The structure of SeviL, a GM1b/asialo-GM1 binding R-type lectin from the mussel Mytilisepta virgata

SeviL is a recently isolated lectin found to bind to the linear saccharides of the ganglioside GM1b (Neu5Ac$$\alpha$$ α (2-3)Gal$$\beta$$ β (1-3)GalNAc$$\beta$$ β (1-4)Gal$$\beta$$ β (1-4)Glc) and its precursor, asialo-GM1 (Gal$$\beta$$ β (1-3)GalNAc$$\beta$$ β (1-4)Gal$$\beta$$ β (1-4)Glc). The crystal structures of recombinant SeviL have been determined in the presence and absence of ligand. The protein belongs to the $$\beta$$ β -trefoil family, but shows only weak sequence similarity to known structures. SeviL forms a dimer in solution, with one binding site per subunit, close to the subunit interface. Molecular details of glycan recognition by SeviL in solution were analysed by ligand- and protein-based NMR techniques as well as ligand binding assays. SeviL shows no interaction with GM1 due to steric hindrance with the sialic acid branch that is absent from GM1b. This unusual specificity makes SeviL of great interest for the detection and control of certain cancer cells, and cells of the immune system, that display asialo-GM1.

Prolactin does not delay early tumour formation of breast cancer cells, but in combination with DNA damage can trigger an anti-tumour immune response

ABSTRACTThere are conflicting reports of the role of prolactin in breast cancer, and its role within the context of the tumour microenvironment is not well understood. In our previous study, we demonstrated a cross-talk between the ataxia telangiectasia-mutated (ATM) DNA damage response pathway and the PRL-Janus-kinase-2 (JAK2)-signal transducer and activator of transcription-5 (STAT5)-heat shock protein-90 (HSP90) pathway. In order to investigate the role of PRL in tumour initiation and the effect of DNA damage in vivo, we used a model of breast cancer initiation that assesses the ability of breast cancer cells to initiate orthotopic xenograft tumour formation after DNA damage. Breast cancer cells engineered to secrete human PRL or the control cells, were treated with the DNA damaging agent doxorubicin or vehicle and injected into mammary fat pad of immune deficient SCID mice. PRL secretion from human breast cancer cells did not change the tumour latency compared to controls, although combined doxorubicin and PRL treatment increased tumour latency. Injection of a natural killer (NK) cell-depleting antibody, anti-asialo GM1, resulted in faster tumour formation only in the PRL-secreting breast cancer cells that were pre-treated with doxorubicin, and not PRL-only or empty vector controls. These results may shed light on the conflicting reports of PRL in breast cancer, and demonstrate that at least within the context of breast cancer cell DNA damage, that PRL exposure in the tumour microenvironment does not delay tumour initiation, but PRL together with DNA damage of breast cancer cells results in reduced tumour volume over time due to asialo-GM1-positive immune cells.

Evaluation of Oncolytic Viral Therapy Combined with or without Artificial Immunoadjuvant Cell Therapy in Immunocompetent Mouse Tumor Models

