Characterization of Single Site Ricin Toxin B Chain Mutants

Bioconjugate Chem. 1996, 7, 30−37

30

Characterization of Single Site Ricin Toxin B Chain Mutants Arthur Frankel,*,† Edward Tagge,‡ John Chandler,‡ Chris Burbage,† Greg Hancock,† Joseph Vesely,§ and Mark Willingham§ Departments of Medicine, Surgery and Pathology, Medical University of South Carolina, Charleston, South Carolina 29425. Received April 5, 1995X

DNA encoding ricin B chain was modified by site-directed mutagenesis, and eight separate mutant RTB cDNAs including four novel mutants were ligated into the baculovirus transfer vector, pAcGP67A. Cotransfection of S. frugiperda Sf9 cells with BaculoGold DNA was followed by limiting dilution isolation of recombinant baculoviruses. Infection of Sf9 cells at a multiplicity of infection of 5 in the presence of 25 mM lactose produced 0.05-1 mg/L of soluble, glycosylated 34 kDa proteins immunoreactive with monoclonal and polyclonal antibodies to ricin B chain. Mutant ricin B chains were partially purified by monoclonal antibody immunoaffinity chromatography to 10-50% purity in near milligram quantities. The mutant ricin B chains had decreased lectin binding relative to plant ricin B chain as measured by binding to immobilized lactose and asialofetuin and cell binding immunofluorescence. The mutant ricin B chains reassociated with plant RTA similarly to plant RTB, and the recombinant heterodimers had slightly reduced cell cytotoxicity relative to ricin.

INTRODUCTION

Ricin toxin is a 65 kDa glycoprotein isolated from the seeds of the castor bean plant, Ricinus communis (1). Ricin consists of an RNA N-glycosidase A chain (RTA)1 disulfide linked to a galactose-specific lectin B chain (RTB). RTB binds the toxin to mammalian cell surface glycoproteins. Ricin is internalized by receptor-mediated endocytosis (2). Internalized toxin is transported to the TR (trans-reticular) Golgi, where further galactosespecific binding facilitates ricin routing to a specific organelle from which RTA translocates to the cytosol and enzymatically inactivates protein synthesis (3, 4). Thus, the lectin function of RTB may be critical for three steps in ricin intoxication (binding, internalization, and intracellular routing). Because of the extreme potency of this toxin, the available extensive biochemical and structural knowledge, and its importance as a component in a number of immunotoxins currently in clinical trials for leukemias and lymphomas, we have been interested in better characterizing its protein-carbohydrate interactions. RTB binds β-D-galactopyranose moieties on cell surface glycolipids and glycoproteins. The hydroxyl groups at positions 3, 4, and 6 of the terminal galactose interact with the protein (5). Affinities for galactose and lactose are approximately 104 M-1, while association constants for complex oligosaccharides and cell surfaces are 3 and 4 logs higher, respectively (6, 7). These results suggest that multiple low affinity sugar binding sites on ricin interact with complex oligosaccharides and cells to yield high affinity binding. * To whom correspondence should be addressed at: Hollings Cancer Center Room 306, Medical University of South Carolina, 171 Ashley Avenue, Charleston, SC 29425. Tel: (803) 792-1450. Fax: (803) 792-3200. † Department of Medicine. ‡ Department of Surgery. § Department of Pathology. X Abstract published in Advance ACS Abstracts, October 1, 1995. 1 Abbreviations: RTA, ricin toxin A chain; RTB, ricin toxin B chain; SPR, surface plasmon resonance.

