Anti-Tac(Fv)-PE38, also called LMB-2, is a very active recombinant immunotoxin that has produced eight responses, including a durable clinical complete remission in a recently completed phase I trial of leukemias and lymphomas. Dose escalation was limited by liver toxicity. We have noted that the Fv of anti-Tac has an isoelectric point (pI) of 10.2. We hypothesize that the overall positive charge on the Fv portion of anti-Tac(Fv)-PE38 contributes to nonspecific binding to liver cells and results in dose-limiting liver toxicity. We have used a mouse model to investigate the basis of this toxicity and found that lowering the pI of the Fv of anti-Tac from 10.2 to 6.82 by selective mutation of surface residues causes a 3-fold decrease in animal toxicity and hepatic necrosis. This change in pI did not significantly alter the CD25 binding affinity, the cytotoxic activity toward target cells, or antitumor activity, resulting in a 3-fold improvement in the therapeutic index. If this decreased toxicity occurs in humans, it should greatly increase the clinical utility of this immunotoxin.

Immunotoxins are composed of Abs or Ab fragments attached to a toxin. Immunotoxins were originally produced by coupling whole Abs to plant or bacterial toxins such as ricin, the α-chain of ricin, saporin, pokeweed antiviral protein, diphtheria toxin, or Pseudomonas exotoxin A (PE)2 (1). Subsequently, immunotoxins were produced using methods of genetic engineering in which the Fv fragments of Abs were fused to the toxin and the recombinant protein was made in Escherichia coli (2). Our laboratory has focused on the development of recombinant toxins in which the Fv fragments of Abs are fused to a 38-kDa mutant form of PE (3). PE is a 66-kDa protein composed of three domains: a binding domain, a translocation domain, and an ADP-ribosylating domain. Recombinant immunotoxins are made by deleting the cell-binding domain of PE and replacing it with the Fv portion of an Ab (4). During the past several years we have made a large number of recombinant toxins using different Abs (5). Three of these recombinant immunotoxins have now been evaluated in phase I trials in patients with cancer. All the recombinant immunotoxins that have been brought to clinical trials have been shown to cure human cancer xenografts growing in nude mice and to be relatively well tolerated by mice and monkeys (5). In a recently completed phase I trial, eight partial responses have been observed in patients with hematopoietic malignancies treated with anti-Tac(Fv)-PE38 (LMB-2). However, side effects have been observed that cannot be attributed to targeting IL-2R positive cells. These side effects limit the amount of immunotoxin that can be given to humans. It seems likely that if we can understand the mechanism or basis of the nonspecific toxicity, we will be able to use genetic engineering to make new molecules with fewer side effects and, therefore, we will be able to give larger doses to humans, resulting in more responses.

The toxic side effects of recombinant immunotoxins are of two types. One type of toxicity results from specific targeting of normal cells which display the same Ag as the cancer cell. The second type of toxicity is nonspecific and usually is characterized by damage to liver cells; this increases the serum levels of serum glutamic oxaloacetic transaminase and serum glutamic pyruvate transaminase (6), although other toxic effects may occur (1).

A phase I trial with anti-Tac(Fv)-PE38 has just been completed. In that trial, dose-limiting toxicity was most frequently due to liver damage. This type of toxicity is nonspecific because liver cells do not express IL-2 receptors. In the current study we have investigated the basis of the nonspecific toxicity of anti-Tac(Fv)-PE38 using mice in which liver toxicity is dose-limiting as in humans and monkeys. We have noticed that the Fv portion of anti-Tac(Fv)-PE38 has a very high isoelectric point (pI) of 10.2, whereas another recombinant immunotoxin, RFB4(Fv)-PE38, which is targeted at the CD22 Ag on human B cells, has a much lower pI of 7.67 and is extremely well tolerated by mice (7). This result suggests that the pI of the Fv portion of the recombinant immunotoxin might contribute to nonspecific toxicity because the high pI makes the protein positively charged at neutral pH and would favor binding to negatively charged groups on the surface of cells in the liver. In the current study, we have used site-directed mutagenesis and molecular modeling to decrease the pI of the Fv portion of anti-Tac(Fv)-PE38 from 10.21 to 6.82. We find that a mutant of anti-Tac(Fv)-PE38 with a pI of 6.82 has the same cytotoxic activity on target cells and the same antitumor activity as the parental anti-Tac(Fv)-PE38. However, the toxicity of this mutant immunotoxin in mice is decreased by more than 3-fold.

The pI of each Fv was calculated using a program in the Genetics Computer Group (Madison, WI) package that is available through the web site http://molbio.info.nih.gov/molbio/gcglite/protform.htm. In the Fv portion, cysteines have no charge because they are disulfide bonded. These were converted to serine for the pI calculation. LMB-2 has a high pI (10.21), whereas the pI of RFB4 is 7.67.

