PEDS Advance Access originally published online on October 21, 2005
Protein Engineering Design and Selection 2006 19(1):37-45; doi:10.1093/protein/gzi073
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cytotoxicity of human RNase-based immunotoxins requires cytosolic access and resistance to ribonuclease inhibition
1Department of Laboratory Medicine and Pathology, Cancer Center and Center for Immunology, University of Minnesota, Minneapolis, MN 55455, USA
2 To whom correspondence should be addressed. E-mail: penne001{at}umn.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Immunotoxins are targeted therapeutics designed to kill cancer cells. The targeting moiety of an immunotoxin selectively binds to a tumor cell and targets it for death via an attached toxin. Because the toxins are typically of plant or bacterial origin, their clinical use is limited by immunogenicity and nonspecific toxicity. To circumvent these problems, we have begun to engineer immunotoxins containing human pancreatic ribonuclease. Here we describe the generation of ribonuclease mutants designed to evade a ubiquitous cytosolic inhibitor that would otherwise block cytotoxicity. Two mutants retained catalytic activity and were relatively resistant to the inhibitor. To deliver them to human T leukemic cells, these ribonuclease variants were fused to a single chain Fv fragment specific for CD7. The ribonuclease–sFv fusion proteins bound CD7+ T cells and were internalized yet were not cytotoxic. Transfection of the proteins directly into the cytosol reduced cell viability, suggesting that the failure of the immunotoxins to kill cells when added externally resulted from the inability of the ribonuclease moiety to access the cytosol efficiently. Our results indicate appropriate intracellular routing, as well as resistance to inhibition, is critical to the cytotoxicity of human ribonuclease-based immunotoxins.
Keywords: cancer therapy/human RNase/immunotoxin/Onconase/ribonuclease inhibitor
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Targeted toxins are chimeric proteins designed to kill malignant cells preferentially (Erickson and Pennell, 2002
). These proteins contain a targeting component covalently linked to a toxin. Typically, the toxin is of plant or bacterial origin while the targeting moiety is a ligand or antibody specific for a receptor expressed on the cell population of interest. Targeted toxins containing antibodies or antibody fragments are called immunotoxins (ITs).
Although numerous ITs have gone into clinical trials, their therapeutic utility has not been fully realized because of immunogenicity and nonspecific toxicity (Frankel et al., 2000). Most toxins damage vascular endothelial cells, liver cells or renal cells directly or indirectly through the induction of inflammatory responses (Baluna and Vitetta, 1997; Onda et al., 2000). To minimize toxicity and immunogenicity, investigators have begun to explore ITs containing human-derived effector proteins, such as those in the pancreatic ribonuclease (RNase) A superfamily.
Members of the RNase A superfamily are small (13 kDa), stable secretory proteins that aid in digestion, host defense and angiogenesis (Fett et al., 1985; D'Alessio et al., 1991; Hooper et al., 2003). Human RNases are present in extracellular fluids and correspondingly are not immunogenic. Host cells are protected from endogenous RNases by ribonuclease inhibitor (RI), a ubiquitous cytosolic protein present at high concentrations (0.01–0.1% of cellular protein). RI binds the active sites of most RNase A family members with femtomolar affinities and is thought to guard cells against rogue RNases that access the cytosol (Lee and Vallee, 1989; Vicentini et al., 1990).
In contrast to human RNases, frog RNases and bovine seminal RNase (BS-RNase) are resistant to RI and are cytotoxic to human cancer cells. BS-RNase depletes multi-drug resistant tumor cells from ex vivo-expanded blood-derived CD34+ hematopoietic progenitor cells (Cinatl et al., 1999; Kotchetkov et al., 2000) while RNases isolated from Rana catesbeiana and Rana japonica selectively kill tumor cell lines (Nitta et al., 1994; Liao et al., 1996). Onconase, which is isolated from the Northern leopard frog Rana pipiens, is being evaluated in a Phase III randomized trial in patients with malignant mesothelioma (Mikulski et al., 2002). Phase I and Phase II trials showed that renal toxicity was dose limiting and that Onconase was immunologically well tolerated with repeated administration, most probably due to its structural similarity to human RNases (Mikulski et al., 1993, 2002).
Although Onconase has a similar tertiary structure to mammalian RNase A family members, its affinity for RI is only in the micromolar range (Boix et al., 1996). This is because many of the RI contact residues identified in mammalian RNase A are absent or substituted in Onconase (Boix et al., 1996). BS-RNase escapes RI via a different mechanism. BS-RNase naturally occurs as a homodimer and, as such, its active sites are sterically inaccessible to RI (Kim et al., 1995a, b). It follows that human RNase could only be effective as a targeted therapeutic if it was modified to retain catalytic activity and avoid RI-mediated inactivation upon internalization.
Antigens targeted by ITs must be preferentially expressed by tumor cells and internalized upon binding. CD7 is a cell surface differentiation antigen expressed by most normal T and NK cells (Barcena et al., 1995; Sempowski et al., 1999). CD7 binds galectin I and is required for the delivery of galectin I's proapoptotic signal to T cells (Pace et al., 2000). CD7 ligation upregulates adhesion molecule expression and plays a role in T and NK cell activation (Shimizu et al., 1992; Leta et al., 1995). It is internalized rapidly upon binding, is not shed from the cell surface and, notably, is present at high densities on almost all T cell acute lymphoblastic leukemias (T-ALL) (Carriere et al., 1989; Flavell et al., 1991; Tonevitsky et al., 1993). Together these qualities make CD7 an ideal target antigen for ITs designed to treat T-ALL. CD7-specific ITs have been evaluated clinically, and although all suffered from dose-limiting toxicities such as vascular leak syndrome, many have had anti-leukemic effects (Frankel et al., 2000).