1043-1802/96/2907-0030$12.00/0

Equilibrium dialysis studies with plant ricin and plant RTB identified two galactose binding sites with either similar or disparate affinities (KA’s ) 103-104 M-1) (8, 9). X-ray diffraction analysis of ricin crystals soaked with low concentrations of R-lactose (5 mM) revealed two lectin sites in the 1R and 2γ subdomains (10). Amino acid residues interacting with the sugars included Asp-22, Trp-37, Gln-35, and Asn-46 in the 1R subdomain and Asp-234, Tyr-248, and Asn-255 in the 2γ subdomain. Only a limited number of mutant RTBs have been reported. In one study, modification of an amino acid residue (Asn-255) in one subdomain (2γ) produced complete loss of sugar binding (11). In another study, mutations of residues in two subdomains (1R and 2γ) were required to abolish sugar binding (12). Thus, results of mutational analysis to date are conflicting. Further, several lines of evidence support three lectin binding sites in ricin. Three distinct sites on ricin were cross-linked by radiolabeled fetuin glycopeptides containing a dichlorotriazine-activated 6-(N-methylamino)-6deoxy-D-galactose moiety (13). The X-ray crystal structure of RTB shows six homologous subdomains with similar folding and primary amino acid sequence that resemble the primitive galactose binding fold in discoidin I from the slime mold Dictyostelium discoideum (14). N-Bromosuccinimide modification of Trp-37 reduced sugar binding, demonstrating a sugar binding site in subdomain 1R (15). N-Acetylimidazole O-acetylation of two tyrosines reduced sugar binding, implicating two additional binding sites in subdomains 1β (Tyr-78) and 2γ (Tyr-248) (16). The disparity between the biochemical/biophysical observations and mutational analyses can be attributed to the previous low level expression of the mutants. Decreased sugar binding due to misfolding or aggregation of recombinant RTBs leads to overestimation of the effect of individual modifications. A major limitation of these previous studies was that they did not report purification or quantitative immunological characterization of the mutants prior to tests of sugar binding. We have previously reported the expression and partial purification of milligram quantities of wild-type RTB from insect cells using recombinant baculovirus (17). The © 1996 American Chemical Society

Ricin Toxin B Chain Mutants

Bioconjugate Chem., Vol. 7, No. 1, 1996 31

Figure 1. Model of lectin binding sites of RTB. Diagram of association of lactose with amino acid side chains of RTB in subdomains 1R and 2γ. Aromatic residues Trp-37 and Tyr-248 interact with the apolar face of terminal galactose. Polar residues Asp-22, Gln-35, Lys-40, Asn-46, Asp-234, and Asn-255 provide hydrogen bonds to hydroxyls of the sugar. Derived from Rutenber and Robertus (10). Wire diagram display of Brookhaven coordinates on SYBYL software on Silicon Graphics Iris Indigo workstation. Stereoviews.

insect RTB had immunologic and biologic properties similar to plant RTB. We have combined this methodology of large scale baculovirus expression of RTB with sitespecific mutagenesis to create a more extensive panel of mutants than previously evaluated. Unique modifications at both the 1R subdomain (W37S and Q35N) and the 2γ subdomain (Y248S and D234E/A237R) were made based on the X-ray crystallographic models of the lectin binding sites (Figure 1). Other modifications were made to test previously constructed mutants in this system (K40M, K40M/N46G, N255A, and N255G). The modifications either changed the aromatic ring residues which provide van der Waals interactions between the protein and sugar (Trp-37 and Tyr-248) or key polar residues which provide hydrogen bonds to sugar hydroxyls (Gln35, Lys-40, Asn-46, Asp-234, and Asn-255). We have added both a purification step and a more detailed analysis of the affinity, cell binding, and intoxication properties of mutant proteins to provide an in-depth quantitative study of the ricin lectin binding sites. MATERIAL AND METHODS

Materials. Restriction endonucleases and T4 ligase were obtained from Promega (Madison, WI). Isotopes and the Sculptor in vitro mutagenesis system were obtained from Amersham (Arlington Heights, IL). Polyclonal antibodies and chemicals were from Sigma (St. Louis, MO). EX-CELL400 medium was obtained from JRH Scientific (Lexena, KS). Sf9 insect cells, TMNFH medium, BaculoGold DNA, and pAcGP67A transfer vector were from PharMingen (San Diego, CA). DNA and

