Efficient killing of CD22+ tumor cells by a humanized diabody–RNase fusion protein

BBRC Biochemical and Biophysical Research Communications 331 (2005) 595–602 www.elsevier.com/locate/ybbrc

Efficient killing of CD22+ tumor cells by a humanized diabody–RNase fusion protein Ju¨rgen Krauss a,*,1, Michaela A.E. Arndt b,1, Bang K. Vu c, Dianne L. Newton c, Siegfried Seeber a, Susanna M. Rybak d a

d

Department of Medical Oncology and Cancer Research, University of Essen, D-45122 Essen, Germany b Institute of Immunology, University of Essen, D-45122 Essen, Germany c SAIC, National Cancer Institute at Frederick, Frederick, MD 21702, USA Developmental Therapeutics Program, National Cancer Institute at Frederick, Frederick, MD 21702, USA Received 29 March 2005 Available online 9 April 2005

Abstract We report on the generation of a dimeric immunoenzyme capable of simultaneously delivering two ribonuclease (RNase) effector domains on one molecule to CD22+ tumor cells. As targeting moiety a diabody derived from the previously humanized scFv SGIII with grafted specificity of the murine anti-CD22 mAb RFB4 was constructed. Further engineering the interface of this construct (VL36Leu fi Tyr) resulted in a highly robust bivalent molecule that retained the same high affinity as the murine mAb RFB4 (KD = 0.2 nM). A dimeric immunoenzyme comprising this diabody and Rana pipiens liver ribonuclease I (rapLRI) was generated, expressed as soluble protein in bacteria, and purified to homogeneity. The dimeric fusion protein killed several CD22+ tumor cell lines with high efficacy (IC50 = 3–20 nM) and exhibited 9- to 48-fold stronger cytotoxicity than a monovalent rapLRI–scFv counterpart. Our results demonstrate that engineering of dimeric antibody–ribonuclease fusion proteins can markedly enhance their biological efficacy.
2005 Elsevier Inc. All rights reserved. Keywords: CD22; Ribonuclease; Fusion protein; Cytotoxicity; Humanized diabody

Monoclonal antibodies (mAb) have substantially expanded the therapeutic options for cancer patients within the last decade (reviewed in [1]). For patients with relapsed or refractory hematological malignancies, immunotoxins that target the CD22 antigen which is expressed in the majority of cases of B-cell non-HodgkinÕs lymphoma belong to the most potent reagents thus far evaluated in clinical trials (reviewed in [2]). A recently published study reported a >65% complete response rate including long-lasting remissions in patients with refractory hairy cell leukemia after treat*

1

Corresponding author. Fax: +49 201 723 5925. E-mail address: [email protected] (J. Krauss). These authors contributed equally to this work.

0006-291X/$ - see front matter
2005 Elsevier Inc. All rights reserved. doi:10.1016/j.bbrc.2005.03.215

ment with a small, recombinant immunotoxin comprising a disulfide stabilized Fv generated from the antiCD22 mAb RFB4 and a Pseudomonas exotoxin A domain derivative [3]. However, immunogenicity and severe systemic toxicity of this and other anti-CD22 immunotoxins [3–7] are major obstacles for the wide clinical application of such reagents in humans. To overcome these problems, the development of similar potent but less immunogenic and toxic reagents is of great importance. Ribonucleases can provide a valuable alternative to the use of plant and bacterial toxins as antineoplastic effector molecules [8]. Among several ribonucleases currently being evaluated as therapeutic agents in a preclinical setting, Ranpirnase (Onconase), a member of the

596

J. Krauss et al. / Biochemical and Biophysical Research Communications 331 (2005) 595–602

pancreatic RNase A superfamily, has progressed to clinical trials. A multicenter phase II study demonstrated clinical activity in approximately 50% of patients with unresectable malignant mesothelioma including the achievement of partial remissions that in some cases lasted for several years [9]. Despite being administered on a weekly basis Ranpirnase was well tolerated by the majority of patients. A phase III clinical trial comparing the efficacy of Doxorubicin versus Doxorubicin plus Ranpirnase in mesothelioma patients is currently ongoing. After conjugating Ranpirnase to an internalizing anti-CD22 antibody (mAb LL2), the selective toxicity of the ribonuclease towards CD22+ tumor cells was increased 10,000-fold [10]. Administration of LL2-Ranpirnase to tumor xenotransplanted SCID mice resulted in tumor regressions comparable to that reported for a bacterial Pseudomonas exotoxin derivative (PE38KDEL) conjugated to the same mAb [11], yet mediated several orders of magnitude lower systemic toxicity than PE38KDEL [10]. Similar results were obtained when conjugating a recombinant homolog of Ranpirnase, Rana pipiens liver RNaseI (rapLRI) [12], to either antiCD22 mAb LL2 [13] or anti-CD22 mAb RFB4 (SMR and DLN, unpublished results). Disadvantages for the clinical application of these chemical mAb–RNase conjugates, however, include non-selective coupling of the cytotoxic moiety that may impair antigen binding [14], poor capacity to penetrate bulky tumor tissue owing to their large size [15], and immunogenicity mediated by murine antibody determinants [16]. To address the large size and immunogenicity of the murine anti-CD22 mAb RFB4 and to make this antibody feasible for the genetic fusion with ribonucleases, we recently generated a panel of humanized scFv fragments that retained the specificity of the murine mAb RFB4, bound to the target antigen with high affinity, and exhibited exceptional biophysical stability [17]. In the present study, we produced and characterized monomeric and dimeric fusion proteins consisting of derivatives of these antibody fragments and rapLRI. Data on soluble bacterial expression, enzymatic activity, and cytotoxicity of the recombinant immunoenzymes are presented.

