Bivalent disulfide-stabilized fragment variable immunotoxin directed against mesotheliomas and ovarian cancer

We have used protein engineering to generate a stable bivalent fragment variable (Fv) molecule from the antimesothelin antibody SS, in which the VH and VL domains of the Fv are linked to each other by a disulfide bond, and the two Fvs are connected by a flexible 15-amino acid (Gly4-Ser)3 linker. The SS (dsFv)2 molecule is fused to a M(r) 38,000 truncated form of Pseudomonas exotoxin to generate a bivalent, disulfide stabilized, (dsFv)2 immunotoxin. The immunotoxin was expressed in Escherichia coli, refolded in vitro, and purified to approximately 95% purity with a high yield of > 10%. Binding studies demonstrated that the (dsFv)2 molecule has 40 times higher apparent affinity for recombinant mesothelin than a monovalent dsFv molecule. The (dsFv)2 immunotoxin was 4-10-fold more cytotoxic to three mesothelin antigen-positive cell lines than the monovalent dsFv immunotoxin. However, when tested in mice bearing tumor cells expressing mesothelin, the antitumor activity of the bivalent immunotoxin is very similar to the activity of the lower affinity monovalent immunotoxin. Our data indicate that increasing affinity of an antibody fragment does not necessarily lead to higher antitumor activity of an immunotoxin in vivo.

Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv

Combinatorial variation of CDR3 of V(H) and V(L), followed by phage display, was used to select affinity mutants of the parental anti-epidermal growth factor receptor-vIII (EGFRvIII) scFv MR1. One mutant, MR1-1(scFv), had increased specific binding affinity for EGFRvIII. It was produced and radiolabeled, and its biodistribution was evaluated in human glioma-bearing athymic mice. MR1-1 targeted the same EGFRvIII epitope as MR1 with an approximately 15-fold higher affinity (K(d) = 1.5 x 10(-9) M) measured by surface resonance analysis. Labeling with (131)I or (125)I was performed, and the immunoreactive fraction of the labeled MR1-1(scFv) was 50% to 55%. After incubation at 37 degrees C for 4 days, the binding affinity was maintained at 60% of initial levels. The specificity of MR1-1 for EGFRvIII was demonstrated in vitro by flow cytometry and incubation of FITC-labeled scFv with the EGFRvIII-expressing U87MG. DeltaEGFR cell line or with the EGFRvIII-negative U87MG cell line in the presence or absence of competing unlabeled MR1-1(scFv). We also investigated the internalization and processing of MR1-1 compared with MR1; MR1-1 exhibited levels of both cell surface retention and internalization up to 5 times higher than those by MR1. In biodistribution studies performed in athymic mice bearing s.c. U87MG. DeltaEGFR tumor xenografts, animals received paired-label intratumoral infusions of (131)I-labeled MR1-1(scFv) and (125)I-labeled MR1(scFv). Our results showed an up to 244% +/- 77% increase in tumor uptake for MR1-1 compared with that for MR1. The improved tumor retention of MR1-1(scFv) combined with its rapid clearance from normal tissues also resulted in sustained higher tumor:normal organ ratios. These results suggest that the improved affinity of MR1-1 can significantly impact in vivo glioma-specific targeting and immunotherapy.

Increased binding affinity enhances targeting of glioma xenografts by EGFRvIII-specific scFv

Combinatorial variation of CDR3 of V(H) and V(L), followed by phage display, was used to select affinity mutants of the parental anti-epidermal growth factor receptor-vIII (EGFRvIII) scFv MR1. One mutant, MR1-1(scFv), had increased specific binding affinity for EGFRvIII. It was produced and radiolabeled, and its biodistribution was evaluated in human glioma-bearing athymic mice. MR1-1 targeted the same EGFRvIII epitope as MR1 with an approximately 15-fold higher affinity (K(d) = 1.5 x 10(-9) M) measured by surface resonance analysis. Labeling with (131)I or (125)I was performed, and the immunoreactive fraction of the labeled MR1-1(scFv) was 50% to 55%. After incubation at 37 degrees C for 4 days, the binding affinity was maintained at 60% of initial levels. The specificity of MR1-1 for EGFRvIII was demonstrated in vitro by flow cytometry and incubation of FITC-labeled scFv with the EGFRvIII-expressing U87MG. DeltaEGFR cell line or with the EGFRvIII-negative U87MG cell line in the presence or absence of competing unlabeled MR1-1(scFv). We also investigated the internalization and processing of MR1-1 compared with MR1; MR1-1 exhibited levels of both cell surface retention and internalization up to 5 times higher than those by MR1. In biodistribution studies performed in athymic mice bearing s.c. U87MG. DeltaEGFR tumor xenografts, animals received paired-label intratumoral infusions of (131)I-labeled MR1-1(scFv) and (125)I-labeled MR1(scFv). Our results showed an up to 244% +/- 77% increase in tumor uptake for MR1-1 compared with that for MR1. The improved tumor retention of MR1-1(scFv) combined with its rapid clearance from normal tissues also resulted in sustained higher tumor:normal organ ratios. These results suggest that the improved affinity of MR1-1 can significantly impact in vivo glioma-specific targeting and immunotherapy.