Mutagenesis of anti-Tac(Fv)-PE38 was done by Kunkel’s method (8) with some modifications. CJ 236 cells were transformed with pRK79. After selection on Luria-Bertani plates containing 2% glucose, 30 μg/ml chloramphenicol, and 100 μg/ml ampicillin, the transformants were grown in 2× YT medium containing glucose and the antibiotics, as mentioned above, and 0.025% uridine at 37°C. At an OD600 of 0.36, the cells were infected with the helper phage M13 at a multiplicity of infection of 5. After incubation at 37°C/110 rpm for 1 h, the culture was maintained at 37°C/300 rpm for another 6 h. The bacterial cells then were precipitated by centrifugation and the phage from the supernatant was precipitated with polyethylene glycol. The single-stranded, uracil-containing DNA from the purified phage was extracted with phenol and chloroform and precipitated with lithium chloride and chilled ethanol at final concentrations of 0.8 M and 75%, respectively. This ssDNA codes for the sense strand of anti-Tac(Fv)-PE38. The following primers were used for anti-Tac(Fv)-PE38 mutagenesis: M1 VL Q1D, 5′-TGCTGGAGACTGGGTGAGAACGATATCAGAGCCGCCGCCACCCGAGCCGCCACCGCCCGAGCCACC-3′; M1 VL Q1D+VL3E, 5′-GATTGCTGGAGACTGGGTGAGTTCGATATCAGAGCCGCCGCCACCCGAGCCGCCACCGCCCGAG-3′; M1 VL S70D, 5′-GCTGATTGTGAGAGAGTAATCGGTCCCGGATCCACTGCCACTGAAGCG-3′, M1 VL S100D, 5′-TTTGAGCTCCAGCTTGGTACCATCACCGAACGTGAGTGGGTA-3′; M1 VH Q1E, 5′-TGAGGCCCCAGGTTTTGCTAGCTCAGCCCCAGACTGCTGCAGATGGACCTCCATATGTATATCTCC-3′; M1 VH Q6E, 5′-TGAGGCCCCAGGTTTTGCTAGCTCAGCCCCAGACTCCTGCAGATGGACCTGCAT-3′; M1 VH Q1E+H3Q+Q5V+Q6E, 5′-TGAGGCCCCAGGTTTTGCTAGCTCAGCCCCAGACTCCACCAGTTGGACCTCCATATGTATATCTCC-3′; and M1 VH G42E, 5′-AATCCATTCCAGACCCTGTTCAGGTCTCTGTTTTACCCAGTGCAT-3′.

The primers were phosphorylated using polynucleotide kinase and T4 DNA ligase buffer from Boehringer Mannheim (Indianapolis, IN). These phosphorylated primers were used to introduce the mutations in the uracil template of pRK79 using Bio-Rad (Richmond, CA) Muta-Gene kit. The product at the end of the mutagenesis reaction was used to directly transform DH5α competent cells. Minipreps were made from single colonies and analyzed with the restriction enzymes, sites which had been introduced by the primers for mutagenesis. Mutations in the clones were confirmed by automated DNA sequencing.

The components of anti-Tac(Fv)-PE38 and 10 mutants were expressed in Escherichia coli BL21(λDE3) and accumulated in inclusion bodies as previously described for other recombinant immunotoxins (7). Inclusion bodies were solubilized in GuCl, reduced with DTE (dithioerythritol), and refolded by dilution in a refolding buffer containing arginine to prevent aggregation, and oxidized and reduced glutathione to facilitate redox shuffling (7). Active monomeric protein was purified from the refolding solution by ion exchange and size exclusion chromatography to near homogeneity as described (7, 9). Protein concentrations were determined by Bradford assay (Coomassie Plus; Pierce, Rockford, IL).

The specific cytotoxicity of the anti-Tac(Fv)-PE38 and its mutants was assessed by protein synthesis inhibition assays (inhibition of incorporation of tritium-labeled leucine into cellular protein) in 96-well plates as previously described (7, 9). The activity of the molecule is defined by the IC50, the toxin concentration that reduced incorporation of radioactivity by 50% compared with cells that were not treated with toxin. The specificity is obtained by comparing the activity toward Ag-positive cells vs toxicity against Ag-negative cells. Fresh malignant cells were partially purified from patients with B-cell leukemia and incubated with recombinant toxins as previously described (10).

Groups of five female (∼20 g) NIH Swiss mice were given injections i.v. through the tail vein of 200 μl of escalating doses of anti-Tac(Fv)-PE38 or its mutant immunotoxins diluted in PBS-human serum albumin (HSA)3 (0.2%). Animals were observed 2 wk for mortality. The LD50 is the calculated dose of immunotoxin that kills 50% of the animals.

NIH Swiss mice were sacrificed 24 h after injection of immunotoxin. Liver and kidney were fixed by 10% Formaline. Sections from each of these organs were stained with hematoxylin and eosin (H&E) and examined histologically.