In an attempt to improve upon CD7-specific ITs by reducing nonspecific toxicity and immunogenicity, we have now generated RI-resistant RNase 1 mutants and linked them to a CD7-specific single chain Fv (sFv) antibody fragment. We describe the enzymatic characteristics of these RNase 1 variants and the requirements they must meet to function as effective ITs.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Materials
Escherichia coli strains XL-1 Blue and BL21(DE3)pLysS, and vector pET17b, were from Novagen (Madison, WI). Restriction enzymes were purchased from Invitrogen (Carlsbad, CA) and New England Biolabs (Beverly, MA). Mutations were introduced with Stratagene's QuikChange Site-Directed Mutagenesis kit (La Jolla, CA) using PAGE-purified oligonucleotides from Integrated DNA Technologies (Coralville, IA). The ribonuclease substrate uridylyl-3'5' adenosine (UpA) was obtained from Sigma (St Louis, MO). Recombinant RI (RNasin) and the TnT Coupled Reticulocyte Lysate System used for the in vitro translation assays were both purchased from Promega (Madison, WI). Alamar Blue, the fluorometric growth indicator used in the in vitro cytotoxicity assays, was from Biosource (Camarillo, CA). The K-562 (chronic myelogenous leukemia) and Molt 3 (T lymphoblastic leukemia) cell lines were obtained from the American Type Culture Collection (Manassas, VA). The CD13+CD19+ human leukemic cell line SEM K2 was a kind gift from Dr John Kersey (University of Minnesota Cancer Center, Minneapolis, MN).
Mutagenesis
E.coli strain XL-1 Blue was used for plasmid propagation. Genetic alterations were made using PCR-mediated mutagenesis. See Table S-I in the supplementary material (Supplementary data available at PEDS online) for sequences of the mutagenic oligonucleotides. All PCR products were sequenced by the Advanced Genetic Analysis Center at the University of Minnesota to ensure only the intended mutations were introduced. A construct encoding RNase 1 with Gly89 mutated to Arg in vector pCR 2.1 was a gift from Dr Daniel Vallera (University of Minnesota Cancer Center, University of Minnesota). The gene encoding G89R RNase 1 was inserted into the vector pET17b and Arg89 was reverted to the native Gly to yield wild-type RNase 1. K7A RNase 1 was generated by mutating Lys7 in the wild-type RNase 1 construct to Ala. K41R RNase 1 was prepared by mutating Lys41 in the wild-type RNase 1 construct to Arg. Q69A RNase 1 was made by mutating Gln69 in the wild-type RNase 1 construct to Ala. The double mutant Q69A/K41R RNase 1 was prepared by mutating Gln69 to Ala in the K41R RNase 1 construct. The triple mutant RDD RNase 1 was generated by altering Asn88, Gly89 and Arg91 in the wild-type RNase 1 construct to Arg, Asp and Asp, respectively. The quadruplet mutant K7A/RDD RNase 1 contains Lys7 mutated to Ala in RDD RNase 1. The quintuplet mutant K41R/Q69A/RDD RNase 1 was made via mutation of Asn88, Gly89 and Arg91 in the K41R/Q69A RNase 1 construct to Arg, Asp and Asp, respectively.
Native Onconase contains a pyroglutamate residue at the N-terminus formed by Gln1 cyclization (Boix et al., 1996). This pyroglutamate residue is essential for catalytic activity. Recombinant Onconase has Met at the N-terminus that prevents Gln from cyclizing, resulting in an enzyme with significantly reduced enzymatic activity. To circumvent this problem, we substituted Ser for the N-proximal Gln because this mutation restored activity to 54% of native Onconase values (Newton et al., 1998).
Fusion protein construction
An oligonucleotide encoding a furin-sensitive peptide (TRHRQPRGWEQL) from Psuedomonas exotoxin A (Goyal and Batra, 2000) was inserted as a SnaBI and XhoI fragment downstream of the RNase in the pET17b construct. A NotI site was included at the 3' end of the protease site to allow the anti-CD7 sFv gene to be inserted as a NotI/XhoI fragment. This sFv gene had previously been constructed using a (Gly4Ser)3 linker to join the immunoglobulin heavy and light chain variable regions expressed by the hybridoma, 3A1f (Pauza et al., 1997).
Protein production and purification
Onconase, RNase 1 and the fusion protein constructs were transformed into E.coli strain BL21(DE3)pLysS cells via electroporation. Protein expression was induced by incubation with 1 mM isopropyl-ß-D-thiogalactopyranoside for 3 h. Bacterial cell pellets were lysed with a French pressure cell (Thermo Lab Systems, Franklin, MA). Inclusion bodies were harvested by centrifugation, washed exhaustively in salt and detergent buffers and solubilized in denaturing buffer containing 100 mM Tris, pH 8.0, 6 M guanidine-HCl, 2 mM EDTA and 300 mM dithioerythritol. Denatured proteins were refolded by their dropwise addition to refolding buffer at 10°C to a final concentration of 30 µg/ml and a final dilution of 1:100 denaturating:refolding buffer. The refolding buffer was 100 mM Tris, pH 8.0, 500 mM arginine, 8 mM oxidized glutathione and 2 mM EDTA. Fusion proteins were refolded in either this buffer or one in which 2 M urea was substituted for arginine (when proteins were to be directly applied to a cation column). Refolded proteins were incubated at 10°C for 48 h. Samples were concentrated by ultrafiltration using YM membranes (Amicon, Billerica, MA).
RNase and RNase–sFv proteins were purified using fast performance liquid chromatography with a S-sepharose cation column (Sigma). Fractions containing proteins of the predicted MW were exchanged into phosphate-buffered saline (PBS) and applied to a heparin column (Pharmacia, Piscataway, NJ). Proteins were eluted with a 0–1 M NaCl gradient in PBS. Fractions were screened via SDS–PAGE, pooled, concentrated using centrifugal filters (Amicon), desalted and glycerol added to 5%. No detectable difference was seen in the activity levels between the two different refolding conditions. Protein concentrations were determined by A280, using 
280 0.53 ml mg–1 cm–1 for RNase and/or a Bradford assay using BSA, to generate a standard curve. All proteins were stored at –80°C.
Ribonucleolytic activity assay
Enzymatic activity was assessed spectrophotometrically based on the assay of Witzel and Barnard (1962). The concentration of the substrate UpA was determined spectrophotometrically using 280 16 300 M–1 cm–1. Kinetic assay buffer consisted of 50 mM MES, pH 6.0 and 125 mM NaCl. Final enzyme concentrations were between 10 nM and 1 µM and UpA concentrations ranged from 25 to 1000 µM. The reaction was started with the addition of RNase. Depletion of UpA was measured at A280 with the Genesys 2 spectrophotometer (Thermo Lab Systems). Initial velocities were entered into the EZ-Fit program (Perella Scientific, Amherst, NH) to calculate Km and Vmax values.