plasmid prep kits were obtained from BioRad (Hercules, CA). The Sequenase kit for dideoxy sequencing was obtained from USB (Cleveland, OH). The Random Primer labeling kit and M13K07 phage were obtained from Stratagene (La Jolla, CA). Purified P2, P8, and P10 murine monoclonal antibodies to RTB and purified RBR12 murine monoclonal antibody to RTA were gifts of Dr. Walter Blattler, ImmunoGen (Cambridge, MA). Sera and media were obtained from GIBCO BRL (Grand Island, NY). 3M Emphaze Biosupport medium AB1 azlactone functionality bis-acrylamide and lactosyl acrylamide were obtained from Pierce (Rockford, IL). Triethylamine was obtained from Aldrich (Milwaukee, WI). The alkaline phosphatase Vectastain kit for Western blots was obtained from Vector Laboratories (Burlingame, CA). Ninety-six-well plates were from Costar (Cambridge, MA). Plant RTB, ricin, and RTA were obtained from Inland Laboratories (Austin, TX). Rhodamine conjugated goat anti-(rabbit Ig) was from ImmunoResearch (Bangor, ME). Construction of Transfer Vectors Encoding Mutant RTBs. pAcGP67A-RTB DNA was prepared using the alkaline lysis method and silica-based matrix purification and then restricted with BamHI and EcoRI. The RTB encoding DNA fragment was ligated to BamHI and EcoRI restricted pUC119 plasmid and used to transform INVRF′ E. coli cells. Single-stranded DNA was produced by infection of transformants with M13K07 phage as previously described. Oligonucleotides were synthesized on an Applied Biosystems 380B DNA synthesizer and desalted with butan-1-ol. 39-mers were prepared with

32 Bioconjugate Chem., Vol. 7, No. 1, 1996

the modified codon flanked by 18 bases on each side matching RTB sequence. Site-specific mutagenesis was performed by the Eckstein method using the Amersham in vitro mutagenesis kit and manufacturer’s instructions (18). Sequences of mutant RTB DNAs were confirmed by double-stranded dideoxy sequencing by the Sanger method using the Sequenase kit (19). Mutant RTB encoding pUC119 DNAs were then restricted with BamHI, and EcoRI and the RTB encoding fragments were subcloned into pAcGP67A plasmid and used to transform INVRF′ cells. Transfer vectors with mutant RTBs were then purified by cesium chloride density-gradient centrifugation. Isolation of Recombinant Baculoviruses. The Sf9 S. frugiperda ovarian cell line was maintained on TMNFH medium supplemented with 10% fetal calf serum. pAcGP67A-mutant RTB DNAs (4 ug) were cotransfected with 0.5 ug of BaculoGold AcNPV DNA into 2 × 106 Sf9 insect cells following the recommendations of the supplier. On day 5 posttransfection, media were centrifuged and the supernatants tested in limiting dilution assays with Sf9 cells. A total of 2 × 104 Sf9 cells were incubated with 10-fold dilutions of supernatants in 96-well plates. Eight days postinfection, supernatants were saved, cells in each assay well were lysed with NaOH, and the lysates were transferred to nitrocellulose. The nitrocellulose was then blocked with Blotto and reacted with random primer 32P-dCTP-labeled RTB DNA. After hybridization for 16 h at 67 °C, the dot blot membranes were washed with 0.1 x (150 mM NaCl/15 mM sodium citrate)/1% SDS, dried, and exposed to X-ray film. Positive wells were identified and supernatants reassayed by limiting dilution until all wells up to 10-7 dilution were positive. Two rounds of selection were required for each mutant. Recombinant viruses in the supernatants were then amplified by infecting Sf9 cells at a multiplicity of infection (moi) of 0.1, followed by collection of day 7 supernatants. Expression of Mutant B Chains in Sf9 Cells. Recombinant baculoviruses were used to infect 2 × 108 Sf9 cells at a moi of 5 in EX-CELL400 media with 50 mM R-lactose in spinner flasks. Media supernatants containing mutant RTBs were collected day 6 postinfection. Purification of Mutant RTBs. Media supernatants were adjusted to 0.01% sodium azide and maintained through all purification steps at 4 °C. The supernatants were ultracentrifuged at 100 000g for 1 h. The final supernatants were then loaded onto a P2 monoclonal antibody-acrylamide column previously described (17). The affinity column was then washed sequentially with NTEAL and 500 mM NaCl, 25 mM Tris pH 9, 1 mM EDTA, 0.01% sodium azide, 25 mM R-lactose, and 0.1% Tween-20 (NTEALT), and mutant RTBs were eluted with 0.1 M triethylamine pH 11. The alkaline eluants were immediately neutralized with 1 M sodium phosphate pH 4.8 and stored at -20 °C until assayed. Optical densities at 280 nm were determined and aliquots mixed with reducing 2 x SDS sample buffer, boiled 4 min, electrophoresed on a 15% SDS-PAGE, stained with Coomassie Blue R-250, and destained with acetic acid/methanol. Gels were scanned on an IBAS automatic image analysis system to estimate the fraction of protein of molecular weight 30 kDa. Immunological Properties of Mutant RTBs. Aliquots of mutant RTB, plant RTB, wild-type recombinant RTB and prestained low molecular weight standards were mixed with reducing 2 x SDS sample buffer, boiled for 4 min loaded on a 15% SDS/PAGE, and electrophoresed for 90 min. Gels, Whatman 3M #1 paper, and

