Characterization of RFB4− Pseudomonas Exotoxin A Immunotoxins

Aug 15, 1996 - Characterization of RFB4rPseudomonas Exotoxin A Immunotoxins ... To develop an immunotoxin for the treatment of B-cell malignancies, we...
1 downloads 0 Views 274KB Size
Download PDF
Bioconjugate Chem. 1996, 7, 557−563

557

Characterization of RFB4-Pseudomonas Exotoxin A Immunotoxins Targeted to CD22 on B-Cell Malignancies Elizabeth Mansfield, Ira Pastan, and David J. FitzGerald* Laboratory of Molecular Biology, DCBDC, National Cancer Institute, National Institutes of Health, 37/4E16, 37 Convent Drive MSC 4255, Bethesda, Maryland 20892. Received December 15, 1995X

To develop an immunotoxin for the treatment of B-cell malignancies, we constructed several candidate conjugates with RFB4, a B-cell specific anti-CD22 IgG1, and truncated forms of Pseudomonas exotoxin (PE). The four versions of PE included PE35 and PE35KDEL, which were linked to RFB4 via a disulfide bond, and PE38 and PE38KDEL, which were linked via a thioether bond. The PE35 truncated forms, which are fully active in ADP ribosylation and lack receptor binding sequences, do not require intracellular proteolytic cleavage in order to be active. PE35KDEL has the consensus endoplasmic reticulum retention signal, KDEL, replacing the wild type PE C-terminal sequence, REDLK. The PE38 forms retain all of domain II and therefore require cleavage to be active within cells. Cytotoxicity experiments on CD22-positive cell lines revealed that the PE35 conjugates were more active than the PE38 versions and the presence of the KDEL sequence generally enhanced toxicity by 5-10-fold compared to that of REDLK. The RFB4-PE35KDEL immunotoxin was most active in cytotoxicity assays against Burkitt’s lymphoma cell lines such as Daudi and CA46 (IC50 ) 0.2 ng/mL) and displayed little cytotoxicity toward human vascular endothelial cells (IC50 > 20 µg/mL). Results of experiments conducted in nude mice showed that both RFB4-PE35KDEL and RFB4-PE35 could inhibit the development of subcutaneous CA46 tumors.

INTRODUCTION

The difficulty of gaining access to all cancer cells within a solid tumor mass may reduce the effectiveness of immunotoxins targeted to solid tumors. Because antigens and receptors found on the surface of B-cell neoplasms are frequently readily accessible from plasma, they make attractive targets for immunotoxin therapy. A number of different B-cell antigens have been successfully targeted both in vitro and in vivo, for example CD19, CD22, and CD40 (1-3). Immunotoxins directed at the B-cell specific marker CD22, in particular, have been made using several different plant toxins such as ricin A chain, saporin, gelonin, and pokeweed antiviral protein, e.g. (4-7), and the bacterial toxin Pseudomonas exotoxin (PE)1 (8). CD22 is a lineage-restricted antigen of the Ig superfamily that is expressed on the surface of normal mature B-lymphocytes and on many malignant B-cells, including chronic B-lymphocytic cells (B-CLL), B-lymphoma cells, * To whom all correspondence should be sent. Phone: (301) 496-4797. Fax: (301) 402-1344. X Abstract published in Advance ACS Abstracts, August 15, 1996. 1 Abbreviations: PE, Pseudomonas exotoxin A; IgG, immunoglobulin G; CLL, chronic lymphocytic leukemia; IC50, 50% inhibitory concentration; SCID, severe combined immunodeficient; sIg+, surface immunoglobulin; EBV, Epstein-Barr virus; dgA, deglycosylated ricin A chain; IPTG, isopropyl β-D-thiogalactopyranoside; SMCC, succinimidyl 4-(N-maleimidomethyl)cyclohexane-1-carboxylate; DTNB, 5,5′-dithiobis(2-nitrobenzoic acid); DTE, dithioerythritol; DMSO, dimethyl sufoxide; PBS, phosphate-buffered saline; TNB, thiobis(2-nitrobenzoic acid); TE, 10 mM Tris and 1 mM EDTA; HSA, human serum albumin; FBS, fetal bovine serum; HUVEC, human umbilical vein endothelial cells; BSA, bovine serum albumin; NAD, nicotinamide adenine dinucleotide; EF-2, elongation factor-2; EDTA, ethylenediaminetetraacetic acid; DTT, dithiothreitol; TCA, trichloroacetic acid; BES, N,N-bis(2-hydroxyethyl)-2-amino-ethanesulfonic acid; SDS, sodium dodecyl sulfate; Fv, antibody variable region; VLS, vascular leak syndrome; LD50, 50% lethal dose; iv, intravenous; TFR, transferrin receptor.

