Lectin Drug Conjugates Targeting High Mannose N-Glycans

Abstract

Cancer-associated alterations to glycosylation have been shown to aid cancer development and progression. An increased abundance of high mannose N-glycans has been observed in several cancers. Here, we describe the preparation of lectin drug conjugates (LDCs) that permit toxin delivery to cancer cells presenting high mannose N-glycans. Additionally, we demonstrate that cancer cells presenting low levels of high mannose N-glycans can be rendered sensitive to the LDCs by co-treatment with a type I mannosidase inhibitor. Our findings establish that an increased abundance of high mannose N-glycans in the glycocalyx of cancer cells can be leveraged to enable toxin delivery.

Graphical Abstract

A toxin delivery system that targets the aberrant presentation of high mannose N-glycans on cancer cells is described. This delivery system leverages lectin drug conjugates (LDCs) that were prepared using a high mannose N-glycan binding lectin that is site specifically functionalized at the C-terminus with peptide toxin via a sortase ligation. In vitro studies demonstrate that the described LDCs potently and preferentially induce apoptosis in high mannose N-glycans presenting cancer cells.

Introduction

The glycocalyx is a carbohydrate-rich coating that decorates the outside of all cells. Alterations to the composition of the glycocalyx are associated with many diseases, including cancer.^[1] The aberrant glycosylation associated with cancer development and progression has been extensively documented, and has prompted efforts to develop therapies targeting tumour-associated carbohydrate antigens (TACAs) and the associated biosynthetic pathways.^[2] TACA targeting vaccines and antibodies, which have been investigated for several decades, demonstrate the therapeutic challenges and potential of TACA targeting modalities.^[2b] Lectins are another class of carbohydrate-binding proteins that have also been investigated in the oncology realm as diagnostic and therapeutic agents.^[3] The inherent toxicity of many lectins prompted the study of their antineoplastic properties.^[4] Unfortunately, indiscriminate cytotoxicity and off-target activities (e.g., erythrocyte agglutination) have curtailed investigations assessing the antineoplastic potential of lectins despite the functional and structural diversity of this vast class of proteins.^[5] The application of highly toxic lectins has subsided in favour of lectin-based strategies that enable the delivery of toxins and toxin loaded nanoparticles to malignant cells.^[6] These strategies continue to gain traction with the aid of promising preclinical studies.^[7]

Unique glycan targets that can be leveraged to enable the delivery of toxins to malignant cells are of considerable interest. An increased abundance of high mannose N-glycans in the glycocalyx of cancer cells has now been described in several studies.^[8] This differs from healthy tissue, in which proteins carrying high mannose N-glycans are predominately located in the endoplasmic reticulum (ER). Matoba and co-workers have elegantly demonstrated that high mannose N-glycans presented on malignant cells can be leveraged as ligands for antitumor agents in vitro and in vivo.^[9] Together, these observations led us to question if high mannose N-glycans on malignant cells could be leveraged to deliver toxins in a specific manner (Figure 1). A similar approach has been applied by Oda and co-workers to facilitate the delivery of a toxins to fucose presenting malignant cells.^[7] We hypothesized that an appropriately designed high mannose N-glycan targeting lectin drug conjugate (LDC) would permit the delivery of a toxin to cancer cells presenting an increased abundance of high mannose N-glycans.

Figure 1.

High mannose N-glycan specific LDC. A graphical description of a high mannose N-glycan specific LDC binding to high mannose N-glycans attached to proteins on the plasma membrane of a cancer cell.

Previously, we engineered a high-mannose N-glycan specific cyanovirin-N homologue from the cyanobacterium Cyanothece7424 (1, Cyt-CVNH). ^[10] The engineered lectin (1) demonstrates specificity for cells presenting high-mannose N-glycans and can be functionalized at the C-terminus via a sortase ligation without disrupting the carbohydrate binding properties of the lectin. As such, we reasoned that this lectin would be an ideal candidate to enable the delivery of cytotoxic agents to high mannose N-glycan presenting malignant cells. The success of targeted delivery strategies is also highly dependent on the cytotoxic agent being delivered. Early antibody drug conjugates (ADCs) carrying traditional chemotherapy drugs (i.e., methotrexate and doxorubicin) were less efficient than standard cytotoxic drugs and often required extremely high dosing.^[11] These observations have led to potent cytotoxic agents with sub-nanomolar activity being applied in targeted delivery systems.^[12] The monomethyl auristatin toxins, MMAE and MMAF, are currently the most prevalent payloads applied in FDA approved ADCs.^[13] MMAE and MMAF are microtubule destabilizing synthetic peptides derived from the dolastatin 10 peptide that is produced by Dolabella auricularia.^[14] Previously, these peptides have been conjugated to targeting molecules via a sortase mediated ligation (SML).^[15] As such, we reasoned that conjugating the monomethyl auristatin toxins to Cyt-CVNH (1) could provide specific and potent antineoplastic constructs.