Site-specific expression of transferrin receptor by human colon cancer cells directly correlates with eradication by antitransferrin recombinant immunotoxin

We determined the efficacy of HB21(Fv)PE40, a single-chain immunotoxin made by fusing the variable regions of a monoclonal antibody directed at the human transferrin receptor (TfR) with a truncated mutant of Pseudomonas exotoxin (PE), against metastatic human colon carcinoma KM12L4 cells growing in the liver or subcutis of nude mice. Organ-specific modulation of TfR expression was examined by immunohistochemistry and flow cytometry using anti-human CD71 antibody. KM12L4 cells expressed human TfR and were lysed in vitro by HB21(Fv)PE40 but not LMB-7 (a control immunotoxin specific for a Lewis Y-related carbohydrate antigen). KM12L4 cells growing in the liver expressed higher levels of TfR than cells growing s.c. Systemic administration of HB21(Fv)PE40 eliminated KM12L4 liver metastasis, whereas administration of LMB-7 did not. Treatment of mice with HB21(Fv)PE40 only delayed the growth of s.c. tumors. KM12L4 cells recovered from liver metastases, expressed higher levels of TfR, and were more sensitive to lysis by HB21(Fv)PE40 than KM12L4 cells recovered from s.c. tumors. Indeed, collectively, the data show that the expression level of the TfR by human colon cancer cells is modulated by the organ microenvironment which can be advantageous for the use of therapeutic immunotoxins.

Immunotoxins with increased activity against epidermal growth factor receptor vIII-expressing cells produced by antibody phage display

Recombinant immunotoxins are fusion proteins composed of Fv regions of antibodies and bacterial or plant toxins that are being developed for the targeted therapy of cancer. MR1(Fv)-PE38 is a single-chain recombinant immunotoxin that targets a mutant form of the epidermal growth factor receptor (EGFR), EGFRvIII, that is frequently overexpressed in malignant glioblastomas. We have used random complementarity determining region (CDR) mutagenesis to obtain mutants of MR1(Fv) with an increased affinity for EGFRvIII and an increased activity when converted to a recombinant immunotoxin. Initially, nine residues of heavy chain CDR3 were randomly mutagenized, and several mutants with increased binding affinity were isolated. All mutations were located at amino acids 98 and 99, which correspond to a DNA hot spot, a DNA sequence that mutates at high frequency during natural antibody maturation. A specific region of variable region of antibody light chain CDR3 was mutagenized that corresponded to a hot spot and a mutant (MR1-1) with an additional increase in affinity, and cytotoxic activity was isolated. These studies show that targeting hot spots in the CDRs of Fvs is an effective approach to obtaining Fvs with increased affinity. The increased affinity of MR1-1(Fv) makes it an attractive candidate for the targeted therapy of glioblastomas.

Improved stability and yield of a Fv-toxin fusion protein by computer design and protein engineering of the Fv

The conversion of the anti-mesothelin monoclonal antibody K1 to a single-chain Fv (scFv) that is fused to a truncated form of Pseudomonas exotoxin A (PE) results in a fusion protein (immunotoxin) that is unstable and refolds very inefficiently. We have devised a method that identifies candidate residues in the framework region of K1 Fv that, when mutated, improved the yield and stability of the protein. The method works by initially aligning the framework sequences of K1 VH and VL with those of other scFvs that are stable and give a good yield as immunotoxins. Then we assigned a character to each residue that indicates its state of exposure based on the known crystal structures of Fabs. This identifies residues that are not compatible with their environment in the folded state of the protein. Next we calculated the frequencies of different amino acids for each position of the Fvs based on the available sequence database. This identifies residues that are not commonly present in the conserved positions. If these residues are compatible with their exposure profile they are left unaltered. Otherwise, they are identified as candidate residues for mutation. We identified two such residues in the VH (T82 and A85) and two in the VL (H36 and V60) of K1 that did not seem appropriate for their respective positions. By mutating these residues in K1 into those that occur most commonly in the sequence database or in stable scFvs, we significantly improved the stability and yield of the K1 scFv immunotoxins. By making single and combined mutations we assessed the relative contribution of mutations at these four sites towards the stability and yield of K1 scFv immunotoxins. The method we devised is probably general and can be used to improve other scFvs.