and hairy cell leukemias. Previously, immunotoxins using the anti-CD22 antibody RFB4 chemically conjugated to ricin A chain were shown to be highly effective in killing CD22-expressing cells in vitro and in preventing and curing experimental B-cell neoplasms induced in SCID and nude mice (2, 4). Pseudomonas exotoxin has been incorporated into immunotoxins targeting a variety of tumor types, and we anticipated it would be comparable to ricin A chain in effectiveness in an anti-CD22 conjugate using RFB4. Because several versions of truncated PE have been produced, we thought that a systematic analysis would be a useful way to optimize the construction of an RFB4 immunotoxin. PE is a three-domain toxin in which domain I at the N-terminus serves as the cell binding moiety, domain II in the middle of the protein allows translocation of PE across membranes, and domain III at the C-terminus harbors the ADP-ribosylating activity. Full length PE molecules require cleavage between amino acids 279 and 280 (9) and subsequent reduction of a disulfide bond in order to be active as toxins. PE is particularly attractive for making immunotoxin conjugates because it permits linkage to antibodies via disulfide bonds, thioether bonds, and peptide bonds. Truncated versions of PE have been used to construct immunotoxins of many different specificities. A truncated form of the toxin called PE35 has been constructed, beginning at amino acid residue 280 of full length PE, that contains the translocating and ADP-ribosylating domains and does not require proteolysis for activity (10). PE35 and PE35KDEL [where REDLK is replaced with the consensus ER retention signal KDEL (R. Kreitman, unpublished data)] have been conjugated through disulfide linkages to several IgGs and shown to have greater cytotoxic activity toward targeted cell lines (10; unpublished data) and CLL patient samples (11) than conjugates made with PE38, another truncated form of PE composed of all of domains II and III. PE38 differs from PE35 in that it retains the requirement for intracellular proteolysis.

S1043-1802(96)00043-2 Not subject to U.S. Copyright. Published 1996 by American Chemical Society

558 Bioconjugate Chem., Vol. 7, No. 5, 1996

RFB4 is a murine IgG1 anti-CD22 monoclonal antibody that reacts exclusively with B-cells in normal human tissues [92-96% of sIg+ cells (12)] and with pre-B, Burkitt’s, and plasmacytoid cell lines but not with prepre-B, EBV-transformed mature B, myeloma, or non-Bcell types in culture (13). Li et al. (13) found that no RFB4-reactive (i.e. CD22) molecules could be detected in normal human sera and sera from patients with CLL, suggesting that this antigen is not shed into the plasma. The immunotoxin RFB4-dgA, in which RFB4 IgG is disulfide-linked to deglycosylated ricin A chain, has been extensively tested in human clinical trials (14-16) and has demonstrated no toxicities except those related to ricin itself. For these reasons, the RFB4 antibody is an excellent candidate for conjugation to PE toxins for treatment of certain B-cell neoplasms. We have examined the efficacy of anti-CD22 immunotoxin conjugates constructed with four different truncated forms of PE. EXPERIMENTAL PROCEDURES

Antibodies. RFB4 IgG was provided by E. Vitetta (UTSMC, Dallas, TX) and was prepared as described (17). Toxins. PE35, PE35KDEL, N-LysPE38, and PE38KDEL were prepared from the periplasm of Escherichia coli as described for LysPE38KDEL (8). Expression of recombinant toxin molecules was induced by the addition of IPTG, which served to increase the T7-mediated transcription of the toxin gene. Cell Lines. HUT102 cells were a gift from T. Waldmann (NCI), and CA46, Raji, and JD38 lines were from I. McGrath (NIH). A431 cells were obtained from G. Todaro (Seattle). Daudi and HL-60 cells were purchased from ATCC (Rockville, MD), and HUVEC were purchased from Clonetics (San Diego, CA). Reagents. 2-Iminothiolane and SMCC were purchased from Pierce (Rockford, IL). PD-10 desalting columns and Mono-Q anion exchange columns were from Pharmacia Biotech (Piscataway, NJ). DTNB, DTE, and DMSO were purchased from Sigma (St. Louis, MO). Centricon concentrators were from Amicon (Bedford, MA), and TSK G3000SW columns were from Toso-Haas (Montgomeryville, PA). Conjugation. RFB4 IgG was modified with a 5-7fold molar excess of 2-iminothiolane in Dulbecco’s PBS at pH 7.4 for 1 h at 37 °C. The modified antibody was desalted over a PD-10 column equilibrated in PBS at pH 7.4. For modified antibody used for making disulfidelinked conjugates, the free sulfhydryls were blocked with DTNB added to a final concentration of 1 mM. Moles of free sulfhydryl per mole of protein were calculated as described (18). Excess DTNB and TNB were removed by desalting as above. PE35 and PE35KDEL were reduced with 100 mM DTE at room temperature and desalted over PD-10 columns as above. PE38 and PE38KDEL were modified in PBS at pH 7.4 with a 10fold molar excess of SMCC at room temperature for 1 h and then desalted over PD-10 columns as above. The antibody and PE were combined and concentrated to 0.5-1 mL in Centricon-30 concentrators (Amicon), incubated for several hours at room temperature, and then refrigerated overnight. Conjugated material was passed over Mono-Q anion exchange resin to separate PE and conjugates from free antibody. The column was eluted with a 0 to 500 mM linear gradient of NaCl in TE. Conjugate (1:1 Ab-PE) eluted at approximately 240 mM NaCl with some contamination from unreacted PE. Gel filtration over a 2.15 × 60 cm G3000SW column equilibrated in PBS at pH 7.4 separated IgG-PE conjugates from unconjugated PE. For the storage of some samples, human serum albumin (HSA) was added to 0.2%, al-

Mansfield et al.