Materials and methods Generation of scFv and diabody genes. The gene coding for the humanized anti-CD22 scFv SGIII [17] was used to generate diabody SGIII by shortening the (G4S)3 encoding linker peptide to 5 residues (GGGGS) as described previously [18]. The VL36Leu fi Tyr mutation was introduced into both SGIII scFv and SGIII diabody by oligonucleotide directed mutagenesis and overlap extension PCR techniques as described elsewhere [19], resulting in scFv SGIIIM and diabody SGIIIM, respectively. Fragments were cloned into bacterial expression vector pHOG21 [20] and correct inserts were verified by DNA sequencing.

Periplasmic expression and purification of scFv fragments. Monovalent and bivalent scFv fragments were expressed as soluble protein in the periplasm of Escherichia coli strain TG1 (Stratagene, La Jolla, CA) and purified by immobilized metal chelate affinity chromatography (IMAC) and size-exclusion chromatography as described previously [17]. Concentrations of purified antibody fragments were spectrophotometrically determined from the absorbance at A280nm using the extinction coefficient e1mg/ml = 1.7 as calculated by the Gene Inspector software (Textco BioSoftware, West Lebanon, NH). Generation, expression, and purification of scFv-ribonuclease fusion proteins. RapLRI and synthetic (G4S)3 spacer encoding gene segments were cloned with flanking NcoI and PvuII restriction sites into bacterial expression vector pHOG21 [20], resulting in plasmid pBJR-1B. RapLRI–scFv 1.110 and rapLRI–diabody 1.111 were obtained by ligating the PvuII/BamHI fragments encoding the monovalent scFv SGIIIM or bivalent Db SGIIIM, respectively, into pBJR-1B. The correct size and sequence of cloned inserts was analyzed by restriction digest and DNA sequencing, respectively. Soluble chimeric fusion proteins were expressed in the periplasm of E. coli TG1 as described [17] except that bacteria were induced for soluble protein expression at 14 C. Periplasmic extracts were thoroughly dialyzed against SP buffer (300 mM NaCl, 50 mM NaH2PO4, pH 8.0) and fusion proteins were purified by immobilized metal affinity chromatography (IMAC) using Ni–NTA columns (Qiagen, Valencia, CA) equilibrated with SP buffer. Absorbed material was eluted in 0.5 ml fractions with increasing imidazole concentrations (10–250 mM) in SP buffer. Final purification and separation of monomeric or dimeric fusion proteins from higher molecular forms was achieved by size-exclusion chromatography in SP-50 buffer (300 mM NaCl, 50 mM NaH2PO4, and and 50 mM imidazole, pH 8.0) using two series-connected Superdex 75 HR 10/30 or Superdex 200 HR 10/30 columns (Amersham-Pharmacia Biotech), respectively. Immunoenzymes were analyzed on 4–20% SDS–PAGE under reducing conditions and stained with Simply Blue Safe Stain (Invitrogen, Carlsbad, CA), or by Western blot, using an anti-(c-myc)-peroxidase-conjugated monoclonal antibody (Roche, Indianapolis, IN) followed by ECL Plus (Amersham-Pharmacia Biotech) detection. Binding assays. Specific binding of the constructs was determined by flow cytometry using the human CD22+ B-cell lines Raji and Daudi (ATTC, Manassas, VA, USA). The human T-cell line Jurkat (ATTC) or breast carcinoma cell line MCF7 (ATTC) were used as negative controls. Staining was performed as previously described [17]. Stained cells were analyzed on a FACScan Flow Cytometer (BD Bioscience, San Jose, CA), and the median fluorescence intensity (MFI) was calculated using the CellQuest software (BD Bioscience). Determination of binding affinity constants. Binding affinity constants (KD) of mono- and bivalent scFv fragments, and the mAb RFB4 (Biopharmaceutical Development Program, SAIC-Frederick, Frederick, MD) were determined as previously described [17]. Serum stability of mono- and bivalent scFv fragments. Monovalent scFv fragments or diabodies were incubated at 37 C in 90% human serum at a concentration of 12 lg/ml for up to 8 days. Samples were removed at different time points and activity for binding to CD22+ Raji cells was determined by flow cytometry. The MFI was determined as described above. tRNA assay. Ribonucleolytic activity of rapLRI or rapLRI–scFv fusion proteins was determined by monitoring the formation of perchloric acid-soluble nucleotides from tRNA as previously described [21,22]. Briefly, 10 ll aliquots of various concentrations of the RNase or scFv–RNase fusion proteins diluted in 0.5 mg/ml human serum albumin (HSA) were added to a 300 ll reaction mixture containing 133 mM Mes, pH 6.5, 0.67 mg/ml HSA, and 0.66 mg/ml yeast tRNA in RNase-free polypropylene microcentrifuge tubes. The mixtures were incubated for 2 h at 37 C. Samples were placed on ice and the reaction was stopped by adding 700 ll of 3.4% perchloric acid. After further incubation on ice for 10 min, the reaction mixtures were centrifuged (12,000 rpm, 10 min, 4 C) and absorbance of the supernatants was