Antitumor activity of anti-Tac(Fv)-PE38 and M1 (Fv)-PE38 was determined in nude mice bearing ATAC4 cells. These cells (2.5 × 106) were injected s.c. into nude mice on day 0. Tumors about 50 mm3 developed in animals by day 4 after tumor implantation. Starting on day 4, animals were treated with i.v. injections of anti-Tac(Fv)-PE38 and M1 (Fv)-PE38 diluted in 0.2 ml of PBS-HSA (0.2%). Therapy was given once every other day (on days 4, 6, and 8) and each treatment group consisted of 5 animals. Tumors were measured with a caliper every 2 or 3 days and the volume of the tumor was calculated by using the formula: tumor volume (mm3) = length × (width)2 × 0.4.

Recombinant immunotoxins are usually composed of the Fv fragments of mouse mAbs fused to a 38,000 m.w. form of PE. The Fv fragment replaces the binding domain of PE and directs the toxin to Ags on cancer cells. Because mouse mAbs are used to make immunotoxins, they are selected by the mouse immune system not to react with mouse Ags. Therefore, the toxicities that occur in mice that have been administered these recombinant immunotoxins are due to nonspecific interactions of the recombinant immunotoxins with mouse tissues. The data in Table I show the LD50 in mice of three different recombinant immunotoxins that are made with three different mouse mAbs, none of which specifically react with mouse cells. Anti-Tac(Fv)-PE38 is targeted at the α subunit of the human IL-2 receptor (CD25) (2, 11). SS(Fv)-PE38 is targeted at an Ag termed “mesothelin” that is found on mesothelial cells, ovarian cancers, and mesotheliomas (9). RFB4(Fv)-PE38 is targeted to CD22. The data in Table I show that when a single dose of each of the recombinant immunotoxins is injected i.v. in mice, the LD50s vary widely. The most toxic is anti-Tac(Fv)-PE38 with an LD50 of 0.34 mg/kg. The least toxic is RFB4(Fv)-PE38 which has an LD50 of 0.81 mg/kg. SS(Fv)-PE38 has intermediate LD50 of 0.68 mg/kg. These three molecules are identical in the PE portion but vary in the Fv. Thus, we concluded that these differences in toxicity were due to some different property in the Fv. The data in Table I also show pI of the Fv portion of these recombinant immunotoxins. They vary widely from a pI of 10.2 for anti-Tac(Fv) to a pI of 7.67 for RFB4(Fv). The PE portion of the molecules is rather acidic, with a pI of 5.4. Thus the nonspecific toxicity increases with pI. The cause of death resulting from injection of all these immunotoxins is acute hepatic necrosis. Therefore, it seemed likely that the toxicity in mice is due to some direct interaction of the Fv portion of the recombinant immunotoxins with the liver. Because the liver cell membranes have a negative charge, we hypothesized that the positive charge at neutral pH on the Fv portion of the recombinant immunotoxins would favor higher nonspecific binding and uptake of the recombinant immunotoxins by cells in the liver. If this were the case, one should be able to decrease the toxicity of anti-Tac(Fv)-PE38 by decreasing its pI by site-directed mutagenesis of the framework regions leaving the CDRs intact and, therefore, not affect the specific interactions of anti-Tac(Fv)-PE38 with the human IL-2 receptor.

Table I.

Toxicity and isoelectric point of three single-chain immunotoxinsa

ImmunotoxinpI of scFvLD50 (mg/kg)
Anti-Tac(Fv)-PE38 10.21 0.34 
SS(Fv)-PE38 8.91 0.68 
RFB4(Fv)-PE38 7.67 0.81 
ImmunotoxinpI of scFvLD50 (mg/kg)
Anti-Tac(Fv)-PE38 10.21 0.34 
SS(Fv)-PE38 8.91 0.68 
RFB4(Fv)-PE38 7.67 0.81 
a

In comparing the toxicity of immunotoxins at the same dose level, significant differences from anti-Tac(Fv)-PE38 are present (p < 0.001, χ2 test).

Accordingly, we compared the amino acid sequence of the Fv portion of anti-Tac with the Fv portion of RFB4 and determined the number of charged amino acids (12). RFB4(Fv) has 10 lysines, 7 arginines, 10 aspartates, 8 glutamates, and 2 histidines, whereas anti-Tac(Fv) has 13 lysines, 6 arginines, 5 aspartates, 8 glutamates, and 3 histidines (Table II). These differences in positive and negative residues are responsible for the differences in the pI between RFB4(Fv) and anti-Tac(Fv). Therefore, we decided to make anti-Tac(Fv) more like RFB4(Fv). Accordingly, we aligned the RFB4 and anti-Tac(Fv) sequences and mutated several surface-exposed residues in the framework region in anti-Tac(Fv) to make them resemble sequences in RFB4(Fv). This had the effect of lowering the pI. Initially, we made a series of recombinant immunotoxins in which single residues were mutated to acidic residues one at a time. The residues in anti-Tac that were changed are Q1, V3, S70, and S100 of VL; and Q1 and G42 of VH (Table III). In general, the residues were changed to aspartate or glutamate. In addition, mutations H3Q, Q5V, and Q6E were introduced in the VH to make the amino terminus more closely resemble the amino terminus of RFB4. After these single-residue mutants were analyzed for cytotoxicity and mouse toxicity, two more recombinant immunotoxins (M1 and M2) were constructed in which mutations were combined (see Table III).