though omission of HSA at immunotoxin concentrations at or above 200 µg/mL did not result in reduced cytotoxic activity. Samples were sterile filtered through MillexGV filters and frozen at -80 °C. Stability of Conjugates in Plasma. Plasma was pooled from three healthy donors who did not have neutralizing antibodies against PE. Immunotoxins were diluted 1:10 in plasma and in PBS/0.2% HSA and incubated for different times at 37 °C. Standard cytotoxicity assays were performed to determine the activity of treated immunotoxins. Cytotoxicity Assays. Cells were maintained in RPMI 1640 containing 10% fetal bovine serum (FBS), 50 u/mL penicillin, 50 µg/mL streptomycin, 1 mM sodium pyruvate, and an additional 2 mM L-glutamine. For cytotoxicity assays, 4 × 104 cells in 50 µL of culture medium were plated in 96-well plates. HUVEC were grown in endothelial cell growth medium (EGM) plus bovine brain extract, both purchased from Clonetics. Cells in the 3rd to 5th passage were seeded in 96-well plates at 4000 cells/ well in 200 µL for 3 days at 37 °C. For Daudi, CA46, Raji, JD38, HL60, and HUT 102 cytotoxicity experiments, immunotoxins were serially diluted in RPMI + 10% FBS and 50 µL was added to cells. For HUVEC experiments, immunotoxins were diluted in PBS/0.2% bovine serum albumin (BSA) and 10 µL was added to cells. Plates were incubated for 20-24 h at 37 °C and then pulsed with 1 µCi/well [3H]leucine in 50 µL of normal or leucine-free RPMI for 4-5 h at 37 °C (Daudi, CA46, HL60, and HUT 102) or 1 µCi/well in 10 µL of PBS for 7 h (HUVEC). Radiolabeled material was captured on filter mats and counted in a Betaplate scintillation counter (Pharmacia, Gaithersburg, MD). Quadruplicate sample values were averaged, and inhibition of protein synthesis was determined by calculating percent incorporation compared to control wells without added toxin. In Vivo Toxicity. Female BALB/c 6-8-week-old mice were obtained from the National Cancer Institute (Frederick, MD). Various amounts of immunotoxins diluted in 200 µL of PBS/0.2% HSA were injected into the tail vein. Two or more mice were injected for each point and were observed over a period of 7 days for deaths or severe toxicity. Inhibition of Tumor Development. Female 6-8week-old athymic nude mice were obtained from the National Cancer Institute. Mice were irradiated with 300 rad of total body irradiation 3 days prior to injection of tumor cells. CA46 human Burkitt’s lymphoma cells were seeded at 1.8 × 105/mL 2 days prior to injection in mice. Cells were washed and suspended in serum-free RPMI, and 8 × 106 to 1 × 107 cells in 200 µL were injected subcutaneously on the dorsal flank on day 0. Mice were injected with immunotoxin or PBS/0.2% HSA diluent in the tail vein every day for four doses starting on day 1 (24 h after injection of tumor cells). Tumor size was measured at 1-3 day intervals with precision calipers, and the volumes in mm3 were calculated by the equation v ) 0.4w2l, where w is width and l is length. ADP-Ribosylating Activity. ADP ribosylation activity of RFB4-PE35, RFB4-PE35KDEL, and RFB4-PE38 was determined by measuring transfer of ADP-ribose from [14C]NAD to EF-2 by the method of Collier and Kandel. Immunotoxins were diluted to 205 µL in 50 mM Tris (pH 8), 1 mM EDTA, 0.1% BSA, and 40 mM DTT and incubated with wheat germ extract in the presence of 2.4 µM [14C]NAD (6 × 105 cpm) for 15 min at 22 °C. Protein was precipitated with TCA and pelleted. Pellets were washed once in 6% TCA, solubilized, and counted in a scintillation counter. Binding Studies. Daudi cells grown to >106/mL were

Characterization of RFB4−PE Immunotoxins

Bioconjugate Chem., Vol. 7, No. 5, 1996 559

Figure 1. Schematic model of forms of PE used to make RFB4 conjugates. Full length PE comprises domains I, II, Ib, and III. The furin cleavage site is within the domain II disulfide loop. N-LysPE38 is deleted for domain I. An 11-amino acid sequence has been added to provide a lysine residue for modification with SMCC. PE35 has a deletion of domain I and a part of domain II up to amino acid 280. No disulfide loops remain. Ser287 is replaced with Cys to provide a free sulfhydryl for conjugation.

washed twice in ice cold binding buffer [RPMI, 50 mM BES (pH 6.8), and 1% BSA] and plated at 106 cells per 150 µL of binding buffer per well on 96-well plates on ice. To the cells were added 0.672 ng of [125I]RFB4 (7.3 × 108 cpm/nmol) in binding buffer and varying concentrations of RFB4-PE35, RFB4-PE35KDEL, and RFB4PE38. Cells were incubated for 3 h on ice, washed twice in cold binding buffer, and solubilized in 200 µL of 0.5% SDS/TE. Bound [125I]RFB4 was quantitated on a Wallac 1470 Wizard γ counter. The means of duplicate samples were used for calculations. RESULTS

Conjugation of RFB4 with Truncated Forms of PE. To evaluate the utility of PE for treating B-cell tumors, several different immunotoxins were made with the anti-CD22 antibody RFB4. RFB4 IgG was chemically conjugated to two truncated forms of PE called PE38 and PE35 (Figure 1). The KDEL C-terminal variants PE35KDEL and PE38KDEL were also tested. Because PE38 and PE38KDEL do not have free sulfhydryls, a reactive maleimide was introduced using the heterobifunctional cross-linking agent SMCC. Reduced PE35 and PE35KDEL each contain a single free sulfhydryl which can be used for conjugation by formation of a stable disulfide bond. This approach causes no loss of toxin activity. Prior to conjugation, RFB4 IgG was reacted with 2-iminothiolane to introduce an average of one free sulfhydryl per IgG and then mixed with either SMCC-modified PE38 or reduced PE35. The resulting conjugates were purified free of unreacted antibody and PE by sequential Mono-Q anion exchange and TSK gel filtration chromatography. RFB4-PE38, RFB4-PE38KDEL, RFB4-PE35, and RFB4-PE35KDEL were indistinguishable in their Mono-Q and TSK elution profiles. The 1:1 and 1:2 RFB4-PE conjugates eluted sequentially from Mono-Q in separate peaks, and the 1:2 conjugates were discarded. RFB4-PE38 and RFB4-PE38KDEL are thioetherlinked conjugates in which the active PE moiety cannot be released without proteolysis. In contrast, RFB4PE35 and RFB4-PE35KDEL and disulfide-linked conjugates, and the active toxin component can be released by reduction without proteolysis. Cytotoxicity. RFB4 conjugates with PE38, PE38KDEL, PE35, and PE35KDEL were tested for their ability to inhibit protein synthesis against the CD22positive Burkitt’s lymphoma cell lines Daudi, CA46, JD38, and Raji and the non-B, CD22-negative lines HL-60