Oncolytic viruses have a great potential to provide clinical improvement for cancer patients who do not respond to conventional therapies. The mechanisms of action include direct infection, immune-mediated cell death, and vascular collapse. We previously reported that a miRNA-regulated, thymidine kinase-deleted vaccinia virus (MDVV) specifically infects multiple myeloma (MM) cells and showed that the intravenous injection of MDVV significantly prolonged survival of severe combined immunodeficiency mice harboring a subcutaneous human RPMI8226 MM tumor (Blood 2014:124:2082, Mol Ther Oncolytics 2017;6:57). Although it was highly effective, there remained to be confirmed whether the MDVV would be similarly effective in the immunocompetent setting. To address this question, we performed several in vivo infection using immunocompetent C57BL/6 (B6) models. A vaccinia virus (VV) strain LC16, which was derived from the Lister strain, was widely used in 1970s for the smallpox eradication program with no severe adverse effect. Because its safety is robust, we used the LC16 for the backbone of the MDVV. To improve its safety and tumor specificity, we deleted viral thymidine kinase gene, and introduced miRNA (let-7a)-target sequences into the critical viral B5R gene. Because let-7a is abundant in normal differentiated tissues, and the endogenous let-7a down-regulates viral B5R gene, the MDVV becomes far more tumor specific (Mol Ther Oncolytics 2017;6:57). To observe whether the MDVV protects immunocompetent mice from MM, we generated a mouse MM model by transplanting 1x10^6 Vk*Myc tumor cells into syngeneic B6 mice. After confirming M-proteins in the serum, 1x10^7 pfu of MDVV was injected intravenously. Contrary to our expectation, we could not observe a clearly infected region, nor a prolongation of survival of these mice; probably due to difference in infectivity between species. Although mouse derived cell lines are generally insensitive to vaccinia virus as compared with human cell lines, a mouse lung carcinoma cell line 3LL showed a relatively high infectivity to VV in our screening in vitro infection experiment (Figure), therefore we used the B6/3LL subcutaneous tumor model instead of B6/Vk*Myc model. The B6/3LL mice were injected with mock or 1x10^7 pfu MDVV intravenously when the subcutaneous tumor became more than 5 mm in diameter. Although viral signal (determined by in vivo imaging using luciferase) disappeared 3 days after infection, the therapeutic benefit had remained, and prolongation of survival was observed Because natural killer (NK) cells were thought to play in the clearance of VV in the B6 mice, we attempted to extend the duration of infection by depleting NK cells using anti-asialo GM1 antibody, which, we thought, enhances viral oncolytic activity. As expected, we observed a persistent viral signal for up to 2 weeks in anti-asialo GM1 antibody treated B6/3LL mice. However, the anti-tumor effect seen with anti-asialo GM1 antibody was rather milder than that of without anti-asialo GM1 antibody, indicating the importance of NK activities in anti-tumor immunity. Based on this finding, we, in reverse, tried to stimulate the anti-tumor immunity using artificial immunoadjuvant cells co-expressing CD1d and tumor antigen/GM-CSF fusion protein. The CD1d is one of non-classical MHC molecules and is involved in the presentation of lipid antigens. When α-GlaCer is loaded on the CD1d, it activates natural killer T (NKT) cells through the interaction with invariant T cell receptor on NKT cells. When activated, NKT cells produce Th1 and Th2 cytokines, and activate NK cells. GM-CSF stimulates antigen-presenting cells (APCs), and fusion proteins of tumor antigen and GM-CSF are more efficiently taken up by APCs than tumor antigen alone (Nature 1993, 362:755). We generated the artificial immunoadjuvant cells by transducing 3LL tumor cells with lentiviral vectors expressing CD1d and ovalbumin (OVA)/GM-CSF fusion protein. After loading with α-GlaCer, the immunoadjuvant cells were irradiated with 50Gy, and injected intravenously into B6/3LL-OVA mice. We observed a significant reduction in tumor mass and prolongation of survival as compared with mock-injected mice. Currently, we are exploring the optimal setting of combination with MDVV virotherapy and immunoadjuvant cell therapy. In this meeting, we will present these results. Disclosures No relevant conflicts of interest to declare.

Binding of CFA/I Pili of Enterotoxigenic Escherichia coli to Asialo-GM1 Is Mediated by the Minor Pilin CfaE

CFA/I pili are representatives of a large family of related pili that mediate the adherence of enterotoxigenicEscherichia colito intestinal epithelial cells. They are assembled via the alternate chaperone-usher pathway and consist of two subunits, CfaB, which makes up the pilus shaft and a single pilus tip-associated subunit, CfaE. The current model of pilus-mediated adherence proposes that CFA/I has two distinct binding activities; the CfaE subunit is responsible for binding to receptors of unknown structure on erythrocyte and intestinal epithelial cell surfaces, while CfaB binds to various glycosphingolipids, including asialo-GM1. In this report, we present two independent lines of evidence that, contrary to the existing model, CfaB does not bind to asialo-GM1 independently of CfaE. Neither purified CfaB subunits nor CfaB assembled into pili bind to asialo-GM1. Instead, we demonstrate that binding activity toward asialo-GM1 resides in CfaE and this is essential for pilus binding to Caco-2 intestinal epithelial cells. We conclude that the binding activities of CFA/I pili for asialo-GM1, erythrocytes, and intestinal cells are inseparable, require the same amino acid residues in CfaE, and therefore depend on the same or very similar binding mechanisms.