| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Cancer Therapy: Preclinical |
Laboratory of Molecular Biology, Center for Cancer Research, National Cancer Institute, NIH, Bethesda, Maryland
Requests for reprints: Ira Pastan, Laboratory of Molecular Biology, National Cancer Institute, NIH, Building 37, Room 5106, 37 Convent Drive, MSC 4264, Bethesda, MD 20892-4264. Phone: 301-496-4797; Fax: 301-402-1344; E-mail: pastani{at}mail.nih.gov.
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Experimental Design: Arg490, which is located in the catalytic domain (III) of the immunotoxin HA22, was mutated to alanine. Purified immunotoxins were produced and tested for cytotoxic activity in cell culture and for antitumor activity and nonspecific toxicity in mice. ADP-ribosylation activity was also measured.
Results: HA22 (R490A) is 
2-fold more cytotoxic than HA22 on several CD22-positive cell lines. When injected i.v., HA22 (R490A) has more potent antitumor activity than HA22 against CA46 tumors in mice. HA22 and HA22 (R490A) have similar LD50s (1.3 mg/kg) and similar plasma half-lives. The R490A mutation also improved the cytotoxicity of the antimesothelin recombinant immunotoxin SS1 (dsFv)-PE38 (SS1P). In vitro ADP-ribosylation assays show that HA22 R490A has increased activity. Increased cytotoxic activity is probably related to this increase in ADP-ribosylation activity.
Conclusion: Protein engineering can be used to increase the efficacy of recombinant immunotoxins. Because HA22 (R490A) has increased antitumor activity without increased animal toxicity, immunotoxins with this mutation are candidates for clinical development.
Key Words: Cytotoxicity • Pharmacokinetics • ADP-ribosylation
|
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
CD22 is a lineage-restricted B-cell antigen that is expressed on B-chronic lymphocytic leukemia, hairy cell leukemia, acute lymphocytic leukemia, and Burkitt's lymphoma. The RFB4 antibody that specifically reacts with CD22 has been used to make a recombinant immunotoxin in which the Fv fragment stabilized by a disulfide bond is fused to a 38 kDa truncated form of Pseudomonas exotoxin A (PE; refs. 3–5). PE38 (a 38 kDa mutant form of PE A) contains the translocating and ADP-ribosylating domains of PE but not the cell-binding portion. In preclinical studies, RFB4 (disulfide-stabilized single-chain antibody fragment [dsFv])-PE38 (BL22) was shown to kill cell lines expressing CD22 as well as leukemic cells from patients with hairy cell leukemia and chronic lymphocytic leukemia. It also induced complete remissions in mice bearing lymphoma xenografts (6, 7). BL22 has been evaluated in a phase I clinical trial at the National Cancer Institute in patients with hematologic malignancies. In the initial report, 16 patients with purine analogue-resistant hairy cell leukemia were treated with BL22 and 11 (68%) achieved complete remission (1).
HA22 is an improved form of BL22 in which the Fv was mutated and antibody phage display used to isolate mutant phage that bound better to CD22. In HA22, residues SSY in heavy-chain CDR3 have been mutated to THW. HA22 has a 5-fold to 10-fold increase in cytotoxic activity on various CD22-positive cell lines and is up to 50 times more cytotoxic to cells from patients with chronic lymphocytic leukemia and hairy cell leukemia (8).
Based on the crystallographic structure of PE (9) and many functional studies, BL22 and other immunotoxins kill target cells by a complex series of steps that are initiated by binding to a cell surface molecule and followed by endocytosis, proteolytic processing, and translocation into the cytosol. In the cytosol, the ADP-ribosylation activity located within domain III of PE catalytically inactivates elongation factor 2, inhibiting protein synthesis and initiating programmed cell death.
Analysis of the three-dimensional structure of domain III of PE has shown that residues 486 to 493 are arranged in a flexible loop with a poorly ordered structure on the surface of domain III. The functional role of this flexible loop has not been defined (10). Brinkmann et al. (11) found that deletion of a trypsin recognition site (R490) in domain III of native PE made it resistant to proteolytic digestion and prolonged its survival in the circulation of mice.
