Introduction: Banana Lectin (BanLec) is a glycoprotein-binding lectin derived from banana fruit that has antiviral activity. BanLec binds high mannose glycans expressed on the viral envelopes of HIV, Ebola, influenza, and coronaviruses. BanLec mitogenicity can be divorced from antiviral activity via a single amino acid change (H84T). The SARS-CoV-2 spike (S) protein is decorated with high mannose N-glycosites that are in close proximity to the viral receptor binding domain (RBD). Our goal was to use the H84T-BanLec as the extracellular targeting domain of a chimeric antigen receptor (CAR). We hypothesized that engineering NK cells to express an H84T-BanLec CAR would specifically direct antiviral cytotoxicity against SARS-CoV-2.
Methods: H84T-BanLec was synthesized and added to a 4-1BB.ζ CAR by subcloning into an existing retroviral vector. To modify primary human NK cells, CD3-depleted peripheral blood mononuclear cells were first activated with lethally irradiated feeder cells (K562.mbIL15.4-1BBL), then transduced with transiently produced replication incompetent γ-retrovirus carrying the H84T-BanLec.4-1BB.ζ CAR construct. Vector Copy Number (VCN) per cell was measured and CAR protein expression detected with Western blotting. 293T cells were engineered to express human ACE2 (hACE2.293T), the binding receptor for SARS-CoV-2. CAR expression on NK cells and SARS-CoV-2 S-protein binding to hACE2.293T were measured using FACS. S-protein pseudotyped lentivirus carrying a firefly Luciferase (ffLuc) reporter was produced. Viral infectivity was measured using bioluminescence (BL) detection in virally transduced cells. H84T-BanLec CAR NK cells were added to our S-protein pseudotyped lentiviral infectivity assay and degree of inhibited transduction was measured. NK cell activation was assessed with detection of IFNγ and TNFα secretion using ELISA.
Results: A median of 4.5 integrated H84T-BanLec CAR copies per cell was measured (range 3.5-7.45, n=4). The CAR was detected by Western blot in NK cell lysates using antibodies to TCRζ and H84T-BanLec. Surface expression of the CAR on primary NK cells was recorded on day 4 after transduction (median [range], 67.5% CAR-positive [64.7-75%], n=6; Fig. 1). CAR expression was maintained on NK cells in culture for 14 days (58.9% CAR-positive [43.6-66.7%], n=6; Fig. 1). ACE2 expression and binding of recombinant S-proteins to hACE2 on hACE2.293T but not parental 293Ts was verified. S-protein pseudotyped lentiviral transduction of hACE2.293T was confirmed with increase in BL from baseline across diminishing viral titer (n=3; Fig. 2). Control 293T cells without hACE2 expression were not transduced, confirming specificity of viral binding and entry dependent on hACE2 (n=3; Fig. 2). S-protein pseudoviral infectivity of hACE2.293T cells was inhibited by both H84T-BanLec CAR-NK and unmodified NK cells, with enhanced inhibition observed in the CAR-NK condition (mean % pseudovirus infectivity +/- SEM of hACE2.293T in co-cultures with unmodified NK vs. H84T-BanLec CAR-NK; 65 +/-11% vs 35%+/- 6% for 1:1 effector-to-target ratio, p=0.05; 78 +/-3% vs 68%+/- 3% for 1:2.5 effector-to-target ratio, p=0.03; n=6; Fig.3). Both unmodified and H84T-BanLec CAR-NK cells were stimulated to secrete inflammatory mediators when co-cultured with pseudoviral particles and virally infected cells. CAR-NK cells showed overall higher cytokine secretion both at baseline and with viral stimulation.
Conclusions: A glycoprotein binding H84T-BanLec CAR was stably expressed on the surface of NK cells. CAR-NK cells are activated by SARS-CoV-2 S-pseudovirus and virally infected cells. Viral entry into hACE2 expressing cells was inhibited by H84T-BanLec CAR-NK cells. Translation of H84T-BanLec CAR-NK cells to the clinic may have promise as an effective cellular therapy for SARS-CoV-2 infection.
Figure 1
Disclosures
Markovitz:
University of Michigan: Patents & Royalties: H84T BanLec and of the H84T-driven CAR construct. Bonifant:
Merck, Sharpe, Dohme: Research Funding; BMS: Research Funding; Kiadis Pharma: Research Funding.