In vitro translation assay
An in vitro translation assay (the TnT Coupled Reticulocyte Lysate System) measured alterations in RNase activity in the presence of RI. Rabbit reticulocyte lysate cocktail was prepared including T7 polymerase, DNA encoding the luciferase gene under control of the T7 promoter, complete amino acids and kit buffer. The cocktail was quartered and RI added to the appropriate concentration (0, 2.5, 25 or 250 nM). Lysate cocktail with RI was then aliquoted and RNases added to a final concentration of 25 nM. Controls included samples with PBS substituted for RNase and samples with no DNA to determine background. Samples were incubated at 30°C for 90 min, diluted and translation of luciferase quantified by addition of the luciferase substrate luciferin (Promega), and light emitted read out on a luminometer (Monolight 3010, Pharmingen; San Diego, CA). Samples were prepared in duplicate and experiments replicated at least two to six times. Values are expressed as fold variant RNase activity over WT RNase activity.
In vitro cytotoxicity
All cell lines were maintained in RPMI 1640 (Mediatech, Herndon, VA) supplemented with 10% fetal bovine serum, 100 U/ml streptomycin and 100 µg/ml penicillin G (Gibco, Carlsbad, CA) in a 37°C humidified incubator with 5% CO2. The cytotoxicities of the mutant RNase and fusion proteins were assessed on K-562 cells, and Molt 3 and SEM K2 cells, respectively. Cells in complete media were plated at 104 cells/ml in 96-well plates and sterile filtered proteins were added to the appropriate concentrations. Cells were incubated for 45 h followed by the addition of Alamar Blue to 10% well volume and incubated an additional 5–9 h. Fluorescence was detected with a Cytofluor II plate reader (Biosearch; Bedford, MA) using wavelengths of 530 nm for excitation and 590 nm for emission. Each treatment was done in triplicate and experiments were replicated at least three times. Values are expressed as percent viable versus control cells treated with PBS ± 5% glycerol unless otherwise indicated.
Binding assays
SFv–RNase binding was measured by flow cytometry in blocking assays. CD7+ Molt 3 cells were incubated in PBS containing 2% fetal bovine serum and 0.02% sodium azide with varying amounts of sFv–RNase fusion proteins, BSA (negative control) or CD7-specific monoclonal antibody 3A1e (positive control). The epitope specificity of 3A1e is indistinguishable from that of 3A1f, the antibody from which the sFv is derived (Pauza et al., 1997). Therefore 3A1e and 3A1f block the binding of each other. Following incubation with the sFv–RNase fusion or control proteins for 30 min at 4°C, the Molt 3 cells were extensively washed in assay buffer and incubated in 100 µl with 5 µg (333 nM) fluorescein isothiocyanate (FITC)-conjugated 3A1e or isotype control antibodies for another 30 min at 4°C. The cells were washed again and analyzed immediately. Data were collected on a FACSCaliber (BD Biosciences, San Jose, CA) and analyzed with Flow Jo version 3.4 software (Tree Star Inc., Ashland, OR).
To assess internalization subsequent to binding, fusion proteins were labeled with AlexaFluor488 according to the manufacturer's protocol (Molecular Probes, Eugene, OR). Cells were incubated with 100 nM of labeled proteins for 30–45 min at 37°C or 4°C, washed extensively in PBS and fixed in 1% paraformaldehyde. Cells were then allowed to adhere to lysine-coated coverslips and washed again before mounting coverslips to slides using Vectashield (Vector Labs, Burlingame, CA). Cells were visualized on a Nikon Eclipse TE300 microscope (Nikon, Melville, NY).
Protein transfection
Cells were transfected with the proteins plus ProteoJuiceTM (Novagen, Darmstadt, Germany) according to the manufacturer's protocol. Transfections were performed in serum-free media. Three hours after exposure to the protein/ProteoJuiceTM complex, media containing 10% fetal bovine serum was added. Cells were incubated 44 h following transfection. Alamar Blue was used to determine cell metabolism as described above. Viability was compared to wild-type RNase 1/sFv fusion protein treated cells and the experiments replicated at least three times.
Disulfide conjugation to IgG
RNase 1 mutants or Onconase were cross-linked to anti-CD7 antibody 3A1e or control antibody mouse IgG (Pierce, Rockford, IL) with N-succinimidyl 3-(2-pyridyldithio) propionate (SPDP; Pierce). Briefly, RNases and antibodies were incubated with SPDP at a molar ratio of 1:4 for 1 h separately. Free SPDP was removed and the RNase samples were reduced with 5 mM DTT. After removal of residual DTT, the RNases and antibodies were mixed at a 1:1 molar concentration. Reactions were allowed to proceed overnight at room temperature. Products were purified via size exclusion chromatography on a HR 200 column (Pharmacia).
Statistics
The one-tailed Student's t-test was used to determine statistical significance.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
RNase 1 mutant design
Our initial goal was to generate active RNase 1 variants resistant to RI. Since the crystal structure of the RNase 1/human RI (hRI) complex has not been published, we identified RNase 1 residues that putatively contact hRI based on the crystal structure of bovine pancreatic RNase A/porcine RI (pRI) and the homology between RNase A and RNase 1 (Kobe and Deisenhofer, 1996). The sequence of RNase 1 and locations of mutated residues are shown in Figure 1 (these residues are illustrated in the structure of the RNase A/pRI complex in Supplementary Figure S1).
|
RNase A and RNase 1 share 70% amino acid identity; most of the other residues are conservative substitutions. A major exception lies in the more basic N-terminal region of RNase 1. This feature has made crystallization of RNase 1 difficult, necessitating deletion or mutation of the N-terminus to determine RNase 1 structure (Pous et al., 2000
, 2001). The N-terminal RNase 1 variants and RNase A are composed of three 
-helices and seven ß-strands with seven connecting loops (L) and are stabilized by four disulfide bonds. Gly88 in L6 of RNase A makes multiple contacts with pRI (Kobe and Deisenhofer, 1996). Because the ability of the corresponding residue (Gly89) in RNase 1 to contact RI is apparently dependent on flanking residues in L6 (Pous et al., 2000; Gaur et al., 2001; Leland et al., 2001), we generated the triple RNase 1 mutant N88R/G89D/R91D (RDD) to determine if these residues combined to confer gains in cytotoxicity due to enhanced RI resistance.
