Activity of Recombinant Mitogillin and Mitogillin Immunoconjugates*

A synthetic gene for the Aspergillus protein toxin mitogillin has been synthesized and expressed in Escherichia coli. The recombinant mitogillin is a potent inhibitor of protein synthesis in vitro with an IC6,, of 9.7 PM. Immunoconjugates of recombinant mitogillin derivatized with S-acetylmercaptosuccinic anhydride and 5-methyl-2-iminothiolane modified H65 antibody kill T cell lines and peripheral blood mononuclear cells expressing the human CD5 surface antigen. Native mitogillin contains 4 cysteine residues which form two disulfide pairs (Fernandez-Luna, J. L., Lopez-Otin, C., Soriano, F., and Mendez, E. (1985) Biochemistry 24, 861-867). Three derivatives of mitogillin have been assembled which substitute alanine residues for cysteine residues 5, 147, or 5 and 147. Each of these molecules retains the ability to inhibit protein synthesis in vitro with at most a 2-fold reduction in activity. The derivative mitogillinc14,A can be conjugated to 5-methyl-2-iminothiolane-modified H65 antibody di- rectly without pretreatment with S-acetylmercapto-succinic anhydride, and the immunoconjugate is active against HSB2 cells. Genetic manipulation of toxin genes to expose an accessible cysteine residue into a recombinant product can thus be used to generate immunotoxins Whole Cell Cytotoxicity Assays-These assays measure the inhibi- tion of macromolecular synthesis in intact T cells by immunoconjugates. H65-RTA and immunoconjugate samples were diluted with RPMI without leucine at half-log concentrations ranging from 2,000 to 0.632 ng/ml. All dilutions were added in triplicate to microtiter plates containing 1 X lo5 HSB2 cells. These cells were incubated 20 h at 37 “C and then pulsed with [’Hlleucine for 4 h before harvesting. Samples were counted on the Inotech Trace 96. By comparison with an untreated sample, ICno can be calculated. The extent of kill was calculated by dividing the [‘HH]leucine uptake (cprn) at the highest concentration of immunoconjugate tested (usually 1 pg/ml) by that uptake in the absence of any added immunoconjugate, subtracting that result from 1, and multiplying by 100%. An identical assay with K562 cells which do not express the CD5 antigen was done as a negative control. The whole cell cytotoxicity assay with human PBMC was per- formed in a similar manner (16). PBMC (1 X lO’/well) were treated in triplicate with various concentrations of immunoconjugates in microtiter plates and stimulated with 3 pg/ml leukoagglutinin (PHA, Pharmacia LKB Biotechnology Inc.). Cultures were incubated for 90 h at 37 “C in a 10% CO:! incubator with humidified air. Tritiated thymidine ([‘HITdR, DuPont-New England Nuclear, 1 pCi/well) was added 16 h prior to harvest. Uptake of [‘HITdR was quantitated with the Inotech Trace 96 Detection System. The ICso was determined from the immunoconjugate concentration resulting in a 50% inhibition of the maximum (no immunoconjugate) [‘HITdR uptake

the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) M94249.
To whom correspondence should he addressed XOMA Corporation, 1545 17th St., Santa Monica, CA 90404. class of ribosome-inactivating proteins (10). Although initial interest in these Aspergillus proteins was related to their potential as chemotherapeutic agents, there has been recent interest in their use as immunoconjugates with monoclonal antibodies. Chemical coupling of restrictocin to antibodies which recognize cancer cell targets has been described, and specific cytotoxicity has been observed (11,12).
Mitogillin is an attractive candidate for linkage to monoclonal antibodies because of its potent inhibition of protein synthesis and relatively small size, with an M, of about 16,900.
The gene encoding mitogillin has not been cloned, and the DNA sequence of the native gene is therefore unknown. However, Fernandez-Luna et al. (8) have reported the complete amino acid sequence of the protein. With this information, we have synthesized a recombinant version of mitogillin and expressed this material in E. coli. Mitogillin immunoconjugates were formed with cross-linking agents which introduce reactive sulfhydryl moieties into both antibody and recombinant mitogillin as each lacks a free cysteine. Derivatives of mitogillin with unpaired cysteine residues were also made which facilitate conjugation to antibody molecules without prior nonspecific derivatization with cross-linking reagents. The properties of these recombinant mitogillins and immunoconjugates are described herein.