Frankel et al.

nitrocellulose were equilibrated for 15 min in Towbin buffer (20 mM Tris/0.1 M glycine/20% methanol) and placed in a Semi-dry Trans-blot cell (BioRad). After electrophoresis at 15 V for 20 min, the nitrocellulose was blocked with 10% Carnation's nonfat dry milk/0.1% BSA/ 0.1% Tween 20/0.02% sodium azide. The blots were then washed with PBS plus 0.05% Tween 20 and PBS, incubated with rabbit anti-ricin antibody at 1:400 in PBS plus 0.5% BSA plus 0.01% sodium azide for 1 h, washed again, incubated with alkaline phosphatase conjugated goat anti-(rabbit IgG) at 1:1000 in PBS plus 0.5% BSA plus 0.01% sodium azide for 1 h, washed again, and developed with the Vectastain alkaline phosphatase kit, following the manufacturer's recommendations. Blots were scanned as above to compare 30 kDa Mr band intensities. Monoclonal antibody P2, P8, or P10 (100 ul) at 5 µg/ mL in PBS was incubated in Costar EIA microtiter wells overnight at 4 °C. Samples of plant RTB, wild-type recombinant RTB, and mutant RTBs were treated for 20 min at room temperature with 5% β-mercaptoethanol and then dilutions made in 50 mM NaCl, 25 mM Tris pH 8, 1 mM EDTA. The antibody-coated microtiter wells were then washed with PBS plus 0.1% Tween 20, blocked with 3% BSA/PBS/0.01% sodium azide, rewashed and incubated with dilutions of the reduced RTB samples, rewashed and incubated with rabbit anti-ricin antibody 1:400 in PBS + 0.5% BSA + 0.01% sodium azide, washed again, incubated with alkaline phosphatase conjugated goat anti-(rabbit IgG) at 1:1000 in PBS + 0.5% BSA + 0.01% sodium azide, washed and developed with pnitrophenyl phosphate at 1 mg/mL in 50 mM diethanolamine buffer pH 9.8, and read on a BioRad 450 Microplate reader at 405 nm. For each experiment 12 different concentrations of plant RTB and recombinant RTBs were tested. A plot of absorbance versus dilution was made for plant RTB and recombinant proteins. Dilutions yielding half-maximal binding were used to calculate concentrations. Lectin Activity of Mutant RTBs. 100 µL volumes of 1 µg/mL asialofetuin in PBS were added to wells of a Costar EIA plate and incubated overnight at 4 °C. Samples of plant RTB, wild-type recombinant RTB, or mutant RTBs in EX-CELL 400 were exposed to 5% β-mercaptoethanol for 20 min at room temperature to remove homodimers and dilutions made in EX-CELL 400 medium with or without 20 µg/mL asialofetuin or 100 mM R-lactose. The asialofetuin coated microtiter wells were then washed with PBS/0.1% Tween-20, blocked with 3% BSA/PBS/0.01% sodium azide, and rewashed. The dilutions of various reduced RTBs were added to wells for 1 h and then removed and the wells washed again. Rabbit anti-ricin antibody was added (1:400 dilution in 0.5% BSA/PBS/0.01% sodium azide) for 1 h; the wells were washed again; alkaline phosphatase conjugated goat anti-(rabbit IgG) (1:5000 in 0.5% BSA/PBS/0.01% sodium azide) incubated in the wells; finally, the wells were washed, reacted with 1 mg/mL of p-nitrophenyl phosphate in 50 mM diethanolamine buffer pH 9.6, and measured in a microtiter plate reader at 405 nm. In each experiment, 12 different concentrations of plant RTB and recombinant protein were tested. As in the antibody elisa, relative reactivity to plant RTB was calculated from concentrations giving half-maximal binding. The effects of 20 µg/mL asialofetuin or 100 mM R-lactose on half-maximal binding were calculated for plant RTB, wild-type recombinant RTB, and mutant RTBs. Samples of freshly reduced plant RTB, recombinant RTB, or mutant RTB were diluted in 50 mM NaCl/25 mM