A second putative RI contact residue in RNase 1 targeted for mutation was Lys41 in L2. Lys41 of RNase A is an active site residue with nine close pRI contacts (Kobe and Deisenhofer, 1996). Altering Lys41 of RNase A by carboxymethylation reduced hRI binding by 88% relative to native RNase A (Blackburn and Gavilanes, 1980). Mutating Lys41 to Arg in G88R RNase A reduced catalytic activity relative to the single G88R mutant, but also resulted in a further reduction in RI inhibition. The decrease in RI inhibition outweighed the loss in catalytic activity and resulted in a 3-fold enhancement of cytotoxicity towards K-562 cells over G88R RNase A alone (Bretscher et al., 2000). Based on these studies and the conservation of Lys41 in RNase 1, we chose it as the second site to mutate.
Residues within RNase A L4 make multiple contacts with a four amino acid loop in pRI (Kobe and Deisenhofer, 1996). Kumar et al. (2004) recently demonstrated the importance of this interface region to RNase/RI binding by altering two residues each to Trp and deleting one residue in the 408–411 loop of hRI. These changes impeded docking of RNase A and RNase 1 to the mutant hRI as neither RNase was detectably inhibited. One of the residues in RNase A that contacts the 408–411 loop of pRI is Gln69 in L4 (Kobe and Deisenhofer, 1996). In an effort to reduce RI binding, we mutated Gln69 to Ala in RNase 1. Q69A was generated both as a single mutant and in combination with K41R in L2 (to disrupt two putative regions of the RNase 1/hRI interface).
The N-terminus of RNase A also contributes significantly to pRI binding. A hybrid protein consisting of the first 11 amino acids of RNase 1 substituted with the first 9 residues of Onconase showed a 100-fold reduction in hRI binding (Boix et al., 1996). In addition, truncating the first seven amino acids of RNase 1 inhibits its binding to hRI (Futami et al., 1995). Lys7 appears to be a key residue in the N-terminal region as it makes seven contacts with RI (Neumann and Hofsteenge, 1994; Kobe and Deisenhofer, 1996). Mutating Lys7 to Ala in RNase A lowered RI affinity 1000-fold and this mutation, in combination with G88R, increased cytotoxicity seven times over G88R RNase A alone (Haigis et al., 2002). Therefore, we generated K7A RNase 1 to assess its binding to hRI.
Protein yields
All recombinant proteins were expressed in E.coli as inclusion bodies, then denatured, refolded and purified by cation chromatography. The final yields of functional, refolded proteins for the RNase mutants and RNase–sFv fusions ranged from 20 to 30 and 1 to 2 mg/l culture, respectively. These amounts represent 
60% and 
10% of the initial inclusion body yields, respectively. The lower amounts of functional fusion proteins relative to the RNase mutants reflect both reduced expression levels and higher losses after refolding that occurred primarily during their concentration.
Enzyme activity
Enzymatic activity was quantified by measuring degradation of the substrate UpA (Table I). The single mutants K7A and Q69A showed no significant changes in catalytic activity versus wild-type RNase 1. K41R resulted in a 70-fold reduction in enzymatic activity, similar to both the 60- and 100-fold reductions reported for K41R RNase 1 and K41R RNase A activity, respectively (Messmore et al., 1995; Leland et al., 2001). K41R also reduced catalytic activity in the K41R/Q69A double mutant 65-fold versus the Q69A single mutant. In the K41R/Q69A/RDD quintuple mutant, K41R resulted in a 105-fold decrease in activity versus RDD.
|
In contrast, the triple RDD mutant exhibited a 2-fold increase in activity. Raines and co-workers reported a similar small increase in the activity of their ERDD RNase 1 mutant (Leland et al., 2001
). Surprisingly, the K7A mutation did not affect the enzymatic efficiency as a single mutant compared to wild type, but it negated the gain in efficiency achieved by the RDD mutation in K7A/RDD. RI-mediated inhibition of RNase 1 mutants
Inhibition of the mutants' enzymatic activity by RI was assessed by quantifying in vitro translation levels of luciferase mRNA in the presence of mutant or wild-type RNase and varying concentrations of RI. The single mutants K41R, Q69A and G89R, and the K41R/Q69A double mutant were equivalent to wild-type RNase 1 in terms of RI sensitivity (Figure 2 and Table II; data not shown). K7A exhibited a slight increase in activity in the presence of RI versus wild-type RNase 1 while RDD showed significant (P < 0.005) reduction in inhibition by RI (Figure 2). The K7A/RDD mutant was more resistant to RI than the triple RDD mutant alone, and the K41R/Q69A/RDD quintuple mutant exhibited even larger reductions in RI inhibition. There were dose-dependent responses of all RNase 1 variants to increasing concentrations of RI (data not shown). Thus, the mutants were still inhibited by RI but to a lower degree than wild-type RNase 1. Onconase served as a positive control in these assays. It was not significantly inhibited by RI at all concentrations tested (data not shown).
|
|
Cytotoxicity of RNase 1 mutants towards K-562 cells
RNase cytotoxicity in vitro was assessed using the human chronic myelogenous leukemic cell line, K-562. K-562 cytotoxicity induced by RNase 1 correlated with RI resistance. Only the K41R/Q69A/RDD and K7A/RDD mutants had IC50 values <10 µM (7.4 ± 1.8 and 6.7 ± 2.0 µM, respectively) (Figure 3). As expected, Onconase was the most cytotoxic with an IC50 value of 0.18 ± 0.14 µM. These data suggest the RI-resistant RNase 1 mutants could function as toxin moieties in ITs.
|
Properties of RNase 1–sFv ITs
In an attempt to increase the potency of the RI-resistant RNase 1 variants, we generated ITs by genetically fusing K41R/Q69A/RDD or K7A/RDD to an sFv fragment derived from the human CD7-specific antibody 3A1f. We previously showed that ricin toxin A chain linked to the same CD7-specific sFv fragment was specifically cytotoxic at low concentrations (IC50 value of 15 pM), consistent with the IT's efficient uptake via receptor-mediated endocytosis (Pauza et al., 1997). We reasoned RNase 1 would similarly be more cytotoxic if internalized specifically by CD7-mediated endocytosis rather than pinocytosis or some other non-specific mechanism. An Onconase–sFv fusion protein was also constructed as a positive control. A furin protease-sensitive site from Pseudomonas exotoxin A was introduced between the RNases and sFv, as it had been shown to increase potency in another RNase-based IT (Goyal and Batra, 2000). Separation of the sFv and toxin by furin-mediated cleavage of the spacer was thought to facilitate translocation of the toxin to the cytosol. The sFv was placed at the RNase C-terminus to minimize interaction with N-terminal catalytic residues (Figure 4A).