MATERIALS AND METHODS
Bacterial Strains and Plasmids-The E. coli host for production of mitogillin is a derivative of W3110, obtained from the E. coli Genetic Stock Center, Yale University, New Haven, CT. The plasmid used for expression of mitogillin is an inducible secretion vector in which gene expression is under the control of the Salmonella typhimurium araB promoter (13). Secretion of protein through the cytoplasmic membrane is conferred by the Erwinia carotouora pelB leader sequence (14). This expression system has been described previously for expression of antibody fragments in E. coli (15).
Mammalian Cells-The HSB2 human T cell line is available from the American Type Culture Collection (Rockville, MD). K562 is a multipotential hematopoetic malignant cell line, ATCC CCL243. Human peripheral blood mononuclear cells (PBMC)' obtained from healthy adults were isolated over Ficoll-Paque as described (16).

FIG. 1. Assembly and expression of recombinant mitogillin.
A, the mitogillin gene was assembled from six oligonucleotides and amplified by PCR with Taq polymerase. Oligonucleotides were mixed in pairs in 100-p1 reactions with 1 pg of each DNA, 2.5 units of Taq polymerase, 50 mM KCI, 10 mM Tris-C1, pH 8.3, and 1.5 mM MgC12. The tube was incubated in a Coy Tempcycler for 1 min at 94 "C, 2 min at 55 "C, and 10 min at 72 "C. A portion of each reaction product (40 pl) was mixed in pairs (Mito-1,2 + Mito-3,4; Mito-3,4 + Mito-5,6), 2.5 units of Taq were added, and the tubes were reincubated at 94 "C for 1 min, 55 "C for 2 min, and 72 "C for 10 min. The mitogillin gene was then assembled by mixing 10 pl of the Mito-1,2,3,4 reaction with 10 pl of the Mito-3,4,5,6 reaction and bringing the volume to 100 p1 of 2.5 units of Taq, 50 mM KC1, 10 mM Tris-C1, pH 8.3, 1.5 mM MgC12, a 200 p~ concentration of each dNTP, and 0.5 *g of each amplification primer Mito-7 and Mito-8. The reaction was overlayed with mineral oil, and the cycle profile used for amplification was denaturation at 94 "C for 1 min, annealing at 55 "C for 2 min, and primer extension at 72 "C for 3 min. Primer extension was carried out for 30 cycles. B, the amplified mitogillin DNA was cut with SphI and XhoI and cloned into an intermediate vector, generating pMITO19, in which its DNA sequence was verified. The relative locations of primers used for DNA sequence verification and sitedirected mutagenesis are shown. C, schematic representation of mitogillin expression vector pING3522. Shown are the positions of the tetracycline resistance gene (ret), the araC regulatory gene, the origin of replication (ori), and thepeZB leader. Also shown are the positions of primers used for PCR amplification of the mitogillin gene segments (HindIXI-2 hybridizes to a region near the tetracycline resistance gene promoter, and araB2 hybridizes to the araB promoter region) and restriction sites used for gene assembly: RI (EcoRI), FspI, SstI, PstI, and XhoI.
DNA sequence of oligonucleotides used to assemble the mitogillin gene is shown in Table I. Oligonucleotides greater than 50 hp in length were purified on a 15% polyacrylamide gel in the presence of 25% urea. DNA strand extension and DNA amplification were with Taq polymerase and the PCR GeneAmp Kit, Perkin-Elmer Cetus Instruments.
The synthetic mitogillin gene was cut with restriction enzymes at sites engineered into Mito-1 and Mito-6 (SphI and XhoI) and cloned into a vector with compatible restriction sites. Candidate genes were sequenced with Sequenase (U. S. Biochemical Corp.) to assure that no base changes had occurred during PCR amplification. The mito-gillin gene was excised as an FspI to XhoI fragment and positioned downstream of the peZB leader sequence in pING1500 (15), and clone pING3520 was identified. The peiB-mitogillin gene was then cloned downstream of the araB promoter into the final expression vector pING3522. Upon arabinose induction, this vector can direct the expression of m i~~l l i n protein which is secreted from E. coli.