Figure 2. Cytotoxicity of RFB4-PE conjugates on Daudi cells. Cytotoxicity of RFB4 monoclonal antibody conjugates containing PE35, PE35KDEL, PE38, and PE38KDEL after 24 h of incubation. Values are representative of at least three experiments for each conjugate.

(promyelocytic leukemia) and HUT 102 (T-cell lymphoma). RFB4-PE35 and RFB4-PE35KDEL were very active on CD22-positive lines [e.g. IC50 values of 1 ng/ mL for RFB4-PE35 and 0.2 ng/mL for RFB4-PE35KDEL on Daudi (Figure 2) and CA46 cells), while RFB4PE38KDEL and RFB4-PE38 had only moderate cytotoxic activity toward the same lines [IC50 values of 10 and 15 ng/mL on Daudi (Figure 2) and CA46 cells]. The CD22-negative lines were insensitive to all RFB4 immunotoxins [IC50 values of >1000 ng/mL for all RFB4 conjugates on HL-60 and HUT 102 (not shown)]. Unconjugated RFB4 IgG, PE35, PE35KDEL, and PE38 were not cytotoxic toward any of the cell lines at concentrations of up to 1000 ng/mL (not shown). PE38KDEL was not tested on these cell lines. Specificity. To confirm that the RFB4 conjugates bound specifically to CD22, competition assays were performed in which excess unconjugated RFB4 IgG was mixed with RFB4-PE35 and RFB4-PE35KDEL and was added to CA46 and Daudi cells. Cytotoxicity caused by either of the conjugates was reduced more than 100fold by addition of 10-30 µg/mL unconjugated RFB4 IgG (values for Daudi cells, which are similar to those for CA46, are shown in Figure 3). No competition occurred when a 10-30 µg/mL mAb directed toward an unrelated antigen was added. The possibility that toxicity could be caused by nonspecific internalization of immunotoxin was excluded by assays showing that a nonrelevant immunotoxin conjugate analogous to RFB4-PE35KDEL did not cause inhibition of protein synthesis in Daudi (Figure 3) and CA46 cells, which do not display the targeted antigen. Cytotoxicity against Endothelial Cells. Cytotoxicity against human vascular endothelial cells (HUVEC) has been suggested as a determinant of the propensity of immunotoxins to cause dose-limiting vascular leak syndrome when administered to human patients (19). Several ricin-containing immunotoxins, in particular RFB4-dgA, have been observed to cause HUVEC toxicity (19) and vascular leak syndrome (VLS) (14-16). We therefore tested RFB4-PE35KDEL for its cytotoxic activity against primary cultures of HUVEC, which arise

560 Bioconjugate Chem., Vol. 7, No. 5, 1996

Figure 3. Specific cytotoxicity of the RFB4-PE35KDEL conjugate toward Daudi cells. RFB4 antibody (30 µg) competes >100-fold of RFB4-PE35KDEL cytotoxicity. Nonrelevant PE35KDEL conjugate is not cytotoxic on Daudi cells at concentrations which are toxic for RFB4-PE35KDEL. Values represent two experiments which gave identical results.

Mansfield et al.

Figure 5. Binding of RFB4-PE conjugates to Daudi cells. [125I]RFB4 antibody binding is competed 50% by 12 nM RFB4 IgG and RFB4-PE35. RFB4-PE38 competes 50% of counts at 40 nM. Competition was performed under equilibrium binding conditions at 4 °C. Binding values were determined in two experiments with nearly identical results. Table 1. Multiple-Dose LD50 Values immunotoxina RFB4-PE35 RFB4-PE35KDEL

dose (µg × 4)b

survivors/total

20 30 40 2 4 5 7.5

2/2 0/2 0/2 2/2 2/2 1/2 0/2

a Immunotoxins were diluted in PBS/0.2% HSA and administered iv in a volume of 200 µL. b Doses shown were given at 24 h intervals for 4 consecutive days. Mice were observed for toxicity for 7 days after last administration in immunotoxin.

Figure 4. Cytotoxicity of RFB4-PE35KDEL on HUVECs. RFB4-PE35KDEL has an IC50 of >20 000 ng/mL on HUVECs after 24 h of exposure. LMB-7, another PE-containing immunotoxin, is more than 10 times more cytotoxic than RFB4PE35KDEL.

outside of the hematopoietic lineage and thus are not expected to express CD22. The IC50 of RFB4-PE35KDEL was greater than 20 000 ng/mL (Figure 4), suggesting that, if HUVEC cytotoxicity can be correlated with VLS, this immunotoxin would not likely cause VLS in humans by killing endothelial cells. ADP-Ribosylating Activity of Conjugates. The relative ADP-ribosylating activity of each conjugate was measured by in vitro assay, and the activities of different RFB4 immunotoxins were compared. RFB4-PE35, RFB4-PE35KDEL, RFB4-PE38, and RFB4-PE38KDEL differed from one another in ADP-ribosylating activity by not more than 2-fold (data not shown), demonstrating that the difference in cytotoxic activities of these conjugates was not due to a reduction in the enzymatic capacity of the PE portion of the molecules.