|
IT cytotoxicity was assessed on the CD7+ human T cell leukemia line, Molt 3. The CD7– leukemic cell line SEM K2 served as a negative control. Contrary to our expectations, the two RNase 1–sFv ITs did not kill Molt 3 cells (Figure 4B). Onconase-sFv was cytotoxic to Molt 3 cells but at equivalent levels to Onconase alone (Figure 4B). The positive control for these experiments was DA7, an IT composed of deglycosylated ricin toxin A chain disulfide-linked to an intact anti-CD7 antibody 3A1e (Vallera et al., 1996
). DA7 reduced Molt 3 cell viability to 3% at 5 nM while SEM K2 cells were not significantly affected (data not shown). These data indicate the Molt 3 cells are sensitive to a CD7-specific IT and suggest that either the RNase 1–sFv ITs are less potent than DA7 or they do not access the cytosol as efficiently. Another CD7+ T cell line, Jurkat, showed similar results (data not shown). The inability of the RNase 1–sFv ITs to kill cells was not due to the loss of function in either the targeting or toxin moieties in the fusion proteins. At 1 and 0.1 µM, the monovalent RDD- and K7A/RDD–sFv fusion proteins were equivalent to each other and the unconjugated divalent CD7-specific antibody 3A1e in their ability to block binding of 0.33 µM of FITC-conjugated 3A1e to Molt 3 cells (Figure S2A; Supplementary data). The ITs blocked binding less efficiently than the unconjugated antibody at the lowest concentration tested (0.01 µM). We speculate this reflects the higher avidity of the divalent antibody rather than reduced immunoreactivity of the ITs (owing to a high proportion of misfolded protein). However, a quantitative comparison between monovalent recombinant sFv and biochemically purified 3A1e Fab fragments is required to determine this directly. Regardless, the ITs bound CD7+ cells and were subsequently internalized (Figure S2B). Moreover, the enzymatic activity of each RNase 1 variant alone was equivalent to its activity when fused to the sFv fragment (data not shown). Together these data indicate both IT moieties were functional.
Because stability is a key factor for IT efficacy, we tested fusion protein integrity. The ITs were incubated at 37°C with media containing serum that was unconditioned or tumor cell-conditioned for 1.5 h before analysis in a far western assay using a CD7/IgG fusion protein probe. Neither RNase 1–sFv IT was detectably degraded (data not shown). As internalization of anti-CD7 antibodies and sFv fragments occurs rapidly (Pauza et al., 1997), our results suggest the fusion proteins were not degraded before uptake.
Protein transfection
To determine if the ITs could kill cells if artificially introduced into the cytosol, Onconase-, K7A/RDD-, RDD- and wild-type RNase 1–sFv fusion proteins were transfected into Molt 3 cells (Figure 5). K7A/RDD–sFv was chosen because the K7A/RDD mutant was the most cytotoxic towards K-562 cells (Table II). Transfected Onconase–sFv was consistently and significantly (P < 0.05) cytotoxic towards Molt 3 cells with an IC50 value of 50 nM. The two RNase 1 fusions showed increased cytotoxicity over wild-type RNase 1–sFv, but only the K7A/RDD–sFv was significantly (P < 0.05) different from wild-type RNase 1. This was consistent with its enhanced cytotoxicity towards K-562 cells relative to the RDD and wild-type RNase 1 (Table II). IC50 values for the RNase 1–sFv proteins were not determined because of limitations imposed by the transfection reagents (i.e. higher protein concentrations necessitated increasing the amount of transfection reagent to toxic levels).
|
These data suggest that although the resistance to RI conferred by the K7A and RDD mutations was detectable in our in vitro translation assay, it was insufficient to overcome inhibition in vivo. The data also indicated that Onconase was cytotoxic if it gained access to the cytosol, suggesting that the Onconase–sFv protein was not being efficiently routed to the cytosol upon internalization.
RNase disulfide conjugates
Onconase has been shown to be an effective IT when conjugated to an anti-CD22 antibody via a disulfide bond (Newton et al., 2001). Therefore we disulfide-linked Onconase, K7A/RDD and K41R/Q69A/RDD to a CD7-specific antibody. Of these conjugates, only Onconase was specifically and potently cytotoxic (Figure 6). It killed CD7+ Molt 3 cells but was not cytotoxic to SEM K2 CD7– cells. These data suggest efficient access to the cytosol and RI evasion are both necessary for RNase-based ITs to be effective as targeted therapeutics.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Our long-term goal is to develop humanized ITs for cancer therapy. As a first step towards achieving this goal, we constructed human RNase 1 variants that retained catalytic activity in the presence of the ubiquitous cytosolic inhibitor RI. Residues in four major regions of RNase 1 predicted to contact hRI were mutated in a stepwise fashion. These four regions were the N-terminal
1 helix, L2, L4 and L6. Single mutations were generally insufficient to evade RI; multiple RNase 1 residues had to be altered to generate mutants active in the presence of RI. CD7-specific ITs containing these RI-resistant RNase 1 variants, however, were not cytotoxic to CD7+ T-ALL lines. Transfection of the ITs directly into the cytosol reduced cell viability, suggesting their failure to kill cells when added externally resulted from the inability of the RNase to access the cytosol efficiently. Our results indicate appropriate intracellular routing and resistance to inhibition both are critical to the cytotoxicity of RNase 1-based ITs.
RI-resistant RNase 1 mutants must retain sufficient enzymatic activity to be cytotoxic. Futami et al. (1995) reported that although deletion of residues 1–7 of RNase 1 reduced RI inhibition, fusing this mutant to fibroblast growth factor did not increase cytotoxicity over wild-type RNase 1 fusions (Futami et al., 1999). The loss in catalytic activity apparently negated the gain resulting from RI evasion. Similarly, Onconase mutants with low catalytic activity have reduced cytotoxicity (Wu et al., 1993; Boix et al., 1996; Lee and Raines, 2003). Therefore, any RNase 1 mutation strategy must balance a gain in resistance to hRI with a potential loss in ribonucleolytic activity to yield a cytotoxic variant. Table II summarizes the characteristics of cytotoxic RNase 1 variants and highlights the importance of RI evasion for cytotoxicity.