Design and A s s e~~y of R e c o m b i~n~ M~t o g~~~~n
with Altered Cysteine R e s~~s -C y s t e i n e residues at positions 5 and/or 147 were changed to alanine by site-directed mutagenesis using PCR as described by Ho et al. (17). To change the Cys' to Ala, pING3522 DNA was amplified with oligonucieotide Mito-11 and araB2, and separately with Mito-10 and Mit3'-5'. Typical reaction conditions were 100 pmol of each primer, a 200 p~ concentration of each dNTP, 0.05 pg of template DNA, 2.5 units of Taq polymerase, 50 mM KCl, 10 mM Tris, pH 8.3, and 1.5 mM MgCl2. These reactions amplified 240-and 295-bp DNA fragments, respectively. The products of these reactions were mixed and amplified with the primers araB2 and Mit3'-5'. This product was cut with PstI and EcoRI, and the 410-bp piece was purified. This was ligated into the vector along with the 179-bp EcoRI to XhoI fragment, which contains the 3' end of mitogillin, to reassemble the altered mitogillin gene.
The Cys residue 147 was changed to alanine similarly by amplifying pING3522 with primers Mito-12 and HindIII-2, and Mito-13 and Mito-250. These reactions amplified 125-and 210-bp fragments, respectively. These fragments were mixed, amplified with Mito-250 and HindIII-2, cut with EcoRI and XhoI, and the 179-hp fragment was purified. This was cloned back into the original vector and verified by DNA sequence analysis.
A mitogillin variant with both Cys to Ala changes was assembled from the individually altered plasmids.
P~r i f~c u t~n of ~e c o~b i~n~ ~i t o g i l~i n -~a c h mitogillin plasmid was transformed into E. coli. Mito~llin was produced by induced cells grown either in a shake flask in tryptone broth or in a Chemap 14liter fermenter in 10 liters of minimal medium supplemented with 0.7% glycerol. For shake flask production, cells were grown to an Am of 0.4 and induced with 0.1% arabinose for 16 h. In the fermenter, the cells were induced by adding arabinose to 0.05% when an Am = 50 was attained, and the cells were incubated for approximately 16 h following induction. Mitogillin was purified from the culture supernatant of a 1-liter induced shake flask culture by Blue Toyopearl (TosoHaas) column chromatography. The pH of the culture supernatant was adjusted to 6.5 with 0.2 M sodium phosphate buffer and diluted with water to a conductivity of 1.0 millisiemans/cm. The sample was filtered through a 0.45-pm filter, and 800 ml was loaded onto a 2-ml Blue Toyopearl column (equilibrated with 10 mM sodium phosphate, pH 6.5) at 1 ml/ min. After sample loading, the column was washed with 30 ml of 10 mM sodium phosphate buffer, pH 6.5, and the bound mitogillin was eluted with 20 mI of 0.5 M NaCI, 10 mM sodium phosphate, pH 6.5. The sample was concentrated and analyzed by SDS-PAGE (see Fig.   3). The mitogiIIin consisted of 56% 36-kDa and 44% 19-kDa proteins under nonreducing gel conditions. Upon reduction, only a 19-kDa protein existed which is monomeric mitogillin. It is estimated that at least 1.0 pg/ml mitogillin was produced in the shake flask supernatant.
Each of the four forms of recombinant mitogillin was purified from the concentrated culture supernatant from a 10-1 bacterial fermentation broth. The concentrated fermen~tion broth was exchanged into 10 mM sodium phosphate, pH 7.5, using a DClO with SlOYlO cartridge (Amicon), and loaded onto a CM-52 hatm man) column preequil~hra~d with 10 mM sodium phosphate, pH 7.5. Protein was eluted from this column with a 0-300 mM NaCl linear gradient in 10 mM sodium phosphate buffer, pH 7.5; mitogillin eluted between 150 and 240 mM NaCl.