J. Krauss et al. / Biochemical and Biophysical Research Communications 331 (2005) 595–602 read at 260 nm. Absorbance of blanks containing tRNA and buffer was subtracted from the absorbance of sample readings. Cytotoxicity assays. Selective cytotoxicity of scFv–ribonuclease fusion proteins towards targeted tumor cells was determined by a protein synthesis inhibition assay as described previously [23]. Briefly, 1 · 104 cells per well were plated into microtiter plates one day prior to treatment. Various concentrations of scFv–rapLRI fusion proteins, rapLRI alone or buffer only were added to target cells in triplicate. Cells were incubated for 72 h at 37 C, 5% CO2 before adding 10 lCi/ well [14C]leucine (Perkin-Elmer NEN, Boston, MA). Incubation was continued for 4 h at 37 C. Cells were harvested onto glass fiber filters using a PHD cell harvester (Cambridge Technology, Cambridge, MA) and incorporated radioactivity was counted with a liquid scintillation counter (Beckman–Coulter, Fullerton, CA). In addition, a cell viability assay was performed by pulsing cells with 10 ll/well water soluble tetrazolium-1 (WST-1) salt (Roche) instead of [14C]leucine. Activity of cellular mitochondrial dehydrogenases in metabolically active cells was measured after 2–4 h by reading the absorbance of the formed formazan dye at 450 nm, using a microplate reader (Bio-Tek Instruments, Winooski, VT). The concentration of recombinant fusion proteins required to inhibit protein synthesis or cell viability by 50% relative to buffer treated control cells was defined as the IC50.

597

Fig. 1. Size-exclusion chromatography. Superimposed are elution profiles of wild-type diabody SGIII (dotted line) and interface engineered diabody SGIIIM (solid line) using a calibrated Superdex 75 column. Arrows indicate elution peaks of calibration reference proteins (aldolase, Mr 158 kDa; apo-transferrin, Mr 81 kDa; bovine serum albumin, Mr 67 kDa; and chymotrypsinogen A, Mr 25 kDa).

Results Generation and characterization of a humanized antiCD22 diabody In order to generate a dimeric immunoenzyme with enhanced cytotoxicity, we first constructed a humanized diabody by shortening the linker peptide of the humanized anti-CD22 scFv SGIII [17] to 5 residues (G4S). This construct, Db SGIII, was produced in bacteria at very low yield (Table 1) and exhibited a strong tendency to aggregate into higher molecular weight species (Fig. 1). These properties indicated that the soluble bacterial expression and purification of a dimeric fusion protein derived from Db SGIII would be impossible. To improve the solubility and possibly interdomain stability of the antibody, interface residue VL36Leu was replaced by tyrosine, the most common human germline amino acid at this location. This resulted in 4.9- and 4.2-fold increase in soluble expression levels of the monovalent scFv SGIIIM, and bivalent diabody SGIIIM, respectively

(Table 1). Size-exclusion chromatography revealed that Db SGIIIM exclusively formed dimers with an apparent molecular size of 54 kDa (Fig. 1). When compared with the wild-type scFv, the introduction of VL36Tyr into the human Vj framework impaired the binding affinity of the monovalent scFv 4.8-fold (KD = 47.5 nM, Table 1). In contrast, the diabody exhibited tight antigen binding (KD = 0.21 nM, Table 1), similar to that of the highly purified mAb RFB4 (KD = 0.26 nM) employed in clinical trials. To assess the stability of mutant scFv SGIIIM and diabody Db SGIIIM, respectively, purified proteins were incubated in human serum for up to 8 days at 37 C and binding activity to CD22+ Raji cells was monitored by flow cytometry. As shown in Fig. 2, scFv SGIIIM lost