Ricin Toxin B Chain Mutants

Tris pH 8/1 mM EDTA/0.01% sodium azide (NTEA), loaded on a lactosyl acrylamide marix, and washed with NTEA, and lactose binding protein was eluted with NTEA plus 50 mM lactose. Fractions were assayed for absorbance at 280 nm and asialofetuin binding as described above. KB cells were washed with PBS and attached to polylysine-coated tissue-culture dishes and centrifuged at 2000g for 10 min. The cells were then incubated live at 4 °C. The cells were washed with 2 mg/mL of BSA in PBS and incubated in PBS plus BSA with or without 100 µg/mL of asialofetuin and with 1 µg/mL of freshly reduced plant RTB, recombinant wild-type RTB or mutant RTB. The incubation was done at 4 °C. The cells were then washed with PBS and incubated with rabbit anti-ricin antibody at 1:400 in PBS plus BSA for 30 min at 4 °C. The cells were then washed with PBS and reacted with goat anti-(rabbit Ig) conjugated to rhodamine at 25 µg/ mL for 30 min at 4 °C. The cells were washed again in PBS and fixed in 3.7% formaldehyde in PBS, mounted under a #1 coverslip in glycerol-PBS (90:10), and examined under a Zeiss Axioplan epifluorescence microscope. Reassociation of Mutant RTB with Plant RTA To Form Heterodimers. Thirty µg portions of plant RTB, wild-type recombinant RTB, and mutant RTBs were mixed with 90 µg of plant RTA in a total volume of 1 mL of PBS, shaking overnight at room temperature. The reaction mixture was then analyzed by a ricin elisa Wells of an EIA plate were coated with 100 µg/mL of asialofetuin in a volume of 100 µL overnight at 4 °C. The wells were washed with PBS plus 0.1% Tween-20, blocked with 3% BSA in PBS plus 0.02% sodium azide, rewashed, and incubated with dilutions of ricin or reassociated heterodimers. The wells were again washed, incubated with RBR12 monoclonal antibody at 1 µg/mL in PBS/0.5% BSA/0.01% sodium azide, rewashed, and incubated with alkaline phosphatase-conjugated goat anti-(mouse Ig) at 1:500 in PBS/0.5% BSA/0.01% sodium azide, washed again, and developed with p-nitrophenyl phosphate at 1 mg/mL in 50 mM diethanolamine pH 9.8. Absorbance at 405 nm was read on a microtiter plate reader. Reassociated mixtures were also analyzed by nonreducing SDS/PAGE followed by immunoblots with RBR12 anti-RTA monoclonal antibody and P10 anti-RTB monoclonal antibody. Densitometric scanning using the IBAS 2000 automatic image analysis system (Kontron, Germany) was done to quantify shift of immunoreactive material from 30 to 60 kDa. Cytotoxicity of Recombinant Mutant Heterodimers. A total of 1.5 × 104 HUT102 human T leukaemia cells in 100 µl were placed in 96-well flatbottomed plates in leucine-poor RPMI1640 containing 10% dialysed fetal bovine serum. Fifty µL of ricin, recombinant wild-type RTB-plant RTA heterodimer, and mutant RTB-plant RTA heterodimers at varying concentrations were added in the same medium and the cells incubated at 37 °C in 5% CO2 for 24 h. [3H]leucine, 0.5 µCi per well (120 mCi/mmol), in 50 µL of the same medium was added and incubated for 4 h. Cells were then harvested with a PhD cell harvester onto glass-fiber filter mats. The filters were dried, mixed with 3 mL of liquid scintillation fluid, and counted in an LKB-Wallac liquid scintillation counter gated for 3H. Cells cultured with medium alone served as controls. All assays were performed in triplicate. In some experiments, duplicate samples were incubated in the presence of 50 mM R-lactose. The ID50 was the concentration of protein which inhibited protein synthesis by 50% compared with control.