Binding of RFB4 Immunotoxins to Daudi Cells. To ensure that conjugation of RFB4 to PE35 and PE38 did not cause a reduction in binding to the CD22 antigen, the capacity of the immunotoxins to displace [125I]RFB4 IgG on Daudi cells was examined. Daudi cells grown to >106 cells/mL (106 cells/well, 0.2 mL) were incubated with 0.0225 nM [125I]RFB4 alone and 0.0225 nM [125I]RFB4 plus 0.036-22.5 nM RFB4-PE35, RFB4-PE38, and unlabeled RFB4 IgG at 4 °C. Both RFB4-PE35 and RFB4 IgG displaced 50% of [125I]RFB4 at 12 nM, demonstrating that conjugation of PE35 to RFB4 through a disulfide linkage resulted in no loss of binding activity (Figure 5). RFB4-PE38 competed 50% of bound [125I]RFB4 at 40 nM. Toxicity of RFB4-PE35KDEL and RFB4-PE35 in Mice. Because RFB4-PE35 and RFB4-PE35KDEL were the most active of the conjugates in tissue culture cytotoxicity experiments, we evaluated their usefulness as in vivo reagents. To select the maximum tolerated doses of RFB4-PE35KDEL and RFB4-PE35 for antitumor experiments, LD50 values were determined by multiple-dose iv injection of varying amounts of immunotoxin into the tail veins of BALB/c mice (Table 1). Two mice were tested for each concentration of immunotoxin, and at each dose level, immunotoxin was administered daily for 4 consecutive days. At doses of 4 × 4 µg of RFB4-PE35KDEL, all mice survived without obvious toxicity. At 4 × 5 µg, one mouse died after the fourth dose. Doses greater than 4 × 5 µg of RFB4-PE35KDEL

Bioconjugate Chem., Vol. 7, No. 5, 1996 561

Characterization of RFB4−PE Immunotoxins

were rapidly lethal (death occurred 2 days after first administration of immunotoxin) to all of the mice. The multiple-dose LD50 of RFB4-PE35 was between 4 × 20 and 4 × 30 µg, because no mice in the group receiving the 4 × 20 µg dose but both mice in the 4 × 30 µg dose group died. Stability of RFB4-PE35 and RFB4-PE35KDEL in Plasma. In order for a disulfide-linked immunotoxin to be effective as an antitumor agent in vivo, the disulfide bond must be stable to reduction in the bloodstream. The stability of the disulfide-linked immunotoxins RFB4PE35 and RFB4-PE35KDEL was examined by incubating these immunotoxins in human plasma at 37 °C for times ranging from 30 min to 24 h. Cytotoxic activity of treated samples on CA46 cells was compared to that of samples that were diluted in plasma and added to cells immediately without incubation at 37 °C. No change in activity compared to unincubated controls was seen in samples incubated for up to 24 h (data not shown), indicating that inactivation of RFB4 disulfide-linked immunotoxins in blood is not rapid. Inhibition of CA46 Tumor Development with RFB4-PE35KDEL. The excellent cytotoxic activities of RFB4-PE35KDEL and RFB4-PE35 in vitro indicated that they should be evaluated for in vivo activity. Therefore, the ability of RFB4-PE35 immunotoxins to prevent or delay formation of CA46 tumors in irradiated nude mice was assessed (Figure 6). Mice that had been inoculated with CA46 cells were randomly assigned to treatment groups, with four or five mice in each group. Mice were administered four iv doses each of RFB4PE35KDEL (1-4 µg) or RFB4-PE35 (1-10 µg), RFB4 IgG (1-5 µg), IgG-PE35KDEL (nonrelevant immunotoxin control), or PBS/0.2% HSA diluent control. Treatment was begun 24 h after injection of cells and administered at 24 h intervals for 4 days. By day 21, tumors in mice receiving diluent, 1 µg of RFB4 IgG or IgGPE35K, grew to a size averaging 1700 mm3. Treatment with 5 µg of RFB4 IgG resulted in tumors averaging 1560 mm3 on day 21. Ninety-one percent of mice that had received 4 × 1-4 µg of RFB4-PE35KDEL had no detectable tumors and did not develop tumors in the subsequent 90 days in which they were observed. One mouse out of 10 treated with 4 × 2 µg RFB4-PE35KDEL developed a slow-growing tumor that was 12% (205 mm3) of the average size of tumors in the control group on day 21. Additionally, two of 14 mice treated with 4 × 1 µg of RFB4-PE35KDEL developed tumors (44 and 8% of average control tumor size), while cage mates remained tumor-free. Two of eight surviving mice treated with 4 × 1 µg of RFB4-PE35 did not develop tumors; four of five mice treated with 2 µg of RFB4-PE35 grew tumors that were on average 38% of the size of those of controltreated mice on day 21. Treatment with 5 µg of RFB4PE35 prevented tumor growth in one of five mice, and the average size of tumors on day 21 was 12% of control. Two of five mice receiving 10 µg of RFB4-PE35 developed tumors that were detectable on day 18 and by day 21 were 2% the size of control tumors. Table 2 lists antitumor and toxicity data for all experiments. We conclude that RFB4-PE35KDEL and RFB4-PE35 were very effective in preventing or slowing the growth of CD22-positive tumors at nontoxic doses. DISCUSSION