Table 1 Yield and affinity of monovalent and bivalent anti-CD22 scFv fragments Clone

SGIII SGIIIM/VL36Leu fi Tyr

scFv

Diabody (Db)

Yield (lg/L)a

KD (nM)b

Yield (lg/L)a

KD (nM)b

42 207

9.8 47.5

20 85

n.d. 0.2

n.d., not determined. a Yield refers to monovalent scFv or bivalent diabody after purification by immobilized metal chelate chromatography and sizeexclusion chromatography. b Binding affinity constants (KD) were calculated from equilibrium binding curves determined by flow cytometry.

Fig. 2. Serum stability of mutant scFv and diabody. Immunoreactivity with CD22+ Raji cells of VL36Leu fi Tyr mutated scFv SGIIIM and diabody SGIIIM after incubation in human serum at 37 C over a period of 8 days was determined by flow cytometry. Binding activity is expressed as percentage of initial median fluorescence intensity (MFImax).

598

J. Krauss et al. / Biochemical and Biophysical Research Communications 331 (2005) 595–602

Fig. 3. Purification of monomeric and dimeric RNase–scFv fusion proteins. (A) Schematic representation of expressed monomeric rapLRI–scFv 1.110 and dimeric rapLRI–diabody 1.111 fusion proteins. C-myc epitope tag and hexahistidine tag are indicated as tags; VH, variable heavy chain; VL, variable light chain; and rapLRI, Rana pipiens liver ribonuclease I. (B) Reducing 4–12% SDS gel of monomeric rapLRI–scFv 1.110 and dimeric rapLRI–diabody 1.111 stained with Simply Blue Safe Stain. Sizes of molecular weight markers (MW) are indicated.

about 20% of its initial binding activity after 8 days of incubation, whereas Db SGIIIM retained full binding activity at this time point. Purification of rapLRI–scFv and rapLRI–diabody fusion proteins Fusion proteins rapLRI–scFv 1.110 or rapLRI–diabody 1.111 (Fig. 3A) were expressed in E. coli, isolated as soluble proteins from the bacterial periplasm after overnight induction at low temperature (14– 16 C), and purified by Ni–NTA columns and two consecutive size-exclusion chromatography steps, respectively. This resulted in homogeneous recombinant fusion protein preparations with >95% purity (Fig. 3B). Specific binding Specific binding of homogeneously purified rapLRI–scFv 1.110 and rapLRI–diabody 1.111 fusion proteins was analyzed by flow cytometry. Both fusion proteins exhibited binding to CD22+ and no binding to CD22
tumor cells (data not shown). Enzymatic activity of fusion proteins The ribonucleolytic activity of rapLRI in the fusion protein configuration was examined by a standard tRNA degradation assay and compared with that of rapLRI alone. As shown in Fig. 4 both monomeric and dimeric fusion proteins exhibited catalytic activity for degrading tRNA substrate.

Fig. 4. Ribonucleolytic activity of monomeric and dimeric fusion proteins. Dose-dependent cleavage of tRNA substrate was measured spectrophotometrically (A260 nm) after incubation of a tRNA containing reaction mixture with various concentrations of rapLRI, rapLRI– scFv 1.110, or rapLRI–diabody 1.111 as described under Materials and methods. Bars representing the standard deviation can be smaller than symbols.

Cytotoxicity of recombinant fusion proteins towards tumor cells The capability of monomeric and dimeric fusion proteins for killing targeted tumor cells was measured by a protein synthesis inhibition and cell death assay, respectively. Both fusion proteins mediated a dose-dependent selective protein synthesis inhibition, resulting in cell death of several CD22+ tumor cell lines (Figs. 5 and 6). However, selective cytotoxicity of the dimeric fusion protein 1.111 exceeded that of the monomeric counterpart between 9- and 48-fold (Fig. 6, Table 2). Cytotoxicity of both fusion proteins was selective, because neither

J. Krauss et al. / Biochemical and Biophysical Research Communications 331 (2005) 595–602

599

the monomeric nor the dimeric fusion protein mediated cytotoxic effects towards CD22
Jurkat cells (Fig. 6).

Discussion

Fig. 5. Protein synthesis inhibition mediated by monomeric and dimeric fusion proteins. Protein synthesis of CD22+ Daudi cells was measured by incorporation of [14C]leucine into cellular proteins after 72 h incubation with indicated concentrations of fusion proteins rapLRI–scFv 1.110, or rapLRI–diabody 1.111, scFv SGIIIM, or diabody SGIIIM, or rapLRI alone, and is expressed as percentage of protein synthesis activity of control cells treated with buffer only. Experiments were done in triplicate, bars represent the standard deviation.