This is especially true for the quintuple K41R/Q69A/RDD RNase 1 mutant where losses in catalytic activity were counterbalanced by gains in RI evasion. The enzymatic activity of K41R/Q69A/RDD RNase 1 was significantly less than RDD RNase 1. However, the quintuple mutant was more active than the triple mutant in the presence of RI and was correspondingly more cytotoxic towards K-562 cells. These data indicate the reduced catalytic activity and enhanced RI resistance introduced by the five mutations yielded a positive tradeoff that translated to increased cytotoxicity towards K-562 cells. It is likely that any changes to RI binding induced by K41R/Q69A alone were minimal as the K41R and Q69A single mutants, the K41R/Q69A double mutant and the wild-type protein were equivalently sensitive to RI in the in vitro translation assay, and none was cytotoxic. It appears that structural changes in L2/L4 owing to the K41R/Q69A mutations, combined with the RI binding strain induced by the RDD changes in L6, more profoundly impact RI binding to RNase 1 than either set of changes alone.
In contrast to the single K41R and Q69A mutants, the single K7A RNase 1 mutant had detectable cytotoxicity, though at a lower level than seen by Gaur et al. (2001). We suspect this results from assay differences as Gaur et al. (2001) found wild-type RNase 1 was also cytotoxic. The K7A mutation did not significantly alter catalytic activity relative to the wild type but it did enhance resistance to RI-mediated inhibition. This is consistent with structural data showing Lys7 of RNase A interacts with RI (Kobe and Deisenhofer, 1996).
In an effort to enhance their cytotoxicity, we linked the RNase 1 mutants to an anti-CD7 sFv targeting moiety to allow efficient internalization of RNase–sFv/CD7 complexes. CD7 appears to be internalized via coated pits upon ligand binding (Carriere et al., 1989). Correspondingly, fluorescently-labeled RNase–sFv proteins were internalized and accumulated in endosomal compartments upon incubation with CD7+ cells at 37°C while 4°C incubation resulted in only surface staining (Figure S2B; Supplementary data). These observations indicated that the fusion proteins bound the cell surface and were internalized. However, they were no more cytotoxic than RNase alone. Their lack of cytotoxicity was not due to decreased RNase enzymatic activity or protein degradation. Together these data suggested that the RNase moieties were not being efficiently routed from the endosomes to the cytosol subsequent to internalization.
To test this hypothesis, the ITs were introduced into the cytosol directly via transfection. Transfected Onconase–sFv was now cytotoxic (IC50 value of 50 nM) and mutant RNase 1 fusions were more cytotoxic than wild-type RNase 1/sFv (Figure 5). However, the transfected RNase 1 fusions were not as consistently cytotoxic as Onconase fusions, and only the K7A/RDD RNase 1–sFv was significantly different from the wild-type RNase 1–sFv. This was expected because K7A/RDD RNase 1 was more active in the presence of RI than RDD (Table II).
ITs composed of native Onconase disulfide-linked to an anti-CD22 antibody are specifically cytotoxic, suggesting that reduction upon internalization allows Onconase to access the cytosol (Newton et al., 2001). We therefore asked if the cytotoxicity of K7A/RDD could be enhanced if it was disulfide-linked to 3A1e, a CD7-specific antibody, rather than fused directly to a CD7-specific sFv fragment. An Onconase-3A1e conjugate was specifically cytotoxic towards Molt 3 cells with an IC50 value of 2 nM. Onconase alone has an IC50 value of 2 µM with Molt 3 cells; therefore, antibody conjugation resulted in nearly a 1000-fold increase in potency. In contrast, RNase 1-3A1e disulfide-linked conjugates were not cytotoxic at any concentration used, even at 100-fold higher concentrations relative to the Onconase conjugate. These data suggest either the RNase 1 proteins did not efficiently access the cytosol or if they did, they were inhibited by RI. Given that K7A/RDD–sFv was cytotoxic at 50 nM in the transfection assay, it seems more likely that the proteins were not efficiently accessing the cytosol.
RNase 1 cytotoxicity may be limited by lysosomal degradation. Endocytosed RNase A and RNase 1 variants accumulate in the lysosomal compartment (McElligott et al., 1985; Bosch et al., 2004), most probably owing to binding to a lysosomal receptor (Cuervo and Dice, 1996). Perhaps a portion of the RNase 1 from the fusions was targeted to the lysosomal compartment directly or indirectly from the cytosol upon translocation. Lysosomal routing issues have been examined with other ITs. Manske et al. (1989) found variability in cytotoxicity between two CD5 targeted ITs was due to differences in the kinetics of the ITs routing to lysosomes.
RNase 1 can generate effective fusion protein ITs in other systems. Recently, it was genetically fused to an anti-ErbB2 sFv (De Lorenzo et al., 2004). This construct had an IC50 value of 12.5 nM on SKBR3 cells, but it should be noted that the sFv alone induces death of these cells with an IC50 value of 200 nM (De Lorenzo et al., 2002). Interestingly, Onconase augments the effects of several cytotoxic agents (Mikulski et al.,1990, 1992; Ryback et al., 1996; Deptala et al.,1998; Lee et al., 2003). Onconase potentiation may be due to specific inhibition of survival pathways (Deptala et al., 1998). It is possible a similar phenomenon may be at play with the RNase 1/anti-ErbB2 sFv construct.
Cytotoxic RNase 1 fusions with epidermal growth factor have been generated with IC50 values of 0.25 and 3 µM on A431 and TE-1 cells, respectively (Psarras et al., 1998). However, the epidermal growth factor alone was toxic to the A431 cell line used in this assays, arguing that these results should be interpreted cautiously (Filmus et al., 1985; Kottke et al., 1999). Cytotoxic RNase 1 fusions with IL-2 and fibroblast growth factor have also been produced (Futami et al., 1999; Psarras et al., 2000). IL-2 and fibroblast growth factor are uniquely internalized. IL-2 receptor undergoes clathrin-independent endocytosis while fibroblast growth factor binds cell surface heparan sulfate proteoglycan and can be routed to the nucleus which does not contain RI (Hawker and Granger, 1991; Woodward et al., 1992; Amalric et al., 1994; Ferriani et al., 1994; Lamaze et al., 2001). In contrast, CD7 appears to undergo clathrin-mediated endocytosis (Carriere et al., 1989). Therefore differences in internalization pathways may explain the variability in potency of RNase 1 fusion proteins. Together these data suggest our RI-resistant RNase 1 variants will be useful as IT components when linked to targeting moieties that permit high enough concentrations of RNase to reach the cytosol.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
We thank John Kersey for the SEM-K2 cell line, John Lipscomb for helpful discussions regarding the enzyme kinetic assays, and Daniel Vallera for DA7 and the G89R RNase 1 construct. This work was supported in part by an NIH training grant (T32 CA09138) to H.A.E. and an NIH grant (R01 CA82766) to C.A.P.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Amalric,F., Bouche,G., Bonnet,H., Brethenou,P., Roman,A.M., Truchet,I. and Quarto,N. (1994) Biochem. Pharmacol., 47, 111–115.[CrossRef][ISI][Medline]
Baluna,R. and Vitetta,E.S. (1997) Immunopharmacology, 3, 117–132.