A sandwich enzyme-linked immunosorbent assay was used for detection of mitogillin in culture supernatants. Plates were coated with affinity-purified rabbit anti-mitogillin antibody overnight at 4 "C. Wells were then blocked with 1% BSA-PBS for 30 min at 37 'C, and sample dilutions were added. After three washes, biotinylated affinity-purified rabbit anti-mitogillin antibody was added for 1 h at   were used to assemble the mitogillin gene. Mito-7 and Mito-8 are PCR primers used to amplify the mitogillin gene DNA.  were used as site-directed mutagenesis primers for construction of mitogillin mutants. Mito-250 and Mit3'-5' were primers used for DNA sequence verification of the mitogillin sequence and were also used as PCR primers for site-directed mutagenesis. HindIII-2 binds near the tet promoter in the vector region of pING3522, and araB2 binds to the araB promoter region of pING3522.  Reticulocyte Lysate Assay-The reticulocyte lysate assay (RLA), which measures the inhibition of protein synthesis in a cell-free system using endogenous globin mRNA from a rabbit red blood cell lysate, was as described by Press et al. (18). Decreased incorporation of ['HH]leucine is measured as a function of toxin concentration. Serial log dilutions of standard RTABO, the 30-kDa form of ricin A chain (19), and sample toxin were tested over a range of 1 pg/ml to 1 pg/ ml. Triplicate samples were prepared on ice, incubated for 30 min at 37 "C, and then quantified by cascade ionization (Trace 96 Detection System INB-384; Inotech, Lansing, MI). By comparison with uninhibited samples, the concentration of toxin (PM) which inhibits protein synthesis by 50% (ICso) can be calculated.

Mito
H65 and cH65-H65 is a murine IgGl/K monoclonal antibody which recognizes the human CD5 antigen (20-22). A chimeric antibody (mouse variable regions, human constant regions) version of H65 antibody was constructed as described (23). Chimeric H65 antibody (cH65) was expressed in mammalian cells and exhibited binding characteristics identical to the murine H65 antibody?
Preparation of H65-RTA-H65-RTA is an immunoconjugate composed of the anti-human CD5 murine monoclonal antibody H65 linked through N-succinimidyl3-(2-pyridyldithio)propionate (Pierce) to RTA. The details of its preparation have been described (20). H65-RTA is cytotoxic to cells expressing the human CD5 antigen.
Preparation of Murine and Chimeric H65-(M2IT)(SAMSA)-mitogillin, and IND2-MS-mitogillin-To prepare this disulfide-linked immunoconjugate, both antibody and toxin were modified at lysine residues to introduce reactive sulfhydryl groups. H65 or cH65 anti-D. M. Fishwild, unpublished observations. body was modified with 5-methyl-2-iminothiolane (M2IT) in the presence of 5,5'-dithiobis(nitrobenzoic acid), DTNB, as described (24). Recombinant mitogillin was treated for 30 min at room temperature with equimolar amounts of S-acetylmercaptosuccinic anhydride (SAMSA, Sigma) (25), and then desalted on Trisacryl GF05 (IBF Biotechnics). For conjugation, a 5-fold molar excess of mitogillin-SAMSA was combined with derivatized H65 in the presence of hydroxylamine (50 mM, pH 7.5) and incubated at room temperature for 3 h (or 18 h a t 4 "C) prior to column purification. A mixture of immunoconjugate and free antibody was separated from unconjugated toxin on an Ultrogel AcA44 column (IBF Biotechnics) equilibrated with 10 mM Tris-C1,150 mM NaC1, pH 7. Free antibody was separated from pure immunoconjugate by loading onto a Blue Toyopearl column equilibrated with 10 mM Tris-C1, 20 mM NaCl, pH 7.5, and eluting with 10 mM Tris-C1, 0.25 M NaCl, pH 7.5. Samples of the final product were examined by 5% nonreducing SDS-PAGE, Coomassie stained, and scanned with a Shimadzu laser densitometer to quantitate the number of toxin molecules/antibody. IND2-MS-Mitogillin was prepared by the method described above. IND2 is an isotype-matched negative control antibody specific for a melanoma antigen and does not bind to human T cells.