Table II.

Number of charged residues in Fvs of anti-Tac, RFB4, and M1

FvVLVHpI
KREDHKREDH
RFB-4 7.67 
Anti-Tac 10.21 
M1 6.82 
FvVLVHpI
KREDHKREDH
RFB-4 7.67 
Anti-Tac 10.21 
M1 6.82 
Table III.

The location of amino acids that were mutated in M1a

VHVL
1234564212370100
RFB-4    
Anti-Tac    
M1       
VHVL
1234564212370100
RFB-4    
Anti-Tac    
M1       
a

Numbering is according to Kabat et al. (10 ).

Overall, we constructed 10 mutants of anti-Tac(Fv)-PE38. The mutations were confirmed by DNA sequencing and the recombinant proteins were expressed in E. coli, where they all accumulated in inclusion bodies. All immunotoxins were purified by ion-exchange and size-exclusion chromatography from renatured inclusion bodies. Each of the 11 immunotoxins (10 mutants and 1 wild type) eluted as a monomer upon TSK gel-filtration chromatography and migrated as a single, major band of about 63 kDa in SDS-PAGE (data not shown). The data in Table IV show the activity of each of the mutant molecules on Ag-positive ATAC-4 cells that express CD25 and the calculated pI of the Fv portion. All of the molecules constructed had the same activity on ATAC-4 cells. Therefore, by introducing negatively charged residues, there was no loss of specific cytotoxic activity on target cells expressing human IL-2Rα. Introduction of single negatively charged residues lead to only a small decrease in the calculated pI. Introduction of one or two negatively charged residues also led to a small decrease of pI from 10.2 to 9.95. Only when several mutations were combined did the pI fall significantly. The pI fell to 6.82 for mutant M1 and to 7.76 for mutant M2.

Table IV.

Mutations in the Fv portion of anti-Tac(Fv)-PE38, cytotoxic activity on ATAC4 cells, mouse toxicity, and pIsa

Fv MutationsIC50 (ng/ml)LD50 (mg/kg)pI of Fv
None 0.07 ± 0.02 0.34 10.21 
VL Q1D 0.07 ± 0.02 0.32 10.09 
VL Q1D, V3E 0.07 ± 0.02 0.24 9.95 
VL S70D 0.07 ± 0.02 0.46* 10.09 
VL S100D 0.07 ± 0.02 0.22 10.09 
VH Q1E 0.05 ± 0.02 0.25 10.09 
    
VH Q6E 0.05 ± 0.02 0.25 10.09 
VH Q1E, H3Q, Q5V, Q6E 0.07 ± 0.02 0.46 9.95 
VH G42E 0.07 ± 0.02 0.17 10.09 
M2 (VH Q1E, H3Q, Q5V, Q6E) 0.07 ± 0.02 0.56* 7.76 
(VL Q1D, V3E, S70D, S100D)    
M1 (VH Q1E, H3Q, Q5V, Q6E, G42E) 0.07 ± 0.02 1.22* 6.82 
(VL Q1D, V3E, S70D, S100D)    
Fv MutationsIC50 (ng/ml)LD50 (mg/kg)pI of Fv
None 0.07 ± 0.02 0.34 10.21 
VL Q1D 0.07 ± 0.02 0.32 10.09 
VL Q1D, V3E 0.07 ± 0.02 0.24 9.95 
VL S70D 0.07 ± 0.02 0.46* 10.09 
VL S100D 0.07 ± 0.02 0.22 10.09 
VH Q1E 0.05 ± 0.02 0.25 10.09 
    
VH Q6E 0.05 ± 0.02 0.25 10.09 
VH Q1E, H3Q, Q5V, Q6E 0.07 ± 0.02 0.46 9.95 
VH G42E 0.07 ± 0.02 0.17 10.09 
M2 (VH Q1E, H3Q, Q5V, Q6E) 0.07 ± 0.02 0.56* 7.76 
(VL Q1D, V3E, S70D, S100D)    
M1 (VH Q1E, H3Q, Q5V, Q6E, G42E) 0.07 ± 0.02 1.22* 6.82 
(VL Q1D, V3E, S70D, S100D)    
a

Agents significantly less toxic than unmodified anti-Tac(Fv)-PE38 by χ2 analysis are indicated: ∗, p < 0.001; †, p < 0.005; ‡, p < 0.01. Based on comparison of toxicity results at different dose levels, M1 was 3-fold less toxic than LMB-2 at a significance level of p < 0.025 and was 2.7-fold less toxic at a significance level of p < 0.001.