In this study, we describe the preparation of high mannose N-glycan specific LDCs. A SML is applied to prepare these constructs, which permits the site-specific modification of Cyt-CVNH and ensures that the high mannose N-glycan binding properties are retained. The cytotoxic activity of the LDCs towards high mannose N-glycan presenting tumorigenic cells and non-tumorigenic cells is determined. Additionally, we assessed if cancer cells that present low levels of high mannose N-glycans can be sensitized to the action of the LDCs by a high mannose N-glycan inducing type I mannosidase inhibitor.

Results and Discussion

The LDCs were prepared via a SML involving Cyt-CVNH (1) and peptides containing the monomethyl auristatin toxins MMAE (2a) and MMAF (2b) (Figure 2A). The C-terminus of 1 carries a sortase motif (i.e., LPETG), and a His tag; while 2a and 2b carry the sortase ligation facilitating triglycine motif, which is linked to the monomethyl auristatin peptides via a cathepsin sensitive valine-citrulline-p-aminocarbonate (VC-PABC) linker. This ligation was catalysed by the sortase A heptamutant (7M SrtA), and we observed that the ligation proceeded smoothly using our previously optimized ligation conditions to provide the desired Cyt-CVNH monomethyl auristatin peptide conjugates (3a-b, Figure 2B).^[16] In addition, a toxin free control, Cyt-CVNH-GGG (4), was prepared in a similar manner (Figure S1).

Figure 2.

LDC synthesis. (A) The SML involving 1 and 2a/2b to provide Cyt-CVNH LDCs 3a/3b. (B) MALDI spectra of the products.

To determine the ability of Cyt-CVNH to bind to high mannose presenting cancer cells, we utilized a fluorophore labelled Cyt-CVNH derivative (i.e., Cyt-CVNH MB 488) that we have previously described and applied in assays to demonstrate the presence of high mannose N-glycans on cells.^[10] Here, Cyt-CVNH MB 488 was applied to assess the ability of Cyt-CVNH to bind to triple negative breast cancer (TNBC) MDA-MB-231 cells, lung adenocarcinoma A549 cells, and adenocarcinoma HT-29 cells, which have been previously described as high mannose presenting cancer cell lines by Matoba and co-workers.^[9] Using flow cytometry, we observed that Cyt-CVNH MB 488 readily bound to MDA-MB-231, A549, and HT-29 cells (Figure 3A). Additionally, we determined the ability of Cyt-CVNH to bind to non-tumorigenic mammary gland epithelial MCF10A cells, and observed that the lectin binds relatively poorly to this cell line, which is in agreement with previous studies.^[9]

Figure 3.

Analysis of LCD cytotoxicity. (A) Histograms depicting the binding of Cyt-CVNH MB488 to the MDA-MB-231, A549, HT-29 and MCF10A cell lines (top to bottom). (B-D) Changes in the viability of MDA-MB-231, A549, HT-29 and MCF10A cell lines up treatment with Cyt-CVN-GGG (4), Cyt-CVN-MMAE (3a), and Cyt-CVN-MMAF (3b) respectively.

The impact of Cyt-CVNH-GGG (4) on the viability of these cell lines was subsequently assessed (Figure 3B). We observed the cytotoxic activity of 4 toward MDA-MB-231, A549, and HT-29 cells to be in the 250-350 nM range, however, MCF10A cells were relatively insensitive to 4. This indicates that the cytotoxic effects of 4 are likely high mannose N-glycan dependent, however, the precise mechanism responsible for the cytotoxic activity of 4 has yet to be defined. We speculate that the toxicity of 4 towards high mannose N-glycan presenting cells may be due to endocytosis and retrograde trafficking of a portion of the lectin to the endoplasmic reticulum (ER), where it may disrupt protein folding and trafficking.^[17] At this time, we have not confirmed this as the mechanism of action and cannot rule out alternative mechanisms.