Fig. 6. Cell death assay. Cell viability of CD22+ and CD22
tumor cells was determined after 72 h treatment with indicated concentrations of monomeric (black symbols) or dimeric (white symbols) fusion proteins. Cells were pulsed with cell proliferation reagent WST-1 and absorbance of the formed formazan dye by living cells was read at 450 nm. Results are expressed as percentage relative to buffer treated control cells.

Table 2 Cytotoxicity of monomeric and dimeric fusion proteins towards human tumor cell lines Cell line

Daudi Raji CA46 Jurkat

IC50 (nM)a rapLRI–scFv

rapLRI–diabody

132 145 185 >500

9 3 20 >500

a Determined by WST-1 cell viability assay as described under Materials and methods.

To preserve the antineoplastic potency and low in vivo toxicity of chemical anti-CD22 mAb–ribonuclease immunoconjugates but to overcome properties that may complicate the clinical application of these reagents (i.e., immunogenicity, non-controllable crosslinkage, large molecular size), we have developed a small-sized recombinant derivative comprising a humanized antiCD22 diabody with specificity of the mAb RFB4 and the ribonuclease rapLRI. We first constructed a humanized anti-CD22 diabody by shortening the linker peptide of the previously generated humanized anti-CD22 scFv SGIII [17] to 5 residues. This construct yielded insufficient soluble protein expression, however, and exhibited a strong tendency for aggregation to higher molecular weight species. Mutagenesis experiments have previously shown that residues contributing to the VH–VL interface of antibodies can impact solubility, stability, and antigen binding properties of scFv molecules [24– 26]. When analyzing the protein sequence of the humanized scFv antibody SGIII for unusual residues located within the VH–VL interface, we found VL36Leu to be uncommon in human sequences [27]. Residue VL36 is part of a set of 10 residues forming an interconnected hydrophobic region (‘‘central zone’’) of the interface and directly interacts with VH residues at a frequency of 87% in 23 analyzed antibody crystal structures [28]. From these data we reasoned that mutagenesis of VL36Leu to the most common human germline residue at this location (tyrosine) may have an impact on adjusting the interface between the VH and VL domain, and hence solubility and aggregation behavior of the antibody. Favorable interface anchoring was predicted to be particularly beneficial for providing increased stability to the two non-covalently associated polypeptide chains of the diabody formatted molecule. Indeed, introduction of VL36Tyr not only markedly improved the solubility of both mono- and bivalent scFv mutants, but also completely prevented aggregation of the bivalent molecule into higher molecular weight species and resulted in enhanced stability. These results demonstrate that engineering the VH–VL interface of a diabody may enhance both bacterial solubility and stability. To generate a dimeric fusion protein comprising two targeting and two effector moieties, the engineered diabody was fused to rapLRI. RapLRI offers a key advantage over all known cytotoxic amphibian ribonucleases, because it retains full enzymatic activity without posttranslational modification of the N-terminal part of the enzyme [13] and therefore can be used as a catalytically active fusion protein effector moiety [23]. Because

600

J. Krauss et al. / Biochemical and Biophysical Research Communications 331 (2005) 595–602

previous attempts to functionally refold dimerizing scFv-seminal bovine RNase fusion proteins from insoluble bacterial inclusion bodies were reportedly unsuccessful [29], we expressed the dimeric fusion proteins soluble in E. coli and purified them from periplasmic extracts. Our data show that two rapLRI-(G4S)3-VH– G4S-VL polypeptide chains were correctly folded and assembled to catalytically active dimeric immunoenzymes in the bacterial periplasm. In comparison with the monovalent counterpart, the rapLRI–diabody fusion protein 1.111 clearly exhibited superior cytotoxicity towards targeted tumor cells. Several factors may have contributed to the enhanced potency of the dimer. It was recently demonstrated that upon binding of an anti-CD22 mAb, the CD22 receptor is re-expressed with slow kinetics [30]. Therefore, simultaneous delivery of two RNase effector molecules per mole diabody may have caused enhanced cytotoxicity owing to greater independence from target antigen reexpression. In addition, increased binding affinity and improved internalization due to the bivalency—as observed for a panel of anti-CD3 immunotoxins with varying affinities [31] and internalization behavior of anti-CD22 F(ab 0 )2 versus Fab 0 fragments [32], respectively—may have contributed to the increased potency of the dimeric molecule. Finally, dimerization may have facilitated better access of the fusion protein to cytosolic RNA substrates. To elucidate these mechanisms in detail, further studies are needed. For potential therapeutic applications the dimeric immunoenzyme is expected to offer important advantages over both the recombinant monomeric counterpart and the chemical immunoconjugate. Animal models comparing the pharmacokinetic behavior of IgGs and derived monovalent and bivalent antibody fragments have shown that the smallest molecules exhibited the best tumor penetration capacity [15] but were most rapidly cleared from the circulation [33]. Owing to their monovalency, monomeric scFvs moreover showed a markedly inferior tumor retention when compared with bivalent counterparts [34,35]. It has therefore been proposed that engineering therapeutic molecules with a molecular size ranging from 60 to 120 kDa provides the most ideal balance between tumor penetration, retention, and clearance [36]. The dimeric immunoenzyme characterized here has a size of about 90 kDa and thus can be expected to fulfill pharmacokinetic properties as demanded for successful therapeutic applications. We have recently demonstrated that chemical immunoconjugates consisting of anti-CD22 mAbs and either Ranpirnase or rapLRI exhibited very low systemic toxicity in tumor bearing mice [10,13]. Animals treated with anti-CD22 mAb–rapLRI immunoconjugates survived a total cumulative dose (TCD) of 500 mg/kg with only reversible weight loss upon cessation of treatment [13].