Barcena,A., Muench,M.O., Roncarolo,M.G. and Spits,H. (1995) Leuk. Lymphoma, 17, 1–11.[ISI][Medline]
Blackburn,P. and Gavilanes,J.G. (1980) J. Biol. Chem., 255, 10959–10965.
Boix,E., Wu,Y., Vasandani,V.M., Saxena,S.K., Ardelt,W., Ladner,J. and Youle,R.J. (1996) J. Mol. Biol., 257, 992–1007.[CrossRef][ISI][Medline]
Bosch,M., Benito,A., Ribo,M., Puig,T., Beaumelle,B. and Vilanova,M. (2004) Biochemistry, 43, 2167–2177.[CrossRef][Medline]
Bretscher,L.E., Abel,R.L. and Raines,R.T. (2000) J. Biol. Chem., 275, 9893–9896.
Carriere,D., Arcier,J.M., Derocq,J.M., Fontaine,C. and Richer,G. (1989) Exp. Cell Res., 182, 114–128.[CrossRef][ISI][Medline]
Cinatl,J. Jr, Cinatl,J., Kotchetkov,R., Vogel,J.U., Woodcock,B.G., Matousek,J., Pouckova,P., Kornhuber,B. (1999) Int. J. Oncol., 15, 1001–1009.[ISI][Medline]
Cuervo,A.M. and Dice,J.F. (1996) Science, 273, 501–503.[Abstract]
D'Alessio,G., Di Donato,A., Parente,A. and Piccoli,R. (1991) Trends Biochem. Sci., 16, 104–106.[CrossRef][ISI][Medline]
De Lorenzo,C., Palmer,D.B., Piccoli,R., Ritter,M.A. and D'Alessio,G. (2002) Clin. Cancer Res., 8, 1710–1719.
De Lorenzo,C., Arciello,A., Cozzolino,R., Palmer,D.B., Laccetti,P., Piccoli,R. and D'Alessio,G. (2004) Cancer Res., 64, 4870–4874.
Deptala,A., Halicka,H.D., Ardelt,B., Ardelt,W., Mikulski,S.M., Shogen,K. and Darzynkiewicz,Z. (1998) Int. J. Oncol., 13, 11–16.[ISI][Medline]
Erickson,H.A. and Pennell,C.A. (2002) Immunol. Res., 25, 177–191.[CrossRef][ISI][Medline]
Ferriani,R.A., Ahmed,A., Sharkey,A. and Smith,S.K. (1994) Growth Factors, 10, 259–268.[ISI][Medline]
Fett,J.W., Strydom,D.J., Lobb,R.R., Alderman,E.M., Bethune,J.L., Riordan,J.F. and Vallee,B.L. (1985) Biochemistry, 24, 5480–5486.[CrossRef][Medline]
Filmus,J., Pollak,M.N., Cailleau,R. and Buick,R.N. (1985) Biochem. Biophys. Res. Commun., 128, 898–905.[CrossRef][ISI][Medline]
Flavell,D.J., Cooper,S., Morland,B. and Flavell,S.U. (1991) Br. J. Cancer, 64, 274.[Medline]
Futami,J., Seno,M., Kosaka,M., Tada,H., Seno,S. and Yamada,H.. (1995) Biochem. Biophys. Res. Comm., 216, 406–413.[CrossRef][ISI][Medline]
Futami,J., Seno,M., Ueda,M., Tada,H. and Yamada,H. (1999) Protein Eng., 12, 1013–1019.
Frankel,A.E., Kreitman,R.J. and Sausville,E.A. (2000) Clinical Cancer Res., 6, 326–334.
Gaur,D., Swaminathan,S. and Batra,J.K. (2001) J. Biol. Chem., 276, 24978–24984.
Goyal,A. and Batra,J.K. (2000) Biochem. J., 345, 247–254.
Haigis,M.C., Kurten,E.L., Abel,R.L. and Raines,R.T. (2002) J. Biol. Chem., 277, 11576–11581.
Hawker,J.R. and Granger,H.J. (1991) Ann. N Y Acad. Sci., 638, 424–426.[ISI][Medline]
Hooper,L.V., Stappenbeck,T.S., Hong,C.V. and Gordon,J.I. (2003) Nat. Immunol., 4, 269–273.[CrossRef][ISI][Medline]
Kim,J.S., Soucek,J., Matousek,J. and Raines,R.T. (1995a) J. Biol. Chem., 270, 31097–31102.
Kim,J.S., Soucek,J., Matousek,J. and Raines,R.T. (1995b) J. Biol. Chem., 270, 10525–10530.
Kobe,B. and Deisenhofer,J. (1996) J. Mol. Biol., 264, 1028–1043.[CrossRef][ISI][Medline]
Kotchetkov,R., Cinatl,J., Matousek,J., Vogel,J., Pouckova,P., Wagner,M., Kornhuber,B., Schwabe,D. and Cinatl,J.Jr (2000) Oncol. Rep., 7, 363–367.[ISI][Medline]
Kottke,T.J., Blajeski,A.L., Martins,L.M., Mesner,P.W.Jr, Davidson,N.E., Earnshaw,W.C., Armstrong,D.K. and Kaufmann,S.H. (1999) J. Biol. Chem., 274, 15927–15936.
Kumar,K., Brady,M. and Shapiro,R. (2004) Proc. Natl. Acad. Sci. USA, 101, 53–58.