Then each of these molecules was evaluated for its nonspecific toxicity in mice (Table IV). Groups of five mice or more were injected with varying doses of immunotoxin and observed for 2 wk. Almost all of the deaths occurred within 72 h after treatment. In general, single amino acid changes did not produce large changes in mouse toxicity, although both the decreases and increases in nonspecific mouse toxicity were observed by single mutations. However, when all of the mutations were combined into one molecule to produce scM1, which had a pI of 6.82, the single-dose toxicity in mice was greatly diminished; we found that the LD50 rose from 0.34 mg/kg to 1.22 mg/kg. Mutant M2 has a pI of 7.76 and contains all the mutations found within M1 except for the G42E mutation in VH. The animal toxicity (LD50) of M2(Fv)-PE38 is 0.56 mg/kg, which is intermediate between the toxicity of anti-Tac(Fv)-PE38 and M1(Fv)-PE38.

To determine the cause of death, mice were sacrificed 24 h after administration of 9 μg (0.45 mg/kg) of anti-Tac(Fv)-PE38 or M1(Fv)-PE38 and sections of the livers were prepared and stained with H&E. In the mice treated with anti-Tac(Fv)-PE38, there was evidence of severe hemorrhagic liver necrosis, whereas the livers from the mice treated with M1(Fv)-PE38 appeared normal (Fig. 1). The finding of hemorrhagic necrosis raises the possibility that the liver endothelial cells are being damaged as part of the process of liver toxicity.

FIGURE 1.

Histological sections of a liver taken from a mouse 48 h after it was treated with one dose of 9 μg (0.45 mg/kg) of LMB-2 (anti-Tac(Fv)-PE38) or M1(Fv)-PE38. The control mouse was treated with HSA. All sections were stained with H&E. Magnification, ×200.

FIGURE 1.

Histological sections of a liver taken from a mouse 48 h after it was treated with one dose of 9 μg (0.45 mg/kg) of LMB-2 (anti-Tac(Fv)-PE38) or M1(Fv)-PE38. The control mouse was treated with HSA. All sections were stained with H&E. Magnification, ×200.

Close modal

We selected M1(Fv)-PE38 for further study because of its low animal toxicity. Before doing further animal studies, we investigated its cytotoxic activity on several other cell lines (Table V). HUT102 is a cell line from a patient with adult T cell leukemia. Anti-Tac(Fv)-PE38 and M1(Fv)-PE38 had similar cytotoxic activities on this cell line with an IC50 value of 0.1 ng/ml. Raji is a cell line of B cell lineage and does not express IL-2 receptors. Neither immunotoxin was active at 1000 ng/ml. We also examined A431 cells which do not have IL-2 receptors and are the parent of ATAC4. The IC50 was 390 ng/ml for anti-Tac(Fv)-PE38 and 510 ng/ml for M1(Fv)-PE38. These very high IC50 values compared with the IC50 of 0.07 ng/ml on ATAC4 cells indicate that the cytotoxic activities of both molecules are highly specific. To determine whether M1(Fv)-PE38 would retain cytotoxicity toward malignant target cells directly obtained from patients, it was incubated with peripheral blood mononuclear cells from two patients with CD25+ B-cell leukemias. In both patients the malignant cells composed >95% of the cell sample tested. As shown in Table V, there was no significant difference in the cytotoxic activity of the two immunotoxins, indicating that the mutations which lowered the pI did not impair cytotoxicity or stability of the protein during its interaction with fresh malignant patient cells.

Table V.

Activity of anti-Tac(Fv)-PE38 on cell lines

CellsTypeIC50 (ng/ml)
scM1anti-Tac(Fv)-PE38
ATAC4 Epidermoid 0.07 ± 0.02 0.07 ± 0.02 
HUT102 ATL 0.1 ± 0.02 0.1 ± 0.02 
A431 Epidermoid 510 ± 90 390 ± 10 
SP2/Tac Plasmacytoma 0.08 ± 0.005 0.08 ± 0.006 
Raji Burkitt’s Lymphoma >1000 >1000 
Patient 1 B-cell leukemia 3 ± 0.3 2.5 ± 0.25 
Patient 2 B-cell leukemia 64 ± 18 56 ± 45 
CellsTypeIC50 (ng/ml)
scM1anti-Tac(Fv)-PE38
ATAC4 Epidermoid 0.07 ± 0.02 0.07 ± 0.02 
HUT102 ATL 0.1 ± 0.02 0.1 ± 0.02 
A431 Epidermoid 510 ± 90 390 ± 10 
SP2/Tac Plasmacytoma 0.08 ± 0.005 0.08 ± 0.006 
Raji Burkitt’s Lymphoma >1000 >1000 
Patient 1 B-cell leukemia 3 ± 0.3 2.5 ± 0.25 
Patient 2 B-cell leukemia 64 ± 18 56 ± 45 