In the present work, we have investigated the effects of mutating Arg490 to alanine on the properties of immunotoxin HA22. We investigated its cytotoxic activity, its half-life in the circulation, and its toxicity and antitumor activity in mice. We chose Arg490 because this residue constitutes a protease recognition site.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, and identified using blue-white screening procedures. Positive clones were sequenced using the primers C and D. The mutated insert was removed from pCR II by digesting the plasmid with SalI and EcoRI endonuclase and the fragment was ligated to identically digested VH-PE38 plasmid.
Expression and Purification of HA22 (R490A)
To produce immunotoxin, the two components, variable domain light chain (VL) and VH-PE38 (R490A), were expressed in E. coli BL21 (
DE3) where the proteins accumulated in inclusion bodies. Immunotoxins were then purified as previously described (5, 12) .
Preparation of SS1P (R490A). R490A was constructed by PCR-based site-directed mutagenesis. After SalI and EcoRI endonuclases digestion, the mutated insert was cloned into the identically digested SSVH-PE38 plasmid DNA. To produce SS1P (R490A) immunotoxin, SS1 VL and SSVH-PE38 (R490A) were expressed in E. coli BL21 (DE3) and purified to near homogeneity as previously described (5, 12).
Cell Lines. CD22-positive human Burkitt lymphoma cell lines (CA46, Daudi, and Raji) were obtained from American Type Culture Collection (Manassas, VA). CD-25-positive HUT-102 was obtained from T. Waldmann (NIH, Bethesda, MD) and human umbilical vascular endothelial cells (HUVEC) were purchased from Clonetics (San Diego, CA). A431-K5 cells were derived from the A431 cell line (human epidermoid carcinoma) by cotransfection of A431 cells with a plasmid encoding mesothelin (13). The ovarian cancer cell line A1847 was obtained from Dr. S. Aaronson (National Cancer Institute, Bethesda, MD).
Cytotoxicity Assay. The specific cytotoxicity of HA22 (R490A) was determined by protein synthesis inhibition assays (7, 8). The concentrations of immunotoxin that reduced [3H]leucine incorporation by 50% relative to untreated control culture were defined as the IC50. The cytotoxic effect of SS1P (R490A) was evaluated on two mesothelin-positive cancer cell lines, A431/K5 (an epidermoid carcinoma cell line transfected with full-length mesothelin cDNA) and A1847, using a protein synthesis inhibition assay previously described (11).
Cell Viability Assay. Inhibition of cell growth upon treatment with immunotoxin was determined in standard 2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium, monosodium salt (WST) assays based on the reduction of tetrazolium salt to formazan by the enzymes from viable cells (14). Viability was expressed as percentage of untreated controls. CA46, A431/K5, and A1847 cells were plated at 8,000 cells/well in a 96-well plate. HUVECs were seeded in 96-well plates at 3,000 cells/well. Plates were incubated for 40 hours or for 72 hours (HUVEC) at 37°C (15). Five microliters WST-8 solution was added to each well of the plate and incubated for 4 hours at 37°C. To correct for background activity, cells were cultured in the presence of cycloheximide at 10 µg/mL.
ADP-Ribosylation Activity Assay
ADP-ribosylation activity of HA22 and HA22 (R490A) was determined by measuring transfer of ADP-ribose from [14C]NAD to EF-2 (16).
Nonspecific Animal Toxicity Assay. On day 0, female Swiss mice (5-6 weeks, 18-22 g) were given a single injection into the tail vein of various amounts of immunotoxin in 0.2 mL PBS containing 0.2% human serum albumin. Animal mortality was observed over 2 weeks.
Pharmacokinetics
NIH Swiss mice were injected into the tail vein with 10 µg HA22 or HA22 (R490A). Blood samples were drawn at different times (2, 5, 10, 20, 30, 40, 50, 60, and 90 min) and the concentration of immunotoxin determined by ELISA. A standard curve was made with each pure immunotoxin. Briefly, microtiter plates were coated with 50 µL CD22-Fc protein (5 µg/mL; ref. 8) in PBS at 4°C overnight. The plates were blocked with PBS containing 3% bovine serum albumin at room temperature for 2 hours, followed by washing five times in PBS containing 0.05% Tween 20. Standards and samples were diluted 1/100, 1/500, and 1/1,000 in PBS with 1% normal mouse serum. One hundred microliters diluted standards or samples were applied, followed by incubation with 50 µL of 1:250 dilution of horseradish peroxidase–conjugated anti-PE antibody for 3 hours at room temperature. After washing, plates were developed using 3,3',5,5'-tetramethylbenzidine for 10 minutes and the absorbance read at 450 and 650 nm. The assays were done in triplicate.