Lamaze,C., Dujeancourt,A., Baba,T., Lo,C.G., Benmerah,A. and Dautry-Varsat,A. (2001) Mol. Cell, 7, 661–671.[CrossRef][ISI][Medline]
Lee,F.S. and Vallee,B.L. (1989) Biochem. Biophys. Res. Commun., 160, 115–120.[CrossRef][ISI][Medline]
Lee,J.E. and Raines,R.T. (2003) Biochemistry, 42, 11443–11450.[CrossRef][Medline]
Lee,I., Lee,Y.H., Mikulski,S.M. and Shogen,K. (2003) Adv. Exp. Med. Biol., 530, 187–196.[ISI][Medline]
Leland,P.A., Staniszewski,K.E., Kim,B.M. and Raines,R.T. (2001) J. Biol. Chem., 276, 43095–43102.
Leta,E., Roy,A.K., Hou,Z. and Jung,L.K. (1995) Cell. Immunol., 165, 101–109.[CrossRef][ISI][Medline]
Liao,Y.D., Huang,H.C., Chan,H.J. and Kuo,S.J. (1996) Prot. Exp. Purif., 7, 194–202.[CrossRef][ISI][Medline]
Manske,J.M., Buchsbaum,D.J. and Vallera,D.A. (1989) J. Immunol., 142, 1755–1766.[Abstract]
McElligott,M.A., Miao,P. and Dice,J.F. (1985) J. Biol. Chem., 260, 11986–11993.
Messmore,J.M., Fuchs,D.N. and Raines,R.T. (1985) J. Am. Chem. Soc., 117, 8057–8060.[CrossRef]
Mikulski,S.M., Viera,A., Ardelt,W., Menduke,H. and Shogen,K. (1990) Cell Tissue Kinetics, 23, 237–246.[Medline]
Mikulski,S.M., Viera,A., Darzynkiewicz,Z. and Shogen,K. (1992) Br. J. Cancer, 66, 304–310.[ISI][Medline]
Mikulski,S.M, Grossman,A.M., Carter,P.W., Shogen,K. and Costanzi,J.J. (1993) Int. J. Oncol., 3, 57–64.
Mikulski,S.M. et al. (2002) J. Clin. Oncol., 20, 274–281.
Neumann,U. and Hofsteenge,J. (1994) Protein Sci., 3, 248–256.[Abstract]
Newton,D.L., Boque,L., Wlodawer,A., Huang,C.Y. and Rybak,S.M. (1998) Biochemistry, 37, 5173–5183.[CrossRef][Medline]
Newton,D.L., Hansen,H.J., Mikulski,S.M., Goldenberg,D.M. and Rybak,S.M. (2001) Blood, 97, 528–535.
Nitta,K., Ozaki,K., Ishikawa,M., Furusawa,S., Hosono,M., Kawauchi,H., Sasaki,K., Takayanagi,Y., Tsuiki,S. and Hakomori,S. (1994) Cancer Res., 54, 920–927.
Onda,M., Willingham,M., Wang,Q.C., Kreitman,R.J., Tsutsumi,Y., Nagata,S. and Pastan,I. (2000) J. Immunol., 165, 7150–7156.
Pace,K.E., Hahn,H.P., Pang,M., Nguyen,J.T. and Baum,L.G. (2000) J. Immunol., 165, 2331–2334.
Pauza,M.E., Doumbia,S.O. and Pennell,C.A. (1997) J. Immunol., 158, 3259–3269.[Abstract]
Pous,J., Canals,A., Terzyan,S.S., Guasch,A., Benito,A., Ribo,M., Vilanova,M. and Coll,M. (2000) J. Mol. Biol., 303, 49–60.[CrossRef][ISI][Medline]
Pous,J. et al. (2001) Acta Crystallogr. D Biol. Crystallogr., 57, 498–505.[CrossRef][Medline]
Psarras,K., Ueda,M., Yamamura,T., Ozawa,S., Kitajima,M., Aiso,S., Komatsu,S. and Seno,M. (1998) Protein Eng., 11, 1285–1292.
Psarras,K., Ueda,M., Tanabe,M., Kitajima,M., Aiso,S., Komatsu,S. and Seno,M, (2000) Cytokine, 12, 786–790.[CrossRef][ISI][Medline]
Rybak,S.M. et al. (1996) J. Nat. Canc. Inst., 88, 747–753.
Sempowski,G.D., Lee,D.M., Kaufman,R.E. and Haynes,B.F. (1999) Crit. Rev. Immunol., 19, 331–348.[ISI][Medline]
Shimizu,Y., van Seventer,G.A., Ennis,E., Newman,W., Horgan,K.J. and Shaw,S. (1992) J. Exp. Med., 175, 577–582.
Tonevitsky,A.G., Agapov,I.I., Mechetner,E.B., Ershova,G.V., Toptygin,A.Y., Shamshiev,A.T. and Pfueller,U. (1993) Biochem. Mol. Bid. Inl., 31, 1059.
Vallera,D.A., Burns,L.J., Frankel,A.E., Sicheneder,A.R., Gunther,R., Gajl-Peczalska,K., Pennell,C.A. and Kersey,J.H. (1996) J. Immunol. Meth., 197, 69–83.[CrossRef][Medline]
Vicentini,A.M., Kieffer,B., Matthies,R., Meyhack,B., Hemmings,B.A., Stone,S.R. and Hofsteenge,J. (1990) Biochemistry, 29, 8827–8834.[CrossRef][Medline]
Witzel,H. and Barnard,E.A. (1962) Biochem. Biophys. Res. Commun., 7, 295–299.[CrossRef][ISI][Medline]
Woodward,W.R., Nishi,R., Meshul,C.K., Williams,T.E., Coulombe,M. and Eckenstein,F.P. (1992) J. Neurosci., 12, 142–152.[Abstract]
Wu,Y., Mikulski,S.M., Ardelt,W., Rybak,S.M. and Youle,R.J. (1993) J. Biol. Chem., 268, 10686–10693.
Received July 6, 2005; revised September 9, 2005; accepted September 13, 2005.
Edited by Sally Ward
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Wang, J. Zhao, J.-L. Ren, L. Zhang, W.-H. Wen, R. Zhang, W.-W. Qin, L.-T. Jia, L.-B. Yao, Y.-Q. Zhang, et al. Recombinant Immunoproapoptotic Proteins with Furin Site Can Translocate and Kill HER2-Positive Cancer Cells Cancer Res., December 15, 2007; 67(24): 11830 - 11839. [Abstract] [Full Text] [PDF] |