To evaluate antitumor activity, mice were implanted s.c. on day 0 with ATAC4 tumor cells and i.v. therapy was initiated on day 4 using increasing doses of anti-Tac(Fv)-PE38 or M1(Fv)-PE38. Each agent was given every other day for three doses. Typical tumor regression results are illustrated in Fig. 2 for mice treated with increasing doses of M1(Fv)-PE38. The data in Table VI shows the toxicity at each dose level and a summary of tumor responses. The data confirm that anti-Tac(Fv)-PE38 is much more toxic to mice than M1(Fv)-PE38. With anti-Tac(Fv)-PE38, 1 of 15 animals died at 0.075 mg/kg × 3; 2 of 15 died at 0.15 mg/kg × 3; and all 15 died at 0.3 mg/kg × 3. In contrast, there were no deaths with M1(Fv)-PE38 even at 0.5 mg/kg × 3. Thus, the three-dose toxicity results in nude mice confirm the one-dose study in normal mice showing that M1(Fv)-PE38 is much less toxic than anti-Tac(Fv)-PE38. The effect of the two immunotoxins on tumor size is also shown in Table VI. The molecules were equally active at the 0.075 mg/kg × 3 and 0.15 mg/kg × 3 doses, producing 6 of 15 or 7 of 15 complete regression (CR) at the lower dose, and 11 of 15 and 14 of 15 CRs at the higher dose. At the dose levels where anti-Tac(Fv)-PE38 caused deaths, M1(Fv)-PE38 produced 14 of 15 CRs (0.3 mg/kg) and 5 of 5 CRs (0.5 mg/kg). These data clearly show the usefulness of being able to give higher doses of immunotoxin.

FIGURE 2.

Anti-tumor effect of M1(Fv)-PE38 in nude mice. Groups of five animals were injected with 2.5 × 106 ATAC4 cells on day 0. On day 4, tumors reached a size of 0.05 cm3. Animals were treated i.v. on days 4, 6, and 8 with 0.025 mg/kg (○), 0.075 mg/kg (▵), or 0.3 mg/kg (□) of M1(Fv)-PE38 in Dulbecco’s modified PBS containing 0.2% HSA. Control groups received carrier alone (▪). No death or toxicity was observed at these doses.

FIGURE 2.

Anti-tumor effect of M1(Fv)-PE38 in nude mice. Groups of five animals were injected with 2.5 × 106 ATAC4 cells on day 0. On day 4, tumors reached a size of 0.05 cm3. Animals were treated i.v. on days 4, 6, and 8 with 0.025 mg/kg (○), 0.075 mg/kg (▵), or 0.3 mg/kg (□) of M1(Fv)-PE38 in Dulbecco’s modified PBS containing 0.2% HSA. Control groups received carrier alone (▪). No death or toxicity was observed at these doses.

Close modal
Table VI.

Effect of increasing doses of anti-Tac(Fv)-PE38 and scM1(Fv)-PE38 on toxicity and tumor responses in micea

Death/TotalbPRCR
Anti-Tac(Fv)-PE38    
0.025 mg/kg× 3 0 /5 5 /5 0 /5 
0.075 mg/kg× 3 1 /15 15 /15 7 /15 
0.15 mg/kg× 3 2 /15 13 /15 11 /15 
0.3 mg/kg× 3 15 /15 NE NE 
scM1(Fv)-PE38    
0.025 mg/kg× 3 0 /5 5 /5 0 /5 
0.075 mg/kg× 3 0 /15 15 /15 6 /15 
0.15 mg/kg× 3 0 /15 15 /15 14 /15 
0.3 mg/kg× 3 0 /15 15 /15 14 /15 
0.5 mg/kg× 3 0 /5 5 /5 5 /5 
Death/TotalbPRCR
Anti-Tac(Fv)-PE38    
0.025 mg/kg× 3 0 /5 5 /5 0 /5 
0.075 mg/kg× 3 1 /15 15 /15 7 /15 
0.15 mg/kg× 3 2 /15 13 /15 11 /15 
0.3 mg/kg× 3 15 /15 NE NE 
scM1(Fv)-PE38    
0.025 mg/kg× 3 0 /5 5 /5 0 /5 
0.075 mg/kg× 3 0 /15 15 /15 6 /15 
0.15 mg/kg× 3 0 /15 15 /15 14 /15 
0.3 mg/kg× 3 0 /15 15 /15 14 /15 
0.5 mg/kg× 3 0 /5 5 /5 5 /5 
a

CR, complete regression of tumor for a minimum of 4 days; PR, reduction of the sum of the tumor length and width by >50% of pre-treatment values; NE, not evaluable.

b

Number of mice that died divided by number injected.

We report that the nonspecific toxicity of the recombinant immunotoxin anti-Tac(Fv)-PE38 has been reduced by introducing mutations in the Fv portion that lower the pI from 10.21 to 6.82. This was accomplished without altering residues in the complementarity-determining regions, and therefore did not affect the specific cytotoxic effect of the immunotoxin on target cells expressing CD25. As a consequence, tumor-bearing mice could be treated with higher doses and enhanced antitumor activity demonstrated.

