used, the hybrid protein is an example of a humanized immunotoxin-like molecule. These results illustrate the possibilities of antibody engineering technology to create and produce improved agents for cancer therapy. Furthermore, they demonstrate for the first time the ability of myeloma cells to secrete an antibody-cytokine chimeric molecule.