Up to 60% of B-cell malignancies are potentially treatable with an immunotoxin targeted to CD22 (20). The murine anti-CD22 monoclonal antibody RFB4 has been extensively tested to confirm its B-cell specificity (13) and its ability to localize at CD22-positive tumor

Figure 6. Antitumor activity of RFB4-PE35 and RFB4PE35KDEL on CA46 subcutaneous tumors in nude mice. (A) Formation of tumors in mice inoculated with CA46 cells was inhibited by four doses of 10, 5, and 2 µg of RFB4-PE35. (B) Formation of CA46 tumors was prevented by four doses of 4, 2, or 1 µg of RFB4-PE35KDEL. See Table 2 for details on numbers of mice used. Table 2. Prevention of CA46 Tumors in Nude Mice immunotoxina RFB4-PE35

RFB4-PE35KDEL

dose (µg × 4)b

survivors

tumor-free day 21

10 5 2 1 4 2 1

5/5 5/5 5/5 8/9 9/10 10/10 14/14

3/5 1/5 1/5 2/8 9/9 9/10 12/14

a Immunotoxins were diluted in PBS/0.2% HSA and administered iv in a volume of 200 µL. b Doses shown were given at 24 h intervals for 4 consecutive days.

sites (21). Furthermore, RFB4-ricin A chain immunotoxins have shown promise in vitro, in murine tumor models, and, to some extent, in the human clinical setting (14-16). However, the dose-limiting toxicity in humans of ricin A chain-based immunotoxins, including RFB4dgA, is a vascular leak syndrome. Trials using RFB4-

562 Bioconjugate Chem., Vol. 7, No. 5, 1996

dgA have shown maximum tolerated dose levels of 28.8 mg/m2 per 192 h when continuously infused (16) and 75 mg/m2 when administered in two to six doses at 48 h intervals (14). Signs of vascular leak were apparent at doses as low as 9.6 mg/m2 per 192 h (continuous infusion) and 12.5 mg/m2 (when doses were given at 48 h intervals) (14, 16). To achieve antitumor effects consistently, higher dose levels are probably needed. Pseudomonas exotoxin itself is not believed to cause VLS. This toxicity has been apparent and dose-limiting with only one PE-based immunotoxin (LMB-1, directed at the Lewisy antigen) (22) but not others (23-25). In in vitro studies (26), LMB-1 and LMB-7 (a recombinant Fv version of LMB-1) were directly toxic to HUVEC in 20 h assays and also caused visible rounding of the cells, while PE38 (a recombinant form of PE that has no cellbinding domains) neither was cytotoxic nor caused cell rounding. Two other PE-based recombinant immunotoxins, ATac(Fv)PE38 and e23(dsFv)PE38 directed at the p55 subunit of IL-2R and the erbB-2 receptor, respectively, were also not cytotoxic to HUVEC. Inhibition of HUVEC protein synthesis by LMB-1 and LMB-7 preceded cell rounding. In the same study, ricin A chain (RTA) caused rapid cell rounding which, in contrast to LMB-1 and LMB-7, apparently precedes inhibition of protein synthesis (19). LMB-1 and LMB-7 cytotoxicity on HUVEC could be prevented by competition with excess B3 antibody, the parent antibody used to make LMB-1 conjugate, demonstrating that binding of these immunotoxins is specific, whereas RTA appears to bind HUVEC in a nonsaturable manner. Finally, dgRTA (also called dgA) inhibited protein synthesis to a similar degree as non-HUVEC-directed immunotoxin ITdgRTA, indicating that toxicity is due to ricin binding and not to binding mediated by the antibody portion of the immunotoxin. Because RFB4-dgRTA immunotoxins have caused VLS in human patients, we examined the ability of RFB4-PE immunotoxins to exert cytotoxic effects on HUVEC by measuring inhibition of protein synthesis. RFB4-PE immunotoxins are at least 20-fold less toxic to HUVEC than LMB-1 and LMB-7. Because we believe that the only PE-based immunotoxin to cause VLS (LMB1) exerts this toxicity through the antibody moiety of the molecule and not the PE itself, we predict that VLS will not occur in patients treated with RFB4-PE immunotoxins since CD22 is not present on endothelial cells. We have shown that, by exchanging Pseudomonas toxin for ricin in an RFB4 immunotoxin, we can obtain specific in vitro toxicities comparable to those reported for the RFB4-dgA immunotoxin. Also, RFB4-PE35 and RFB4-PE35KDEL immunotoxin conjugates are remarkably effective in their ability to inhibit the establishment of lymphoma tumors in nude mice inoculated with CA46 cells. The CA46 tumors that form in untreated inoculated mice are not grossly vascularized and appear macroscopically and microscopically similar to human lymphoma masses (A. Wellmann, unpublished observations). RFB4-PE immunotoxins, therefore, seem promising as therapeutic agents in circumstances of minimal residual disease. Cytotoxicity of RFB4-PE immunotoxins toward B-cells is particularly dependent on the molecular form of PE that is used. In this study, RFB4 coupled to PE35 or PE35KDEL was quite toxic to cultured CD22-positive B-cells, while RFB4 conjugated to PE38 was at least 10fold less active. This difference probably represents a reduced ability of this cell type to proteolytically process PE, although some of the reduction in cytotoxicity must be attributed to reduced binding of RFB4-PE38 to CD22.

Mansfield et al.