Antitumor Activity and Statistics
The antitumor activity of the immunotoxins was determined in severe combined immunodeficiency (SCID) mice bearing CA46 cells (6). Cells (1 x 107) were injected s.c. into SCID mice (5 weeks, body weight 18 g) on day 0. Tumors 100 mm3 developed in animals by day 6 after tumor implantation when treatment was initiated. All data are presented as mean ± SD. Differences in data from animal experiments were analyzed for significance with Wilcoxon scores (rank sum) test. All mentioned changes in distribution qualified for the probability level P < 0.05.
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
2-fold more active on Daudi and on CA46 cells than HA22 and is 3-fold more active on Raji cells. BL22 is much less active than either immunotoxin. The immunotoxins were also tested on the CD22-negative L540 or HUT-102 cell lines and found to be >1,000-fold or 10,000-fold less toxic, demonstrating that these immunotoxins are specific for CD22-expressing cells.
|
|
To further assess specificity, we examined whether HA22 (R490A) could result in induction of endothelial cell death. We examined endothelial cells because they do not express CD22 but may have a role in the toxic side effects of some immunotoxins (15). The endothelial cell line HUVEC was treated with either HA22 (R490A), HA22, BL22, HB21 (Fv)-PE40, or LMB-7. HB21 (Fv)-PE40 targets the transferrin receptor that is widely expressed on many cell types and was expected to be cytotoxic to HUVEC. LMB7 targets the LeY antigen previously shown to be expressed on HUVEC (16). HA22 (R490A), HA22, or BL22 did not decrease the viability of HUVECs (Fig. 2D). However, as expected, both HB21 (Fv)-PE40 and LMB7 were cytotoxic to the cells (16). These results further confirm the specificity of HA22 (R490A).
Mouse Studies. Because HA22 (R490A) was more cytotoxic to the CD22-positive cell lines than HA22, we compared its antitumor activity with HA22. Before doing this, we determined its LD50. Groups of five or more mice received a single i.v. injection of various doses of HA22 (R490A), HA22, or BL22 and were observed for 2 weeks. HA22 and HA22 (R490A) have very similar animal toxicities with LD50s of 1.3 mg/kg (Table 2). BL22 was slightly more toxic with all mice dying at a dose of 1.25 mg/kg. The calculated (interpolated) LD10 and LD50 are 1.01 and 1.12 for BL22, 1.06 and 1.33 for HA22, and 1.08 and 1.38 for HA22 (R490A). Almost all of the deaths occurred within 72 hours after treatment. These data show that the R490A mutation has very little effect on mouse toxicity.
|
Antitumor Activity. To determine if the improved in vitro cytotoxic activity was translated into increased antitumor activity, HA22 (R490A) and HA22 were compared using tumor xenografts of CA46 cells growing in SCID mice. CA46 cells (1 x 107) were implanted into the flanks of SCID mice on day 0. On day 6, when the tumors reached 100 mm3 in size, the animals were injected i.v. with either 300 µg/kg (n = 10) or 150 µg/kg (n = 8) HA22 or HA22 (R490A) every other day x3. As shown in Fig. 3, treatment with HA22 (R490A) or HA22 decreased tumor size compared with controls. By day 10, tumors in mice receiving 300 µg/kg of HA22 (R490A) fell to 85 ± 28.5 mm3 in size, whereas in mice treated with 300 µg/kg of HA22 the tumors were 126 ± 31.2 mm3 in size. Treatment with 150 µg/kg HA22 (R490A) resulted in tumors averaging 302 ± 143.6 mm3 on day 18, whereas tumors treated with HA22 were significantly larger, averaging 592 ± 189 mm3 (Fig. 3A). Antitumor activity was also dependent on dose; 150 µg/kg was less effective than 300 µg/kg for both immunotoxins. Without treatment, CA46 tumors grew rapidly. A significant difference in tumor size (Wilcoxon test, P < 0.001) was found between mice that received HA22 and mice that received HA22 (R490A) at 150 µg/kg on treatment day 12 (P = 0.0007), on day 14 (P = 0.0008), and on day 16 (P = 0.0008); and with 300 µg/kg treatment on day 14 (P = 0.0009; Fig. 3C). To show that the inhibition of tumor growth by HA22 and HA22 (R490A) is specific, we also treated mice with CA46 tumors with SS1P (R490A) that targets the mesothelin antigen. SS1P (R490A) did not inhibit the growth of CA46 tumors (data not shown).