Preparation of H65-(M2IT)-mitogillin~,~~~-H65 antibody was modified with a 12-fold molar excess of MZIT in the presence of 2 mM DTNB. The mitogillin~~47.~ was treated briefly with DTT to expose a reactive sulfhydryl group at position 5 and then desalted (the presence of a free sulfhydryl was verified by reaction with DTNB). H65 derivatized with M2IT was incubated with reduced then 16 h at 4 'C. The conjugate was purified on AcA44 followed by mitogillinc1471\ a t a molar ratio of 1 5 at room temperature for 3 h and Blue Toyopearl columns. Samples of the final product were run on 5% nonreduced SDS-PAGE, Coomassie stained, and scanned with a Shimadzu laser densitometer to quantitate the number of toxin molecules/antibody. Whole Cell Cytotoxicity Assays-These assays measure the inhibition of macromolecular synthesis in intact T cells by immunoconjugates. H65-RTA and immunoconjugate samples were diluted with RPMI without leucine a t half-log concentrations ranging from 2,000 to 0.632 ng/ml. All dilutions were added in triplicate to microtiter plates containing 1 X lo5 HSB2 cells. These cells were incubated 20 h at 37 "C and then pulsed with ['Hlleucine for 4 h before harvesting. Samples were counted on the Inotech Trace 96. By comparison with a n untreated sample, ICno can be calculated. The extent of kill was calculated by dividing the ['HH]leucine uptake (cprn) at the highest concentration of immunoconjugate tested (usually 1 pg/ml) by that uptake in the absence of any added immunoconjugate, subtracting that result from 1, and multiplying by 100%. An identical assay with K562 cells which do not express the CD5 antigen was done as a negative control.
The whole cell cytotoxicity assay with human PBMC was performed in a similar manner (16). PBMC (1 X lO'/well) were treated in triplicate with various concentrations of immunoconjugates in microtiter plates and stimulated with 3 pg/ml leukoagglutinin (PHA, Pharmacia LKB Biotechnology Inc.). Cultures were incubated for 90 h at 37 "C in a 10% CO:! incubator with humidified air. Tritiated thymidine (['HITdR, DuPont-New England Nuclear, 1 pCi/well) was added 16 h prior to harvest. Uptake of ['HITdR was quantitated with the Inotech Trace 96 Detection System. The ICso was determined from the immunoconjugate concentration resulting in a 50% inhibition of the maximum (no immunoconjugate) ['HITdR uptake as follows.

50% inhibition
The extent of kill was calculated as described above, substituting ['HI TdR uptake for ("H]leucine uptake.

RESULTS
Design and Assembly of Recombinant Mitogillin-Based on the reported amino acid sequence of mitogillin, oligonucleotides were synthesized and used to assemble a gene which encodes the mitogillin amino acid sequence. These oligonucleotides ranged in length from 92 to 98 bp and were assembled into a complete gene and cloned into an expression vector for regulated expression of recombinant product. The procedure used to assemble mitogillin is outlined in Fig. 1, and the complete DNA sequence of the synthetic mitogillin gene is shown in Fig. 2. Mitogillin expression is under the control of the araBAD promoter which is tightly regulated by the inducer arabinose. Mitogillin could be detected in the culture supernatant of induced E. coli cultures with specific antibodies either by Western analysis or enzyme-linked immunosorbent assay or by its ability to inhibit protein synthesis. Expression was efficient (at least 1 pg/ml mitogillin was produced in a 1liter shake flask), and this material was purified directly from culture supernatants.
Three additional mitogillin genes were constructed which contained alanine substitutions for cysteine residues at positions 5, 147, or 5 and 147. These residues form an intrachain disulfide bond, are near the N and C termini of the protein and may be accessible to the surface of the molecule. Each of these mitogillin proteins was produced in the same E. coli expression system and could be purified directly from induced culture supernatants.
Characterization of the Recombinant Mitogillin Proteins- Fig. 3, A and B, shows the typical SDS-PAGE band patterns seen for each of the recombinant mitogillin proteins. All four recombinant forms were efficiently produced in E. coli, but the yield of wild-type mitogillin is about four times that of the modified molecules (data not shown). Wild-type recombinant mitogillin can associate in E. coli supernatants to form an apparent dimer which can be reduced to monomer of M, 19,000, although the fraction of dimer found in the fermen-A. tation broth is variable (compare lunes 1 and 3, Fig. 3A). MitOgillincSA and mi tog ill in^,^^^ also form substantial amounts of reducible dimer (Fig. 3B, lanes 2 and 3 ) ) whereas mitogillin~~A~~~,A was greater than 99% monomer. Only monomeric mitogillin was used for preparation of immunoconjugates. Without reduction, both mit~gillinc,~~A and mitogillincsA have 30-35% available free SH as determined with DTNB. M i t~g i l l i n C~~C ,~~~ had only 7% free thiol. In Vitro Actiuity of Recombinant Mitogillin-Recombinant mitogillin and its analogs were assayed for inhibition of protein synthesis in a rabbit reticulocyte assay (RLA). As shown and Mitogillin Immunoconjugates in Table 11, the IC50 was 9.7 PM. This value for mitogillin is similar to that reported for restrictocin of 10 PM (12). The three derivatives of mitogillin in which 1 or 2 of the 4 cysteine residues were converted to alanines were also tested in this assay. Each of these altered forms of mitogillin retained the ability to inhibit protein synthesis in vitro. The IC50 for each of the single alanine derivatives was 10.2 pM (mitogill i n~~~~~) and 13.7 PM (mitogillincsA), whereas the IC50 for mi tog ill in^^^^^^^^ was 18.7 PM. The ICSo of RTA30 was determined for comparison, and the relative potency of the four mitogillin species to RTAso is shown in Table I1 as a ratio. Recombinant mitogillin and mi tog ill in^^^^^ are roughly 40% as active as RTAao, whereas the activity of mitogillinc5A is slightly lower. Mitogi1linc5Acl47A is approximately 20% as active as RTASO.