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

Mesothelin is a differentiation antigen present on the surface of ovarian cancers, mesotheliomas, and several other types of human cancers. Because among normal tissues, mesothelin is present only on mesothelial cells, it represents a good target for antibody-mediated delivery of cytotoxic agents. In the present study mice were immunized with an eukaryotic expression vector coding for mesothelin. When high serum antibody titers were obtained, a phage display library was made from the splenic mRNA of these mice. After three rounds of panning on recombinant mesothelin, a single-chain Fv (scFv)-displaying phage was selected that bound specifically to recombinant mesothelin and mesothelin-positive cells. The scFv was used to construct an immunotoxin by genetically fusing it with a truncated mutant of Pseudomonas exotoxin A. The purified immunotoxin binds mesothelin with high affinity (Kd 11 nm), is stable for over 40 hr at 37 degrees C and is very cytotoxic to cells expressing mesothelin. It also produces regressions of tumors expressing mesothelin. This combination of selective cytotoxicity, high activity, and stability makes the immunotoxin a good candidate for development as a therapeutic agent. This work also shows that DNA immunization can be used to isolate and clone antibodies against epitopes present on human proteins in their native conformation.

Cloning and expression of the recombinant FAb fragment of monoclonal antibody K1 that reacts with mesothelin present on mesotheliomas and ovarian cancers

The monoclonal antibody (MAb) K1 recognizes an approximate 40 kDa glycoprotein, mesothelin, that is present on the surface of human mesothelial cells, mesotheliomas and ovarian cancers. We have cloned the cDNAs encoding the variable regions of MAb K1 and constructed plasmids for expression of recombinant K1(FAb). Recombinant FAb was produced in Escherichia coli in inclusion bodies that were solubilized and refolded to active protein. Binding of K1 MAb and FAb was compared by radioactive binding and competition assays and by surface plasmon resonance (BIAcore). Recombinant K1(FAb) binds to cells expressing K1-antigen with a similar affinity as papain derived FAb from K1(IgG) and with a 4- to 10-fold reduced affinity compared with bivalent IgG. The cloned FAb can be used to make higher affinity antibodies and immunoconjugates that could be useful for various types of immunotherapies.

Stabilization of a recombinant Fv fragment by base-loop interconnection and V(H)-V(L) permutation

We have developed a novel method to stabilize a recombinant antibody Fv fragment. The V(H) and V(L) domains of this Fv fragment, called pFv (permutated Fv), are covalently interconnected to each other at the two "base-loops" that normally connect V(H) beta strand 3 to 3b and V(L) beta strand 3 to 3b. To produce the base-loop stabilized Fv fragment, we connected the N-terminal half of the V(L) domain (V(L) 1-40) of murine antibody anti-Tac to the C-terminal half of V(H) (V(H) 42-115). We also fused the C terminus of V(H) by a (Gly4Ser)3 linker to the N-terminal half of V(H) (V(H) 1-40, thereby generating a permutated V(H) domain). Finally we connected the base loop of V(H) (N-terminal half) to the C-terminal half of V(L) (V(H) 42-115). The anti-Tac pFv fragment was fused to a truncated form of Pseudomonas exotoxin to generate a pFv-immunotoxin. Fvs with the correct structure were produced by refolding of recombinant inclusion body protein using a renaturation protocol that was originally developed for Fab and scFv fragments. Due to the artificially connected and permutated primary sequence, the folding pathway for the pFv structure may possibly be different from the conventional folding of antibody domains. Analysis of antigen binding of anti-Tac pFv, and of the specific cytotoxicity of pFv-immunotoxin towards antigen expressing cancer cells demonstrated that the anti-Tac pFv retained most of its affinity and full specificity when compared to anti-Tac scFv. Also anti-Tac pFv was relatively stable, retaining 25% of its binding activity after a 24 hour incubation in human serum at 37 degrees C. This indicates that connection of base loops can be a useful alternative to linker or disulfide stabilization of Fv fragments.