|
Immmunotoxin SS1P. Mesothelin is an antigen that is highly expressed on pancreatic and ovarian cancers and mesotheliomas. SS1P is an immunotoxin that binds to mesothelin and kills mesothelin-expressing cells (13). To determine if the R490A mutation would also increase the cytotoxic activity of an immunotoxin target in an epithelial cancer, we introduced the R490A mutation into SS1P to produce SS1P (R490A). The data in Table 3 shows that SS1P (490A) was significantly more active than SS1P on two mesothelin-expressing cell lines with an 2-fold increase in activity.
|
2-fold increase in ADP-ribosylation activity with the R490A mutation. |
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
We found that mutation of Arg490 to alanine in the toxin portion of two different recombinant immunotoxins, HA22 and SS1P, resulted in a 2-fold to 3-fold increase in cytotoxic activity and with HA22 a corresponding increase in antitumor activity. This enhanced antitumor activity occurs without any change in nonspecific toxicity in normal mice or in the tumor-bearing animals. The mutation is in the catalytic domain (domain III) of the toxin. Arg490 is part of a flexible loop on the surface of domain III of PE and this loop was shown to be more susceptible to proteolytic digestion than other regions of the molecule (11). ADP-ribosylation assays show that the R490A mutant has an increase in ADP-ribosylation activity probably accounting for the increased cytotoxic activity of immunotoxins containing this mutation. Several amino acids have been shown to be important for the activity of domain III of PE. These include H426, H440, W466, Y470, Y481, and E553. Because R490 is not in the active site cleft, it is unlikely to interfere sterically in NAD binding or to be directly involved in the chemical catalytic process. However, it could affect binding of EF2 to domain III. Residue 490 was chosen for mutation to alanine for two reasons. One is that alanine destroys a recognition site for trypsin and related proteases. The second is that it does not alter the main chain conformation and does not alter protein structure by imposing extreme electrostatic or steric effects. This strategy seemed to be successful because we were able to use a protein renaturation method to produce HA22 (R490A) with the same yield as the parental molecular HA22 immunotoxin.
Our initial biological characterization of HA22 (R490A) consisted of evaluating its cytotoxic activity on three CD22-positive cell lines. We found an increase in cytotoxic activity measured by a protein synthesis inhibition assay, ADP-ribosylation activity assay, and a cell viability assay. This increase in cytotoxic activity was not dependent on the nature of the target antigen because the activity of an immunotoxin (SS1P) containing an Fv targeting the mesothelin antigen was also increased by the R490A mutation (Table 3). After internalization, there are a series of steps required for a fragment of PE38 to react to the cytosol, inactivate EF2, and cause cell death. These include proteolytic processing, reduction of a disulfide bond, transport to the endoplasmic reticulum, and inactivation of elongation factor 2. Any of these steps could be affected by the R490A mutation.
Several factors control the lifetimes of immunotoxins in the circulation, including binding to tissues, filtration in the kidney, and inactivation or degradation of proteins by proteolysis. Previous studies had shown that elimination of a protease recognition site in domain III of native PE prolonged its half-life in the circulation of mice, and this was attributed to increased protease resistance resulting in decreased degradation in the circulation. We have now extended these studies to recombinant immunotoxins in which the binding domain of native PE is replaced by the Fv portion of an antibody. The antibody portion is human specific so that in mice there are no specific binding sites, and degradation of the immunotoxin is entirely by nonspecific pathways. In contrast to the results with native PE in which removal of a protease site resulted in an increase in half-life in the circulation, there was a very small and not significant decrease in the half-life of HA22 (R490A) compared with HA22 (R490).