These data very favorably compare with anti-CD22 immunotoxins with reported LD50 values ranging from 0.5 to 80 mg/kg administered TCD [37–40]. Because the recombinant fusion protein derivative characterized in the present study can be expected to exhibit a similarly favorable toxicity profile as the corresponding mAb-RNase immunoconjugate, its therapeutic index will be significantly broader than that of fusion proteins made with plant or bacterial toxins. To summarize, we have generated a diabody–ribonuclease fusion protein that exhibits potent selective toxicity towards targeted cancer cells. Because the immunoenzyme is additionally expected not to elicit appreciable immunogenicity and systemic toxicity but to exhibit favorable pharmacokinetics when administered to humans, we believe this reagent merits further evaluation as a novel immunotherapeutic drug.

Acknowledgments The authors thank D. Ruby and M. Hursey for excellent technical assistance and Dr. E. Sausville for his continued interest and support. This project has been funded in whole or in part with Federal Funds from the National Cancer Institute, National Institutes of Health, under Contract No. N01-CO-12400, and grants from the Deutsche Krebshilfe/Dr. Mildred-Scheel-Stiftung Grant D/00/39301 (to J.K.), and Deutsche Akademie der Naturforscher Leopoldina Grant BMBF-LPD 9901/8-33 (to M.A.E.A.). The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. The publisher or recipient acknowledges the right of the US government to retain a non-exclusive, royaltyfree license in and to any copyright covering the article.

References [1] M. Mehren Mv, G.P. Adams, L.M. Weiner, Monoclonal antibody therapy for cancer, Annu. Rev. Med. 54 (2003) 343–369. [2] A.E. Frankel, D.M. Neville, T.A. Bugge, R.J. Kreitman, S.H. Leppla, Immunotoxin therapy of hematologic malignancies, Semin. Oncol. 30 (2003) 545–557. [3] R.J. Kreitman, W.H. Wilson, K. Bergeron, M. Raggio, M. Stetler-Stevenson, D.J. FitzGerald, I. Pastan, Efficacy of the antiCD22 recombinant immunotoxin BL22 in chemotherapy- resistant hairy-cell leukemia, N. Engl. J. Med. 345 (2001) 241–247. [4] P.L. Amlot, M.J. Stone, D. Cunningham, J. Fay, J. Newman, R. Collins, R. May, M. McCarthy, J. Richardson, V. Ghetie, A phase I study of an anti-CD22-deglycosylated ricin A chain immunotoxin in the treatment of B-cell lymphomas resistant to conventional therapy, Blood 82 (1993) 2624–2633. [5] E.S. Vitetta, M. Stone, P. Amlot, J. Fay, R. May, M. Till, J. Newman, P. Clark, R. Collins, D. Cunningham, et al., Phase I

