Double-lectin site ricin B chain mutants expressed in insect cells have residual galactose binding: evidence for more than two lectin sites on the ricin toxin B chain

Structural Basis of Antibody-Mediated Inhibition of Ricin Toxin Attachment to Host Cells

Monoclonal antibodies, JB4 and SylH3, neutralize ricin toxin (RT) by inhibiting the galactose-specific lectin activity of the B subunit of the toxin (RTB), which is required for cell attachment and entry. It is not immediately apparent how the antibodies accomplish this feat, considering that RTB consists of two globular domains (D1, D2) each divided into three homologous subdomains (α, β, γ) with the two functional galactosyl-specific carbohydrate recognition domains (CRDs) situated on opposite poles (subdomains 1α and 2γ). Here, we report the X-ray crystal structures of JB4 and SylH3 Fab fragments bound to RTB in the context of RT. The structures revealed that neither Fab obstructed nor induced detectable conformational alterations in subdomains 1α or 2γ. Rather, JB4 and SylH3 Fabs recognize nearly identical epitopes within an ancillary carbohydrate recognition pocket located in subdomain 1β. Despite limited amino acid sequence similarity between SylH3 and JB4 Fabs, each paratope inserts a Phe side chain from the heavy (H) chain complementarity determining region (CDR3) into the 1β CRD pocket, resulting in local aromatic stacking interactions that potentially mimic a ligand interaction. Reconciling the fact that stoichiometric amounts of SylH3 and JB4 are sufficient to disarm RTB's lectin activity without evidence of allostery, we propose that subdomain 1β functions as a "coreceptor" required to stabilize glycan interactions principally mediated by subdomains 1α and 2γ. Further investigation into subdomain 1β will yield fundamental insights into the large family of R-type lectins and open novel avenues for countermeasures aimed at preventing toxin uptake into vulnerable tissues and cells.

Sites of vulnerability on ricin B chain revealed through epitope mapping of toxin-neutralizing monoclonal antibodies

Ricin toxin's B subunit (RTB) is a multifunctional galactose (Gal)-/N-acetylgalactosamine (GalNac)-specific lectin that promotes uptake and intracellular trafficking of ricin's ribosome-inactivating subunit (RTA) into mammalian cells. Structurally, RTB consists of two globular domains (RTB-D1, RTB-D2), each divided into three homologous sub-domains (α, β, γ). The two carbohydrate recognition domains (CRDs) are situated on opposite sides of RTB (sub-domains 1α and 2γ) and function non-cooperatively. Previous studies have revealed two distinct classes of toxin-neutralizing, anti-RTB monoclonal antibodies (mAbs). Type I mAbs, exemplified by SylH3, inhibit (~90%) toxin attachment to cell surfaces, while type II mAbs, epitomized by 24B11, interfere with intracellular toxin transport between the plasma membrane and the trans-Golgi network (TGN). Localizing the epitopes recognized by these two classes of mAbs has proven difficult, in part because of RTB's duplicative structure. To circumvent this problem, RTB-D1 and RTB-D2 were expressed as pIII fusion proteins on the surface of filamentous phage M13 and subsequently used as "bait" in mAb capture assays. We found that SylH3 captured RTB-D1 (but not RTB-D2) in a dose-dependent manner, while 24B11 captured RTB-D2 (but not RTB-D1) in a dose-dependent manner. We confirmed these domain assignments by competition studies with an additional 8 RTB-specific mAbs along with a dozen a single chain antibodies (VHHs). Collectively, these results demonstrate that type I and type II mAbs segregate on the basis of domain specificity and suggest that RTB's two domains may contribute to distinct steps in the intoxication pathway.

Partial identification of carbohydrate-binding sites of a Galalpha1,3Galbeta1,4GlcNAc-specific lectin from the mushroom Marasmius oreades by site-directed mutagenesis

The Galalpha1,3Galbeta1,4GlcNAc-specific lectin from the mushroom Marasmius oreades (MOA) contains a ricin B chain-like (QXW)(3) domain at its N-terminus that is composed of three identical subdomains (alpha, beta, and gamma) and a C-terminal domain of unknown function. Here, we investigate the structure-function relationship of MOA to define the number and location of its carbohydrate-binding sites. Based on the sequence alignment of MOA to the ricin B-chain lactose-binding sites, we systematically constructed mutants by site-directed mutagenesis. We have used precipitation and hemagglutination assay for the primary analyses, and surface plasmon resonance for the kinetic analysis. Among amino acid residues at the putative carbohydrate-binding sites, Gln(46) in the alpha subdomain and Trp(138) in the gamma subdomain have been identified to be important amino acid residues directly or indirectly involved in carbohydrate recognition. By surface plasmon resonance, Q46A and W138A were 2.4- and 4.3-fold less active than that of the wild-type MOA (K(a) = 2 x 10(7)), respectively. A double-site mutant (Q46A/W138A) had activity similar to W138A. The C-terminal deletion mutant MOADeltaC showed hemagglutination and precipitation activity, although its binding constant was 12.5-fold less active (K(a) = 1.6 x 10(6)) than that of the wild-type MOA. A C-terminal deletion mutant with mutations at both Gln(46) and Trp(138) (MOADeltaC-Q46A/W138A) was 12,500-fold less active (K(a) = 1.6 x 10(3)) than that of the wild-type MOA. On the basis of this observation, we conclude that both alpha and gamma subdomains are most probably involved in carbohydrate binding, but the beta subdomain appears to be inactive.