The MyoD family of myogenic factors is regulated by electrical activity: isolation and characterization of a mouse Myf-5 cDNA

A full-length cDNA coding for a homolog of the human Myf-5 was isolated from a BC3H-1 mouse library and characterized. The clone codes for a protein of 255 amino acids that is 89%, 88% and 68% identical to the human, bovine and Xenopus myf-5, respectively. The mouse Myf-5 cDNA (mmyf-5), as well as sequences coding for MyoD, myogenin and Mrf-4, were used to probe Northern blots to analyze the effects of innervation on the expression of the MyoD family of myogenic factors. Mouse myf-5, MyoD and myogenin mRNAs levels were found to decline in hind limb muscles of mice between embryonic day 15 (E15) and the first postnatal week, a period that coincides with innervation. In contrast, Mrf-4 transcripts increase during this period and reach steady-state levels by 1-week after birth. To distinguish if the changes in myogenic factor expression are due to a developmental program or to innervation, mRNA levels were analyzed at different times after muscle denervation. Mmyf-5 transcripts begin to accumulate 2 days postdenervation; after 1 week levels are 7-fold higher than in innervated muscle. Mrf-4, MyoD and myogenin transcripts begin to accumulate as soon as 8h after denervation, and attain levels that are 8-, 15- and 40-fold higher than found in innervated skeletal muscle, respectively. The accumulation of these three mRNAs precedes the increase of nicotinic acetylcholine receptor alpha subunit transcripts, a gene that is transcriptionally regulated by MyoD-related factors in vitro. Using extracellular electrodes to directly stimulate in situ the soleus muscle of rats, we found that 'electrical activity' per se, in absence of the nerve, represses the increases of myogenic factor mRNAs associated with denervation.

Membrane topology of Escherichia coli prolipoprotein signal peptidase (signal peptidase II)

The lsp gene of Escherichia coli encodes the inner membrane enzyme, signal peptidase II (SPase II). SPase II is comprised of 164 amino acid residues and contains four hydrophobic domains. A series of lsp-phoA and lsp-lacZ gene fusions have been constructed in vitro to determine the topology of SPase II. The fusion junction for each of these gene fusions was determined by DNA sequencing. The lengths of the SPase II fragment in the fusions varied from 12 to 159 amino acid residues. Strains containing SPase II-PhoA fusions to the two predicted periplasmic loops exhibited higher levels of alkaline phosphatase activity than fusions to the predicted cytoplasmic domains. In contrast, SPase II-LacZ fusions at the cytoplasmic and the periplasmic domains of SPase II showed high and low levels of beta-galactosidase activity, respectively, a result opposite to those shown by SPase II-PhoA fusions located at precisely the same amino acid of SPase II. Taken together, these results strongly support the predicted model for SPase II topology, i.e. this enzyme spans the cytoplasmic membrane four times with both the amino and the carboxyl termini facing the cytoplasm.

Nucleotide sequence of the Pseudomonas fluorescens signal peptidase II gene (lsp) and flanking genes

The lsp gene encoding prolipoprotein signal peptidase (signal peptidase II) is organized into an operon consisting of ileS and three open reading frames, designated genes x, orf149, and orf316 in both Escherichia coli and Enterobacter aerogenes. A plasmid, pBROC128, containing a 5.8-kb fragment of Pseudomonas fluorescens DNA was found to confer pseudomonic acid resistance on E. coli host cells and to contain the structural gene of ileS from P. fluorescens. In addition, E. coli strains carrying pBROC128 exhibited increased globomycin resistance. This indicated that the P. fluorescens lsp gene was present on the plasmid. The nucleotide sequences of the P. fluorescens lsp gene and of its flanking regions were determined. Comparison of the nucleotide sequences of the lsp genes in E. coli and P. fluorescens revealed two highly conserved domains in this enzyme. Furthermore, the five genes which constitute an operon in E. coli and Enterobacter aerogenes were found in P. fluorescens in the same order as in the first two species.

Isolation and characterization of a new globomycin-resistant dnaE mutant of Escherichia coli

We isolated a globomycin-resistant, temperature-sensitive mutant of Escherichia coli K-12 strain AB1157. The mutation mapped in dnaE, the structural gene for the alpha-subunit of DNA polymerase III. The in vivo processing of lipid-modified prolipoprotein was more resistant to globomycin in the mutant strain 307 than in its parent. The prolipoprotein signal peptidase activity was also increased twofold in the mutant, and there was a threefold increase in the activity of isoleucyl-tRNA synthetase. The results suggest that a mutation in dnaE may affect the expression of the ileS-lsp operon in E. coli. In addition, strain 307 showed a reduced level of streptomycin resistance compared with its parental strain AB1157 (rpsL31). Strain 307 was killed by streptomycin at a concentration of 200 micrograms/ml, which did not affect the rate of bulk protein synthesis in this mutant. A second mutation which was involved in the reduced streptomycin resistance in strain 307 was identified and found to be closely linked to or within the rpsD (ramA, ribosomal ambiguity) gene. Both dnaE and rpsD were required for the reduced streptomycin resistance in strain 307.