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 antigen as the tumor cells. The best solution to overcome this toxicity is to find a different target antigen 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. We hypothesized that the protease-sensitive site near Arg490 in domain III could contribute to the toxicity of HA22 in mice because plasmin and thrombin are associated with the surface of liver cells and endothelial cells (18, 19) and, therefore, HA22 might bind to protease recognition sites on those cells. However, there was very little change in the nonspecific toxicity of immunotoxins with the R490A mutation in mice. One key parameter in considering the clinical potential for HA22 (R490A) is its therapeutic window or the ratio of the maximum tolerated dose to the therapeutic dose. Initial toxicity evaluation in mice showed that HA22 (R490A) and HA22 were well tolerated, with no signs of toxicity at doses up to 1 mg/kg. At 1.75 mg/kg, both HA22 (R490A) and HA22 were toxic.
We evaluated the antitumor activity of HA22 and HA22 (R490A) by assessing their ability to decrease the size of CA46 tumors in SCID mice. We chose this model because the s.c. tumors are structurally similar to human soft tissue lymphomas and because this model produces quantitative, rapid, and reproducible results in terms of tumor reduction rather than depending on development of terminal disease for measuring antitumor activity in a disseminated model (20). Figure 3 shows that treatment with HA22 (R490A) significantly retarded the growth of CA46 tumors in SCID mice compared with HA22 treatment (Wilcoxon test, P < 0.001). This is consistent with the results obtained in vitro on the CD22-positive cell lines.
In conclusion, we have improved the cytotoxicity of HA22 on the CD22-positive cell lines by 2-fold without decreasing its nonspecific toxicity or antitumor activity in mice. These studies show that it is possible to use protein engineering to increase the efficacy of recombinant immunotoxins.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received 9/21/04; revised 10/29/04; accepted 11/ 1/04.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. Matsushita, I. Margulies, M. Onda, S. Nagata, M. Stetler-Stevenson, and R. J. Kreitman Soluble CD22 as a tumor marker for hairy cell leukemia Blood, September 15, 2008; 112(6): 2272 - 2277. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Onda, S. Nagata, D. J. FitzGerald, R. Beers, R. J. Fisher, J. J. Vincent, B. Lee, M. Nakamura, J. Hwang, R. J. Kreitman, et al. Characterization of the B Cell Epitopes Associated with a Truncated Form of Pseudomonas Exotoxin (PE38) Used to Make Immunotoxins for the Treatment of Cancer Patients J. Immunol., December 15, 2006; 177(12): 8822 - 8834. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. Li, S. P. Mahesh, D. F. Shen, B. Liu, W. O. Siu, F. S. Hwang, Q.-C. Wang, C.-C. Chan, I. Pastan, and R. B. Nussenblatt Eradication of Tumor Colonization and Invasion by a B Cell-Specific Immunotoxin in a Murine Model for Human Primary Intraocular Lymphoma Cancer Res., November 1, 2006; 66(21): 10586 - 10593. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Zhang, L. Xiang, R. Hassan, C. H. Paik, J. A. Carrasquillo, B.-s. Jang, N. Le, M. Ho, and I. Pastan Synergistic Antitumor Activity of Taxol and Immunotoxin SS1P in Tumor-Bearing Mice Clin. Cancer Res., August 1, 2006; 12(15): 4695 - 4701. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
, BL22; 
, HA22; 
, HA22 (R490A). D, viability of HA22 (R490A) on HUVEC. HUVEC 3,000 cells/well incubated with various concentrations of immunotoxins for 72 hours. Cell viability was determined by a WST assay as described in Materials and Methods. Results are given as %viability of incubation without immunotoxin; points, mean of triplicate values; bars, SD. 
, HB21Fv-PE40; 
, LMB-7. E, comparison of ADP-ribosylation activity between HA22 and HA22 (R490A). ADP-ribosylation assays were done as described 