J. Krauss et al. / Biochemical and Biophysical Research Communications 331 (2005) 595–602

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

immunotoxin trial in patients with B-cell lymphoma, Cancer Res. 51 (1991) 4052–4058. R.A. Messmann, E.S. Vitetta, D. Headlee, A.M. Senderowicz, W.D. Figg, J. Schindler, D.F. Michiel, S. Creekmore, S.M. Steinberg, D. Kohler, E.S. Jaffe, M. Stetler-Stevenson, H. Chen, V. Ghetie, E.A. Sausville, A phase I study of combination therapy with immunotoxins IgG-HD37-deglycosylated ricin A chain (dgA) and IgG-RFB4-dgA (Combotox) in patients with refractory CD19(+), CD22(+) B cell lymphoma, Clin. Cancer Res. 6 (2000) 1302–1313. E.A. Sausville, D. Headlee, M. Stetler-Stevenson, E.S. Jaffe, D. Solomon, W.D. Figg, J. Herdt, W.C. Kopp, H. Rager, S.M. Steinberg, Continuous infusion of the anti-CD22 immunotoxin IgG-RFB4-SMPT-dgA in patients with B-cell lymphoma: a phase I study, Blood 85 (1995) 3457–3465. S.M. Rybak, D.L. Newton, Natural and engineered cytotoxic ribonucleases: therapeutic potential, Exp. Cell Res. 253 (1999) 325–335. S.M. Mikulski, J.J. Costanzi, N.J. Vogelzang, S. McCachren, R.N. Taub, H. Chun, A. Mittelman, T. Panella, C. Puccio, R. Fine, K. Shogen, Phase II trial of a single weekly intravenous dose of ranpirnase in patients with unresectable malignant mesothelioma, J. Clin. Oncol. 20 (2002) 274–281. D.L. Newton, H.J. Hansen, S.M. Mikulski, D.M. Goldenberg, S.M. Rybak, Potent and specific antitumor effects of an antiCD22-targeted cytotoxic ribonuclease: potential for the treatment of non-Hodgkin lymphoma, Blood 97 (2001) 528–535. R.J. Kreitman, H.J. Hansen, A.L. Jones, D.J. FitzGerald, D.M. Goldenberg, I. Pastan, Pseudomonas exotoxin-based immunotoxins containing the antibody LL2 or LL2-Fab 0 induce regression of subcutaneous human B-cell lymphoma in mice, Cancer Res. 53 (1993) 819–825. S. Chen, S.Y. Le, D.L. Newton, J.V. Maizel Jr., S.M. Rybak, A gender-specific mRNA encoding a cytotoxic ribonuclease contains a 3 0 UTR of unusual length and structure, Nucleic Acids Res. 28 (2000) 2375–2382. M. Hursey, D.L. Newton, H.J. Hansen, D. Ruby, D.M. Goldenberg, S.M. Rybak, Specifically targeting the CD22 receptor of human B-cell lymphomas with RNA damaging agents: a new generation of therapeutics, Leuk. Lymphoma 43 (2002) 953–959. W. Debinski, I. Pastan, Monovalent immunotoxin containing truncated form of Pseudomonas exotoxin as potent antitumor agent, Cancer Res. 52 (1992) 5379–5385. T. Yokota, D.E. Milenic, M. Whitlow, J. Schlom, Rapid tumor penetration of a single-chain Fv and comparison with other immunoglobulin forms, Cancer Res. 52 (1992) 3402–3408. R.W. Schroff, K.A. Foon, S.M. Beatty, R.K. Oldham, A.C. Morgan Jr., Human anti-murine immunoglobulin responses in patients receiving monoclonal antibody therapy, Cancer Res. 45 (1985) 879–885. J. Krauss, M.A. Arndt, A.C. Martin, H. Liu, S.M. Rybak, Specificity grafting of human antibody frameworks selected from a phage display library: generation of a highly stable humanized anti-CD22 single-chain Fv fragment, Protein Eng. 16 (2003) 753– 759. P. Holliger, T. Prospero, G. Winter, ‘‘Diabodies’’: small bivalent and bispecific antibody fragments, Proc. Natl. Acad. Sci. USA 90 (1993) 6444–6448. S.N. Ho, H.D. Hunt, R.M. Horton, J.K. Pullen, L.R. Pease, Sitedirected mutagenesis by overlap extension using the polymerase chain reaction, Gene 77 (1989) 51–59. S.M. Kipriyanov, O.A. Kupriyanova, M. Little, G. Moldenhauer, Rapid detection of recombinant antibody fragments directed against cell-surface antigens by flow cytometry, J. Immunol. Methods 196 (1996) 51–62. R. Shapiro, S. Weremowicz, J.F. Riordan, B.L. Vallee, Ribonucleolytic activity of angiogenin: essential histidine, lysine, and

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36] [37]