Our strategy was to change framework residues so that Ag binding would not be affected. We selected residues that differed between the Fv of mAb RFB4 and that of anti-Tac because RFB4(Fv) has a pI of 7.67 was much less toxic to mice than anti-Tac(Fv)-PE38 (Table II). In this approach we only change one marginally basic residue VH H3; this change was not expected to alter the pI because histidine should be largely unprotonated at pH 7.2, which is the pH at the surface of the cells in the liver. The data in Table II show that most of the changes were from neutral to acidic residues. We plan to investigate the effect of changing basic residues to acidic or neutral residues in the next series of experiments. We had previously made a compilation of the frequency with which each amino acid is present in the framework regions of VH and VL (13). Many residues are highly conserved and were not changed because such mutations might affect Fv folding or stability. We have chosen only framework residues which are variable and which lie on the surface of the molecule so they would not affect protein folding or stability. To examine the effects of the change in pI on the electrostatic potential of the surface of anti-Tac(Fv), we employed a model of the Fv portion of the Ab. Fig. 3 shows the electrostatic potential mapped to the molecular surface of wild-type and M1-mutant models of anti-Tac(Fv), with red designating negative and blue designating positive. Note the large increase in the negative potential on both the front and back sides of the M1 mutant. The figure also shows regions of remaining positive charge, which are candidates for further mutagenesis.

FIGURE 3.

Electrostatic potential mapped to the molecular surface of wild-type and M1-mutant models of anti-Tac(Fv). The electrostatic potential ranges from −5.00 kT (red) to 5.00 kT (blue). Upper panel shows the surface region where most mutations are located. Note the large increase in negative potential in the M1 mutant. Lower panel shows the back view and the decrease in positive potential due to mutations that are largely on the opposite, front side of the protein. The ribbon diagrams in the middle indicate the orientation of the view. The Ag binding site is located at the top center in all views. The model of the wild type and mutant were produced using GRASP (19 ) and Ribbons (20 ) and were generated by homology modeling starting from the crystal structure of McPC603, a phosphocholine-binding mouse myeloma protein (21 ).

FIGURE 3.

Electrostatic potential mapped to the molecular surface of wild-type and M1-mutant models of anti-Tac(Fv). The electrostatic potential ranges from −5.00 kT (red) to 5.00 kT (blue). Upper panel shows the surface region where most mutations are located. Note the large increase in negative potential in the M1 mutant. Lower panel shows the back view and the decrease in positive potential due to mutations that are largely on the opposite, front side of the protein. The ribbon diagrams in the middle indicate the orientation of the view. The Ag binding site is located at the top center in all views. The model of the wild type and mutant were produced using GRASP (19 ) and Ribbons (20 ) and were generated by homology modeling starting from the crystal structure of McPC603, a phosphocholine-binding mouse myeloma protein (21 ).

Close modal

Several investigators have shown that when Abs or albumin are cationized by treatment with an amine, their pharmacokinetic properties are altered and their uptake by liver and other tissues is greatly enhanced (14, 15, 16). Our data are consistent with this finding. We assume that the positive charge in the Fv not only enhances liver uptake but also uptake by other organs. We plan to use immunohistochemistry and radiolabeled immunotoxins to examine this hypothesis. Alterations in animal toxicity could also be due to changes in half-life and stability, which will be addressed in future experiments.

The toxic side effects of immunotoxins in animals and humans are of two types. One side effect arises from the targeted killing of normal cells that have the same Ag as the tumor cells. The best solution to overcome this toxicity is to find a different target Ag that is not expressed on normal cells. The second type of toxicity arises from undefined nonspecific binding to normal cells. The liver is particularly vulnerable because it is susceptible to apoptosis induced by toxic substances, it has a high blood content, and its capillaries are fenestrated allowing immediate access of the high concentrations of immunotoxins that are in the blood just after injection. However, capillary damage could also occur leading to hemorrhagic liver necrosis and also contributing to the vascular leak syndrome that has been observed with immunotoxins containing ricin as well as those containing PE (17, 18). We have begun to decrease the pI of the Fvs of other recombinant immunotoxins to determine whether the decrease in animal toxicity observed with anti-Tac(Fv)-PE38 occurs with other molecules. Preliminary evidence indicates that this is the case.

In summary, we have been able to decrease the nonspecific toxicity of anti-Tac(Fv)-PE38 (LMB-2) in mice by about 3-fold without decreasing its specific cellular or antitumor activity. If this change also reduces toxicity in humans by 3-fold, we should be able to greatly increase the response to anti-Tac(Fv)-PE38 which already has shown good antitumor activity in patients with CD25+ leukemias and lymphomas (1).

We thank Dr. James Vincent for preparing molecular models and Dr. Irwin Arias for his suggestion that the liver endothelial cells might be the site of liver damage. We also thank J. Evans for expert editorial assistance.

2

Abbreviations used in this paper: PE, Pseudomonas exotoxin A; pI, isoelectric point; HSA, human serum albumin; H&E, hematoxylin and eosin; CR, complete regression.