Next, we determined the cytotoxic activity of LDCs 3a and 3b. High mannose N-glycan presenting tumorigenic cell lines were highly sensitive to the LDCs 3a and 3b, while the non-tumorigenic MCF10A cells, which present low levels of high mannose N-glycans, are relatively insensitive (Figure 3C and 3D). This demonstrates that the Cyt-CVNH monomethyl auristatin toxin conjugates can be leveraged to deliver a toxic payload to tumorigenic cells. The cytotoxic effects of the LDCs 3a and 3b is inhibited by a high mannose N-glycan, which indicates that the observed activity of the LDCs is likely high mannose N-glycan dependent (Figure S3). The MMAE LDC 3a exhibits greater potency than the MMAF LDC 3b, and when the potency of 3a and 3b toward the TNBC MDA-MB-231 and the non-tumorigenic MCF10A cell lines are compared, a greater in vitro therapeutic index (TI) is also observed with 3a (TI_3a: 80.5 Vs TI_3b: 10.4). Importantly, the LDCs are considerably more potent than the Cyt-CVNH-GGG (4) control. It is also interesting to note that the free MMEA peptide is more potent than the MMAE LDC 3a, however, the free MMAF peptide is approximately one order of magnitude less potent than the MMAF LDC 3b (Figure S4). MMAF is less membrane permeable and potent than MMAE, and we hypothesize that the MMAF LDC 3b exhibits greater potency relative to the free MMAF peptide due to the ligand mediated uptake of the LDC. Comparing the activity of the MMAE LDC (3a) with TACA targeting antibody drug conjugates (ADCs) in assays carried out with MDA-MB-231 cells indicates that the LDCs exhibit similar activity to ADCs.^[18] Additionally, the referenced ADCs have an MMEA-to-antibody ratio of ~4, while the LDC 3a only carries one molecule of MMEA per lectin, which further highlights the efficiency of this delivery strategy.

Recognizing that the sensitivity of the malignant cells to the LDCs will vary depending on the prevalence of high mannose N-glycan presentation, we questioned if this variability could be addressed by increasing the abundance of high mannose N-glycans. Disrupting the N-glycosylation biosynthetic pathway using a type I mannosidase inhibitor can increase the abundance of high mannose N-glycans.^[19] To test this hypothesis, we utilized TNBC MDA-MB-468 cells, as this cell line presents low levels of high mannose N-glycans, and the kifunensine analogue JDW-II-010 (5) to inhibit type I mannosidase (MAN I) enzymes.^[10] When MDA-MB-468 cells were treated with 5 a significant increase in the binding of Cyt-CVNH MB 488 to the cells was observed by flow cytometry (Figure 4A), indicating that 5, increased the abundance of high mannose N-glycans. Next, we determined if treating MDA-MB-468 cells with 5 altered the sensitivity of these cells to Cyt-CVNH-GGG (4) and the LDCs 3a and 3b (Figure 4B–C). In the absence of 5 the MDA-MB-468 cells were relatively insensitive to Cyt-CVNH-GGG and the LDCs; however, the addition of JDW-II-010 (5) dramatically increased the sensitivity of the MDA-MB-468 cells to 4, 3a and 3b. For example, the IC₅₀ of LDC 3a was ~135 fold higher upon co-treatment of MDA-MB-468 cells with JDW-II-010 (5).

Figure 4.

Sensitizing cells to LDCs targeting high mannose N-glycans. (A) Histograms depicting the binding of Cyt-CVNH MB488 to MDA-MB-468 cells post incubation with JDW-II-010 (1 mM) or vehicle for 48 hours. (B-D) Changes in the viability of MDA-MB-468 cells upon co-treatment with JDW-II-010 (1 mM) or vehicle plus Cyt-CVN-GGG (4), Cyt-CVN-MMAE (3a), and Cyt-CVN-MMAF (3b) respectively.

Conclusion

In summary, we have prepared high mannose N-glycan targeting LDCs via a sortase mediated ligation. These constructs deliver toxins to cell lines presenting high mannose N-glycans and we observed that the LDCs preferentially impact the viability of tumorigenic cells. Additionally, we demonstrate that disrupting N-glycosylation via the inhibition of MAN I enzymes significantly increases the sensitivity of the TNBC MDA-MB-468 cell line, which present low levels of high mannose N-glycans, to the cytotoxic effects of the LDCs. While we recognize that this sensitization strategy would likely also alter the susceptibility of non-tumorigenic tissue to the LDCs, our findings demonstrate the potential of pharmacologically increasing the abundance of high mannose N-glycans to enable toxin delivery strategies. Efforts to develop an approach that permits the selective inhibition of MAN I enzymes in malignant cells are ongoing.

Experimental Section

The materials and methods utilized in the experiments described in this manuscript can be found in the associated supplementary information.

Biography

Mark P. Farrell received his B.Sc. (2010) and Ph.D. (2014) from the National University of Ireland, Galway. His Ph.D. research, carried out under the tutelage of Paul V. Murphy, focused on studying glycosyl bond anomerization. Subsequently, he joined the lab Amos B. Smith, III at the University of Pennsylvania where he developed new reaction methods and HIV-1 entry inhibitors. In 2017, he joined the Department of Medicinal Chemistry at the University of Kansas, where he studies carbohydrate-protein interactions, and develops molecular strategies to leverage the potential of the immune system.