601

arginine residues, Proc. Natl. Acad. Sci. USA 84 (1987) 8783– 8787. D.L. Newton, Y. Xue, K.A. Olson, J.W. Fett, S.M. Rybak, Angiogenin single-chain immunofusions: influence of peptide linkers and spacers between fusion protein domains, Biochemistry 35 (1996) 545–553. J. Krauss, M.A. Arndt, Z. Zhu, D.L. Newton, B.K. Vu, V. Choudhry, R. Darbha, X. Ji, N.S. Courtenay-Luck, M.P. Deonarain, J. Richards, S.M. Rybak, Impact of antibody framework residue V(H)-71 on the stability of a humanised anti-MUC1 scFv and derived immunoenzyme, Br. J. Cancer 90 (2004) 1863–1870. P.H. Tan, B.M. Sandmaier, P.S. Stayton, Contributions of a highly conserved VH/VL hydrogen bonding interaction to scFv folding stability and refolding efficiency, Biophys. J. 75 (1998) 1473–1482. P.S. Chowdhury, G. Vasmatzis, R. Beers, B. Lee, I. Pastan, Improved stability and yield of a Fv-toxin fusion protein by computer design and protein engineering of the Fv, J. Mol. Biol. 281 (1998) 917–928. N. Hugo, M. Weidenhaupt, M. Beukes, B. Xu, J.C. Janson, T. Vernet, D. Altschuh, VL position 34 is a key determinant for the engineering of stable antibodies with fast dissociation rates, Protein Eng. 16 (2003) 381–386. E.A. Kabat, T.T. Wu, Identical V region amino acid sequences and segments of sequences in antibodies of different specificities. Relative contributions of VH and VL genes, minigenes, and complementarity-determining regions to binding of antibodycombining sites, J. Immunol. 147 (1991) 1709–1719. E. Vargas-Madrazo, E. Paz-Garcia, An improved model of association for VH-VL immunoglobulin domains: asymmetries between VH and VL in the packing of some interface residues, J. Mol. Recogn. 16 (2003) 113–120. M.P. Deonarain, A.A. Epenetos, Design, characterization and anti-tumour cytotoxicity of a panel of recombinant, mammalian ribonuclease-based immunotoxins, Br. J. Cancer 77 (1998) 537– 546. J. Carnahan, P. Wang, R. Kendall, C. Chen, S. Hu, T. Boone, T. Juan, J. Talvenheimo, S. Montestruque, J. Sun, G. Elliott, J. Thomas, J. Ferbas, B. Kern, R. Briddell, J.P. Leonard, A. Cesano, Epratuzumab, a humanized monoclonal antibody targeting CD22: characterization of in vitro properties, Clin. Cancer Res. 9 (2003) 3982S–3990S. J.M. Hexham, D. Dudas, R. Hugo, J. Thompson, V. King, C. Dowling, D.M. Neville Jr., M.E. Digan, P. Lake, Influence of relative binding affinity on efficacy in a panel of anti-CD3 scFv immunotoxins, Mol. Immunol. 38 (2001) 397–408. L.B. Shih, H.H. Lu, H. Xuan, D.M. Goldenberg, Internalization and intracellular processing of an anti-B-cell lymphoma monoclonal antibody, LL2, Int. J. Cancer 56 (1994) 538–545. D.E. Milenic, T. Yokota, D.R. Filpula, M.A. Finkelman, S.W. Dodd, J.F. Wood, M. Whitlow, P. Snoy, J. Schlom, Construction, binding properties, metabolism, and tumor targeting of a singlechain Fv derived from the pancarcinoma monoclonal antibody CC49, Cancer Res. 51 (1991) 6363–6371. G.P. Adams, J.E. McCartney, M.S. Tai, H. Oppermann, J.S. Huston, W.F.d. Stafford, M.A. Bookman, I. Fand, L.L. Houston, L.M. Weiner, Highly specific in vivo tumor targeting by monovalent and divalent forms of 741F8 anti-c-erbB-2 single-chain Fv, Cancer Res. 53 (1993) 4026–4034. U.B. Nielsen, G.P. Adams, L.M. Weiner, J.D. Marks, Targeting of bivalent anti-ErbB2 diabody antibody fragments to tumor cells is independent of the intrinsic antibody affinity, Cancer Res. 60 (2000) 6434–6440. P.J. Hudson, C. Souriau, Engineered antibodies, Nat. Med. 9 (2003) 129–134. A. Bolognesi, P.L. Tazzari, F. Olivieri, L. Polito, R. Lemoli, A. Terenzi, L. Pasqualucci, B. Falini, F. Stirpe, Evaluation of

602

J. Krauss et al. / Biochemical and Biophysical Research Communications 331 (2005) 595–602

immunotoxins containing single-chain ribosome-inactivating proteins and an anti-CD22 monoclonal antibody (OM124): in vitro and in vivo studies, Br. J. Haematol. 101 (1998) 179–188. [38] M.A. Ghetie, K. Tucker, J. Richardson, J.W. Uhr, E.S. Vitetta, The antitumor activity of an anti-CD22 immunotoxin in SCID mice with disseminated Daudi lymphoma is enhanced by either an anti-CD19 antibody or an anti-CD19 immunotoxin, Blood 80 (1992) 2315–2320.

[39] E. Mansfield, P. Amlot, I. Pastan, D.J. FitzGerald, Recombinant RFB4 immunotoxins exhibit potent cytotoxic activity for CD22bearing cells and tumors, Blood 90 (1997) 2020–2026. [40] P.J. Van Horssen, F.W. Preijers, Y.V. Van Oosterhout, T. De Witte, Highly potent CD22-recombinant ricin A results in complete cure of disseminated malignant B-cell xenografts in SCID mice but fails to cure solid xenografts in nude mice, Int. J. Cancer 68 (1996) 378–383.