1
Kreitman, R. J., I. Pastan.
1998
. Immunotoxins for targeted therapy.
Adv. Drug Delivery Rev.
31
:
53
2
Chaudhary, V. K., C. Queen, R. P. Junghans, T. A. Waldmann, D. J. FitzGerald, I. Pastan.
1989
. A recombinant immunotoxin consisting of two antibody variable domains fused to Pseudomonas exotoxin.
Nature
339
:
394
3
Siegall, C. B., V. K. Chaudhary, D. J. FitzGerald, I. Pastan.
1989
. Functional analysis of domains II, Ib and III of Pseudomonas exotoxin.
J. Biol. Chem.
264
:
14256
4
Brinkmann, U., I. Pastan.
1994
. Immunotoxins against cancer.
Biochim. Biophys. Acta
1198
:
27
5
Reiter, Y., I. Pastan.
1998
. Recombinant Fv immunotoxins and Fv fragments as novel agents for cancer therapy and diagnosis.
Trends Biotechnol.
16
:
513
6
Kreitman, R. J., I. Pastan.
1995
. Targeting Pseudomonas exotoxin to hematologic malignancies.
Semin. Cancer Biol.
6
:
297
7
Kreitman, R. J., Q. C. Wang, D. J. P. FitzGerald, I. Pastan.
1999
. Complete regression of human B-cell lymphoma xenografts in mice treated with recombinant anti-CD22 immunotoxin RFB4(dsFv)PE38 at doses tolerated by cynomolgus monkeys.
Int. J. Cancer
81
:
148
8
Kunkel, T. A., K. Bebenek, J. McClary.
1991
. Efficient site-directed mutagenesis using uracil-containing DNA.
Methods Enzymol.
204
:
125
9
Chowdhury, P. S., J. L. Viner, R. Beers, I. Pastan.
1998
. Isolation of a high affinity stable single chain Fv specific for mesothelin from DNA-immunized mice by phage display and construction of a recombinant immunotoxin with anti-tumor activity.
Proc. Natl. Acad. Sci. USA
95
:
669
10
Kreitman, R. J., C. B. Siegall, D. J. P. FitzGerald, J. Epstein, B. Barlogie, I. Pastan.
1992
. Interleukin 6 fused to a mutant form of Pseudomonas exotoxin kills malignant cells from patients with multiple myeloma.
Blood
80
:
2344
11
Kreitman, R. J., P. Bailon, V. K. Chaudhary, D. J. P. FitzGerald, I. Pastan.
1994
. Recombinant immunotoxins containing anti-Tac(Fv) and derivatives of Pseudomonas exotoxin produce complete regression in mice of an interleukin-2 receptor-expressing human carcinoma.
Blood
83
:
426
12
Kabat, E. A., T. T. Wu, H. M. Perry, K. S. Gottesman, and C. Foeller. 1991. Sequence of Proteins of Immunological Interest. U.S. Department of Health and Human Services, Washington, DC, Publ. no. 1-1-1137.
13
Chowdhury, P. S., G. Vasmatzis, B. Lee., I. Pastan.
1998
. Improved stability and yield of a Fv-toxin fusion protein by computer design and protein engineering of the Fv.
J. Mol. Biol.
281
:
917
14
Pardridge, W. M., J. Buciak, J. Yang, D. Wu.
1998
. Enhanced endocytosis in cultured human breast carcinoma cells and in vivo biodistribution in rats of a humanized monoclonal antibody after cationization of the protein.
J. Pharmacol. Exp. Ther.
286
:
548
15
Pardridge, W. M., Y. S. Kang, J. Yang, J. L. Buciak.
1995
. Enhanced cellular uptake and in vivo biodistribution of a monoclonal antibody following cationization.
J. Pharm. Sci.
84
:
943
16
Pardridge, W. M., Y. S. Kang, A. Diagne, J. A. Zack.
1996
. Cationized hyperimmune immunoglobulins: pharmacokinetics, toxicity evaluation and treatment of human immunodeficiency virus-infected human-peripheral blood lymphocytes-severe combined immune deficiency mice.
J. Pharmacol. Exp. Ther.
276
:
246
17
Soler-Rodriguez, A.-M., M.-A. Ghetie, N. Oppenheimer-Marks, J. W. Uhr, E. S. Vitetta.
1993
. Ricin α-chain and ricin α-chain immunotoxins rapidly damage human endothelial cells: implications for vascular leak syndrome.
Exp. Cell Res.
206
:
227
18
Kuan, C., L. H. Pai, I. Pastan.
1995
. Immunotoxins containing Pseudomonas exotoxin targeting LeY damage endothelial cells in an antibody specific mode: relevance to vascular leak syndrome.
Clin. Cancer Res.
1
:
1589
19
Nicholls, A., K. Sharp, B. Honig.
1991
. Protein folding and association: insights from the interfacial and thermodymanic properties of hycrocarbons.
Proteins Struct. Funct. Genet.
11
:
281
20
Carson, M..
1991
. Ribbons 2.0.
J. Appl. Crystallogr.
24
:
958
21
Rudikoff, S., Y. Satow, E. Padlan, D. Davies, M. Potter.
1981
. κ-Chain structure from a crystallized murine Fab′: role of joining segment in hapten binding.
Mol. Immunol.
18
:
705
22
Kreitman, R. J., Wilson, D. Robbins,I. Margulies, M. Stetler-Stevenson, T. Walmann, and I. Pastan. 1999. Responses in refactory hairy cell leukemia to a recombinant immunotoxin. Blood, In press.