Recently, native PE was shown to require furin-mediated cleavage to produce the C-terminal fragment that translocates to the cell cytosol (27). Failure to generate this fragment results in reduced or no toxicity. If cells lack furin or furin is present only in subcellular locations separate from the endocytic pathway of the toxin, one would expect reduced toxicity compared to immunotoxins in which PE does not need to be cleaved by furin. Conjugation of RFB4 IgG to the PE mutants PE35 and PE35KDEL through disulfide bonds is efficient and results in immunotoxin molecules that are very active against CD22-positive cells both in vitro and in vivo. Recombinant immunotoxins made of PE and RFB4 are currently being developed. Previous results have indicated that recombinant immunotoxins are often more active than their corresponding chemical conjugates (2830). In addition, recombinant immunotoxins are 25% of the size of chemical conjugates using whole antibody and thus are more likely to penetrate tumor masses. ACKNOWLEDGMENT

We thank Bob Kreitman for his help in initiating animal experiments and for providing the plasmid pRK35K and Ellen Vitetta for providing the RFB4 antibody and for helpful comments and discussion. LITERATURE CITED (1) Uckun, F. M., Manivel, C., Arthur, D., Chelstrom, L. M., Finnegan, D., Tuel-Ahlgren, L., Irvin, J. D., Myers, D. E., and Gunther, R. (1992) In vivo efficacy of B43 (anti-CD19)pokeweed antiviral protein immunotoxin against human pre-B cell acute lymphoblastic leukemia in mice with severe combined immunodeficiency. Blood 79, 2201-2214. (2) Ghetie, M.-A., Richardson, J., Tucker, T., Jones, D., Uhr, J. W., and Vitetta, E. S. (1991) Antitumor activity of Fab′ and IgG-anti-CD22 immunotoxins in disseminated human B lymphoma grown in mice with severe combined immunodeficiency disease: Effect on tumor cells in extranodal sites. Cancer Res. 51, 5876-5880. (3) Francisco, J. A., Gilliland, L. K., Stebbins, M. R., Norris, N. A., Ledbetter, J. A., and Siegall, C. B. (1995) Activity of a single-chain immunotoxin that selectively kills lymphoma and other B-lineage cells expressing the CD40 antigen. Cancer Res. 55, 3099-3104. (4) Ghetie, M.-A., May, R. D., Till, M., Uhr, J. W., Ghetie, V., Knowles, P. P., Relf, M., Brown, A., Wallace, P. M., Janossy, G., Amlot, P., Vitetta, E. S., and Thorpe, P. E. (1988) Evaluation of ricin A chain-containing immunotoxins directed against CD19 and CD22 antigens on normal and malignant human B-cells as potential reagents for in vivo therapy. Cancer Res. 48, 2610-2617. (5) Bonardi, M. A., French, R. R., Amlot, P., Gromo, G., Modena, D., and Glennie, M. J. (1993) Delivery of saporin to human B-cell lymphoma using bispecific antibody: Targeting via CD22 but not CD19, CD37, or immunoglobulin results in efficient killing. Cancer Res. 53, 3015-3021. (6) French, R. R., Penney, C. A., Browning, A. C., Stirpe, F., George, A. J. T., and Glennie, M. J. (1995) Delivery of the ribosome-inactivating protein, gelonin, to lymphoma cells via CD22 and CD38 using bispecific antibodies. Br. J. Cancer 71, 986-994. (7) Uckun, F. M., Ramakrishnan, S., and Houston, L. L. (1985) Immunotoxin-mediated elimination of clonogenic tumor cells in the presence of human bone marrow. J. Immunol. 134, 2010-2016. (8) Kreitman, R. J., Hansen, H. J., Jones, A. L., FitzGerald, D. J. P., Goldenberg, D. M., and Pastan, I. (1993) Pseudomonas exotoxin-based immunotoxins containing the antibody LL2 or LL2-Fab′ induce regression of subcutaneous human B-cell lymphoma in mice. Cancer Res. 53, 819-825. (9) Ogata, M., Fryling, C. M., Pastan, I., and FitzGerald, D. J. (1992) Cell-mediated cleavage of Pseudomonas exotoxin between Arg279 and Gly280 generates the enzymatically active