GCGATGGCG
Characterization of Recombinant Mitogillin Immunoconjugates-Recombinant mitogillin, mi tog ill in^^^, and mito-gillir~C~~~A were conjugated with H65, a murine antibody which recognizes the human CD5 antigen. Mitogillin was also conjugated to chimeric H65 antibody. Antibody lysine residues were derivatized to produce an activated thiol for conjugation, using M2IT in the presence of DTNB. Recombinant mitogillin was derivatized with SAMSA to provide a reactive thiol, whereas mitogillinc5A and mitogillinc147A were conjugated directly to the activated antibody through the unique unpaired cysteine a t C Y S '~~ or Cys', respectively. (The available cysteine at position 5 in rnit~gillinc~~~A was confirmed by protein microsequencing after in situ alkylation with 4-vinylpyridine (26).) Conjugation efficiency (shown in all b lanes in Fig. 4) was 60-70% for H65-MS-mitogillir1, cH65-MS-mitogillin and H65"-mitogillin~~~~~, but only 8% for H65-M-mitogillinc5~. The purity and average toxin to antibody ratio of each final The IC6o values reported are averages of a t least three assays. The for mitogillin reflects an average of five assays, and the value for RTAnn is an average of 20 assays. product (except H65"-mitogillinc5~) were determined by densitometry of Coomassie-stained SDS-PAGE. Because of poor conjugation efficiency, the mitogillinc5A conjugate was lost during purification. The remaining conjugates were all greater than 98% pure after column chromatography (shown in all c lanes in Fig. 4). The toxin to antibody ratios were 1.3 for H65-MS-mitogillin and cH65-MS-mitogillin, and 1.6 for H65-M-mitogillin~~~~~. Activity of Recombinant Mitogillin Immunoconjugates-The purified immunoconjugates were tested for specific cytotoxicity on the human T cell line HSBB and on PBMC. A typical PBMC assay with H65-MS-mitogillin, cH65-MS-mitogillin, IND2-MS-mitogillin, and H65-RTA is shown in Fig. 5. The figure illustrates the concentration-dependent cytotoxicity, the ICso values, and the extent of kill achieved with these conjugates. The negative control IND2-MS-mitogillin was nontoxic. The results of cytotoxicity assays for all anti-CD5 mitogillin conjugates tested are summarized in Table 111. On HSBB cells, the IC50 values of cH65-and H65-MS-mitogillin (-200 pM toxin) were 90-96% of that measured for the H65-RTA standard, which is a well characterized immunoconjugate (16,(20)(21)(22). The extent of kill for the two mitogillin conjugates was also comparable (70-75%) to that of H65-RTA (78%). The cytotoxicity of the H65-MS-mitogillinc147~ conjugate was %fold lower (640 PM, toxin) and the extent of kill was not as great (55%) on the T cell line. Against PBMC, H65-MS-mitogillin (IC50 = 97.8 pM toxin) was twice as cytotoxic as H65-RTA, whereas the chimeric H65-MS-mitogillin conjugate had an intermediate IC50 (158 PM toxin). The extent of kill for all immunoconjugates was similar (approximately 75%). H65 conjugates were not cytotoxic to the antigennegative cell line K562 (data not shown).

DISCUSSION
Cytotoxic proteins are produced by prokaryotes and eukaryotes, plants and animals, presumably conferring some selec-