Bioconjugate Chem., Vol. 7, No. 1, 1996 33

Figure 2. Insect-derived wild-type and mutant RTBs. (A) Coomassie stained 15% reducing SDS/PAGE of wild-type and mutant RTBs: lane 1, low molecular weight prestained BioRad protein standards; lane 2, wild-type; lane 3, N255A; lane 4, Y248S. (B) Immunoblots using rabbit anti-ricin antibody of 15% reducing SDS/PAGE of wild-type and mutant RTBs: lane 1, low molecular weight prestained BioRad protein standards; lane 2, wild-type; lane 3, N255A; lane 4, D234E/A237R; lane 5, N255G; lane 6, K40M; lane 7, N46G/K40M; lane 8, N46G/K40M; lane 9, Q35N; lane 10, ∆32; lane 11, W37S; lane 12, Y248S; lane 13, AcNPV polyhedrin; lane 14, DRA protein. RESULTS AND DISCUSSION

Yields and Immunoreactivity of Mutant RTBs. Mutant RTBs were produced with modifications in single putative lectin sites in subdomain 2γ (N255A, N255G, D234E/A237R, and Y248S) and subdomain 1R (W37S, K40M, Q35N, and N46G/K40M). An additional mutant had a nonsense mutation (an additional adenine) at codon 32. The changes in amino acid residues were selected based on their presumed role in lectin function based on X-ray crystallography and, in some cases, their previous analysis in other expression systems (10-12). Four of the mutants have not been previously described (D234E/ A237R, Y248S, W37S, and Q35N). The yields were estimated from the optical density at 280 nm of neutralized alkaline eluants’ postaffinity chromatography (plant RTB OD ) 1.44 for 1 mg/mL) and densitometry of Coomassie-stained reducing SDS/PAGE (10-50% of the protein migrated at 33 kDa, Figure 2A). These results were confirmed by densitometry of immunoblots with polyclonal rabbit anti-ricin antibody. As shown in Figure 2B, all of the mutants except the nonsense mutant were reactive with the polyclonal antibody. Finally, a monoclonal antibody anti-RTB elisa was used to verify concentrations of each mutant. All three assays gave similar values. Purified wild-type RTB was obtained at 300-500 µg/L culture fluid, while the mutants varied in concentration about 10-fold. The yield of Y248S was 640 µg/L, and the yields of W37S, D234E/ A237R, and N255G were 264 µg/L, 240 µg/L, and 160 µg/ L, respectively. Lower yields were obtained with Q35N, N255A, K40M/N46G, and K40M. These mutants were produced at 80 µg/L, 40 µg/L, 32 µg/L and 24 µg/L, respectively. The nonsense mutant control did not yield

34 Bioconjugate Chem., Vol. 7, No. 1, 1996

levels above background (