Bioconjugate Chem., Vol. 7, No. 5, 1996 563

Characterization of RFB4−PE Immunotoxins fragment which translocates to the cytosol. J. Biol. Chem. 267, 25396-25401. (10) Theuer, C. P., Kreitman, R. J., FitzGerald, D. J., and Pastan, I. (1993) Immunotoxins made with a recombinant form of Pseudomonas exotoxin A that do not require proteolysis for activity. Cancer Res. 53, 340-347. (11) Kreitman, R. J., Chaudhary, V. K., Kozak, R. W., FitzGerald, D. J. P., Waldmann, T. A., and Pastan, I. (1992) Recombinant toxins containing the variable domains of the anti-Tac monoclonal antibody to the interleukin-2 receptor kill malignant cells from patients with chronic lymphocytic leukemia. Blood 80, 2344-2352. (12) Campana, D., Janossy, G., Bofill, M., Trejdosiewicz, L. K., Ma, D., Hoffbrand, V., Mason, D. Y., Lebacq, A.-M., and Forster, H. K. (1985) Human B cell development. I. Phenotypic differences of B lymphocytes in the bone marrow and peripheral lymphoid tissue. J. Immunol. 134, 1524-1530. (13) Li, J.-L., Shen, G.-L., Ghetie, M.-A., May, R. D., Till, M., Ghetie, V., Uhr, J. W., Janossy, G., Thorpe, P. E., Amlot, P., and Vitetta, E. S. (1989) The epitope specificity and tissue reactivity of four murine monoclonal anti-CD22 antibodies. Cell. Immunol. 118, 85-99. (14) Vitetta, E. S., Stone, M., Amlot, P., Fay, J., May, R., Till, M., Newman, J., Clark, P., Collins, R., Cunningham, D., Ghetie, V., Uhr, J. W., and Thorpe, P. E. (1991) Phase I immunotoxin trial in patients with B cell lymphoma. Cancer Res. 51, 4052-4058. (15) Amlot, P., Stone, M. J., Cunningham, D., Fay, J., Newman, J., Collins, R., May, R., McCarthy, M., Richardson, J., Ghetie, V., Ramilo, O., Thorpe, P. E., Uhr, J. W., and Vitetta, E. S. (1993) A phase I study of anti-CD22 deglycosylated ricin A chain immunotoxin in the treatment of B cell lymphomas resistant to conventional therapy. Blood 82, 2624. (16) Sausville, E. A., Headlee, D., Stetler-Stevenson, M., Jaffe, E. S., Solomon, D., Figg, W. D., Herdt, J., Kopp, W. C., Rager, H., Steinberg, S. M., Ghetie, V., Schindler, J., Uhr, J., Wittes, R. E., and Vitetta, E. S. (1995) Continuous infusion of the anti-CD22 immunotoxin IgG-RFB4-SMPT-dgA in patients with B-cell lymphoma: A phase I study. Blood 85, 34573465. (17) Ghetie, V., Thorpe, P., Ghetie, M. A., Knowles, P., Uhr, J. W., and Vitetta, E. S. (1991) The GLP large scale preparation of immunotoxins containing deglycosylated ricin A chain and a hindered disulfide bond. J. Immunol. Methods 142, 223230. (18) FitzGerald, D. J. P. (1987) Construction of immunotoxins using Pseudomonas exotoxin A. Methods Enzymol. 151, 139145. (19) Soler-Rodriguez, A.-M., Ghetie, M.-A., OppenheimerMarks, N., Uhr, J. W., and Vitetta, E. S. (1993) Ricin A-chain and ricin A-chain immunotoxins rapidly damage human endothelial cells: Implications for vascular leak syndrome. Exp. Cell. Res. 206, 227-234. (20) Anderson, K. C., Bates, M. P., Slaughenhoupt, B. L., Pinkus, G. S., Schlossman, S. F., and Nadler, L. M. (1984)

Expression of human B cell-associated antigens on leukemias and lymphomas: A model of human B cell differentiation. Blood 63, 1424-1433. (21) Ghetie, M.-A., Richardson, J., Tucker, T., Jones, D., Uhr, J. W., and Vitetta, E. S. (1990) Disseminated or localized growth of a human B-cell tumor (Daudi) in SCID mice. Int. J. Cancer 45, 481-485. (22) Pai, L. H., Wittes, R. E., Setser, A., Goldspiel, B., FitzGerald, D., Willingham, M. C., and Pastan, I. (1994) Phase I study of the immunotoxin LMB-1, an anti-cancer murine MAb B3, coupled to a recombinant form of Pseudomonas exotoxin. Proc. Am. Assoc. Cancer Res. 35, 507. (23) Davey, R. T., Boenning, C. M., Herpin, B. R., Batts, D. H., Metcalf, J. A., Wathen, L., Cox, S. R., Polis, M. A., Kovacs, J. A., and Falloon, J. (1994) Use of recombinant soluble CD4 Pseudomonas exotoxin, a novel immunotoxin, for treatment of persons infected with human immunodeficiency virus. J. Infect. Dis. 170, 1180-1188. (24) Ramachandran, R. V., Katzenstein, D. A., Wood, R., Batts, D. H., and Merigan, T. C. (1994) Failure of short-term CD4PE40 infusions to reduce virus load in human immunodeficiency virus-infected persons. J. Infect. Dis. 170, 1009-1013. (25) Pai, L. H., Bookman, M. A., Ozols, R. F., Young, R. C., Smith, J. W., II, Longo, D. L., Gould, B., Frankel, A., McClay, E. F., Howell, S., Reed, E., Willingham, M. C., FitzGerald, D. J., and Pastan, I. (1991) Clinical evaluation of intraperitoneal Pseudomonas exotoxin immunoconjugate OVB3-PE in patients with ovarian cancer. J. Clin. Oncol. 9, 2095-2103. (26) Kuan, C.-T., Pai, L. H., and Pastan, I. (1995) Immunotoxins containing Pseudomonas exotoxin that target LeY damage human endothelial cells in an antibody-specific mode: relevance to vascular leak syndrome. Clin. Cancer Res. 1, 15891594. (27) Chiron, M. F., Fryling, C. M., and FitzGerald, D. F. (1994) Cleavage of pseudomonas exotoxin and diphtheria toxin by a furin-like enzyme prepared from beef liver. J. Biol. Chem. 269, 18167-18176. (28) Batra, J. K., Jinno, Y., Chaudhary, V. K., Kondo, T., Willingham, M. C., FitzGerald, D. J., and Pastan, I. (1989) Antitumor activity in mice of an immunotoxin made with anti-transferrin receptor and a recombinant form of Pseudomonas exotoxin. Proc. Natl. Acad. Sci. U.S.A. 86, 8545-8549. (29) Chaudhary, V. K., Queen, C., Junghans, R. P., Waldmann, T. A., FitzGerald, D. J., and Pastan, I. (1989) A recombinant immunotoxin consisting of two antibody variable domains fused to Pseudomonas exotoxin. Nature 339, 394-397. (30) Brinkmann, U., Pai, L. H., FitzGerald, D. J., Willingham, M., and Pastan, I. (1991) B3(Fv)-PE38KDEL, a single-chain immunotoxin that causes complete regression of a human carcinoma in mice. Proc. Natl. Acad. Sci. U.S.A. 88, 8616-8620.

BC960043Y