Conjugation of Blocked Ricin to an Anti-CD19 Monoclonal Antibody Increases Antibody-Induced Cell Calcium Mobilization and CD19 Internalization

CD19 (B4) is a signal transduction molecule restricted to the B-cell lineage and the target of the immunotoxin anti-B4–blocked ricin (anti-B4–bR), which is composed of the monoclonal antibody (MoAb) anti-B4 and the modified plant toxin blocked ricin. To explore the influence of conjugation of blocked ricin to anti-B4 on functional activation of CD19, we investigated the effects of anti-B4–bR, and that of unconjugated anti-B4, on intracellular calcium mobilization and ligand/receptor internalization. The data showed that anti-B4–bR was more potent than anti-B4 in triggering cell calcium mobilization. Two other immunotoxins that bind to the B-cell surface, anti-CD20–bR and anti-CD38–bR, were devoid of the calcium increasing effect of anti-B4–bR. Furthermore, anti-B4 conjugated to ricin A-chain was also without effect in Namalwa cells, indicating that the ricin B-chain component was required for anti-B4–bR effect. Anti-B4–bR-induced calcium mobilization was inhibited in the presence of lactose, yet the calcium response induced by cross-linking anti-B4–bR with a second step antibody was not affected. The extent of CD19 modulation induced by anti-B4–bR was higher than that induced by anti-B4, and lactose dampened the effect of the immunotoxin down to that of the MoAb. Moreover, the number of internalized immunotoxin molecules was higher than that of unconjugated MoAb. Although a mechanism involving dimerization of the immunotoxin cannot be excluded, our findings suggest that the residual binding activity of the blocked ricin B-chain to cell surface molecules plays an important role in the greater calcium fluxes and greater internalization rate of anti-B4–bR, and is of functional significance in the mechanism of intoxication of cells by the immunotoxin.

Conjugation of Blocked Ricin to an Anti-CD19 Monoclonal Antibody Increases Antibody-Induced Cell Calcium Mobilization and CD19 Internalization

CD19 (B4) is a signal transduction molecule restricted to the B-cell lineage and the target of the immunotoxin anti-B4–blocked ricin (anti-B4–bR), which is composed of the monoclonal antibody (MoAb) anti-B4 and the modified plant toxin blocked ricin. To explore the influence of conjugation of blocked ricin to anti-B4 on functional activation of CD19, we investigated the effects of anti-B4–bR, and that of unconjugated anti-B4, on intracellular calcium mobilization and ligand/receptor internalization. The data showed that anti-B4–bR was more potent than anti-B4 in triggering cell calcium mobilization. Two other immunotoxins that bind to the B-cell surface, anti-CD20–bR and anti-CD38–bR, were devoid of the calcium increasing effect of anti-B4–bR. Furthermore, anti-B4 conjugated to ricin A-chain was also without effect in Namalwa cells, indicating that the ricin B-chain component was required for anti-B4–bR effect. Anti-B4–bR-induced calcium mobilization was inhibited in the presence of lactose, yet the calcium response induced by cross-linking anti-B4–bR with a second step antibody was not affected. The extent of CD19 modulation induced by anti-B4–bR was higher than that induced by anti-B4, and lactose dampened the effect of the immunotoxin down to that of the MoAb. Moreover, the number of internalized immunotoxin molecules was higher than that of unconjugated MoAb. Although a mechanism involving dimerization of the immunotoxin cannot be excluded, our findings suggest that the residual binding activity of the blocked ricin B-chain to cell surface molecules plays an important role in the greater calcium fluxes and greater internalization rate of anti-B4–bR, and is of functional significance in the mechanism of intoxication of cells by the immunotoxin.

Ricin fusion toxin targeted to the human granulocyte-macrophage colony stimulating factor receptor is selectively toxic to acute myeloid leukemia cells

Treatment failure of patients with acute myelogenous leukemia (AML) is frequently due to the development of multidrug resistance phenotype blasts. We have expressed a fusion protein consisting of human granulocyte-macrophage colony stimulating factor (GMCSF) fused to the N-terminus of a lectin-deficient ricin toxin B chain (RTB) in Spodoptera frugiperda insect cells. The fusion protein was purified by immunoaffinity chromatography and reassociated with chemically deglycosylated ricin toxin A chain (RTA). The resulting fusion toxin was found to react with antibodies to GMCSF, RTB and RTA and had the predicted molecular mass of 80 kDa. GMCSF-ricin bound poorly to asialofetuin (Kd = 10(6) M-1) and receptor negative cells indicating loss of lectin activity, but bound strongly to GMCSF receptor positive HL60 cells. Ligand displacement assays showed fusion toxin affinity 2.6-fold less than native GMCSF. Selective inhibition of protein synthesis was observed on receptor positive cells. Induction of apoptosis was also observed on receptor positive cells. Cells expressing multidrug resistance gene products (P-gp, Bcl2 and BclXL) were also sensitive to fusion toxin. These results suggest that GMCSF-ricin deserves further preclinical development.