Mitoxantrone modulates a heparan sulfate-spike complex to inhibit SARS-CoV-2 infection

Spike-mediated entry of SARS-CoV-2 into human airway epithelial cells is an attractive therapeutic target for COVID-19. In addition to protein receptors, the SARS-CoV-2 spike (S) protein also interacts with heparan sulfate, a negatively charged glycosaminoglycan (GAG) attached to certain membrane proteins on the cell surface. This interaction facilitates the engagement of spike with a downstream receptor to promote viral entry. Here, we show that Mitoxantrone, an FDA-approved topoisomerase inhibitor, targets a heparan sulfate-spike complex to compromise the fusogenic function of spike in viral entry. As a single agent, Mitoxantrone inhibits the infection of an authentic SARS-CoV-2 strain in a cell-based model and in human lung EpiAirway 3D tissues. Gene expression profiling supports the plasma membrane as a major target of Mitoxantrone but also underscores an undesired activity targeting nucleosome dynamics. We propose that Mitoxantrone analogs bearing similar heparan sulfate-binding activities but with reduced affinity for DNA topoisomerases may offer an alternative therapy to overcome breakthrough infections in the post-vaccine era.


Scientific Reports
| (2022) 12:6294 | https://doi.org/10.1038/s41598-022-10293-x www.nature.com/scientificreports/ L 8,15 . In vitro, spike-and ACE2-mediated membrane fusion could even be activated by trypsin, a non-specific protease 15 . Thus, the protease sensitivity of the S2 domain is a major determinant of viral entry efficiency 4,7,16 . In addition to above-mentioned receptors, we and others recently reported that SARS-CoV-2 also uses the cell surface heparan sulfate proteoglycans (HSPGs) as an attachment factor during cell entry [17][18][19] . HSPGs refers to a family of membrane proteins conjugated with negatively charged heparan sulfate 20 , which can attract a variety of cargos to the cell surface. This presumably increase the cargo dwell time, and therefore, promote cargo entry via endocytosis 18,21 . Intriguingly, heparan sulfate can bind directly to spike 18,19,[22][23][24][25][26] , forming a ternary complex that also includes ACE2 17 . Because heparan sulfate is essential for the recruitment of viral particles to the cell surface, drugs targeting the spike-heparan sulfate complex may interfere with the SARS-CoV-2 entry process independently of viral entry receptors.
Although viral receptors are generally considered as an ideal target for antiviral drug development, the existence of redundant SARS-CoV-2 receptors has imposed a challenge for this drug development strategy [27][28][29] .
Our recent drug repurposing screen identified an FDA-approved anti-cancer drug named Mitoxantrone, which binds heparan sulfate and the heparan sulfate analog heparin with high affinities 18 . Mitoxantrone inhibits the entry of pseudo-viral particles coated with the SARS-CoV-2 S protein, but the underlying mechanism is unclear. Importantly, whether Mitoxantrone can block the entry of authentic SARS-CoV-2 virus is unknown. In this study, we show that Mitoxantrone targets a spike-heparin/heparan sulfate complex, restricting the membrane fusion competency of spike. As a result, Mitoxantrone inhibits the entry of an authentic SARS-CoV-2 strain in human lung epithelial cells and in a human lung EpiAirway 3D tissue model. Thus, a small-molecule COVID-19 therapeutics could be potentially developed by modifying Mitoxantrone, removing the undesired topoisomerase binding activity while keeping the heparin/heparan sulfate binding moiety.

Mitoxantrone targets a spike-heparin complex. Because heparin/heparan sulfate binds to both
Mitoxantrone and spike, we thought that these polysaccharides might use the same binding site to interact with Mitoxantrone and spike. If so, Mitoxantrone might compete with spike for binding to HSPG on the cell surface. To test this idea, we conducted pulldown experiments by incubating the purified extracellular domain (ECD) of spike with heparin-conjugated beads in the presence of 150 mM salt or in a low salt buffer (see methods). Coomassie blue staining showed efficient co-precipitation of spike with heparin under low salt conditions, but the interaction was significantly attenuated by salt (Fig. 1A), consistent with the proposed charge-charge interactions between spike and heparin/heparan sulfate. As a control, Sepharose lacking heparin did not precipitate spike even under low salt conditions. In the presence of 150 mM salt, the interaction of spike with heparin could only be detected by immunoblotting. Interestingly, when heparin beads were first incubated with excess Mitoxantrone, prebound Mitoxantrone did not inhibit the spike-heparin interaction under this condition. Instead, Mitoxantrone enhanced spike binding to heparin by ~ 7-fold. By contrast, Banoxantrone, a compound structurally homologous to Mitoxantrone, did not show such an effect ( Fig. 1B-D). Under low salt conditions, neither Mitoxantrone nor Banoxantrone affected the spike-heparin interaction (Fig. 1E). These data suggest that Mitoxantrone may change heparin (e.g. charge distribution) to increase its affinity to spike. However, our data cannot rule out the possibility that Mitoxantrone may act on spike as well.
Our data raised the possibility that Mitoxantrone binding might cause a change in the conformational dynamics of the spike-heparin complex. To test this hypothesis, we treated purified spike ECD with low concentrations of proteinase K in the absence or presence of Mitoxantrone and/or heparin. Both the full-length spike ECD and the furin-processed S2 domain were readily degraded by proteinase K regardless of whether Mitoxantrone or heparin was present ( Fig. 2A). However, when Mitoxantrone and heparin were both present, the S2 domain of spike was more resistant to proteolysis (Fig. 2B), suggesting that the conformational dynamics of spike may be altered by heparin and Mitoxantrone. Alternatively, the bound drug and polysaccharide may partially shield the spike S2 cleavage site.
Mitoxantrone inhibits spike-mediated membrane fusion. The reduced sensitivity of the S2 domain to protease by heparin and Mitoxantrone treatment raised the possibility that Mitoxantrone may block spikemediated membrane fusion. To test this, we established an in vitro membrane fusion assay by mixing two populations of HEK293T cells expressing spike and ACE2-GFP, respectively (Fig. 3A). To identify spike-positive cells, the spike-expressing plasmid was co-transfected with a mCherry-expressing plasmid. When mCherry and spike co-transfected cells were incubated with GFP-transfected cells or when mCherry-expressing cells were mixed with ACE2-GFP positive cells, no cell fusion was observed as green and red cells were segregated (Fig. 3B, panels 1, 2). However, when spike-, mCherry-positive cells are incubated with ACE2-GFP cells at 1:1 ratio for 90 min, we observed rapid cell-cell fusion, resulting in the mixing of green and red fluorescence in giant syncytia each filled with multiple nuclei (panels 3, 5). Interestingly, when cell fusion was carried out in the presence of Mitoxantrone, the cell-cell fusion was significantly inhibited as indicated by increased number of unfused cells or reduced syncytium size (panel 4, quantification in Fig. 3C). Altogether, we concluded that Mitoxantrone modulates a heparan sulfate-spike complex, changing the protease sensitivity of the S2 domain, which inhibits spike-mediated membrane fusion.

Mitoxantrone inhibits the cell entry of an authentic SARS-CoV-2 strain. To examine whether
Mitoxantrone could block the entry of an authentic SARS-CoV-2 strain, we treated cells with Mitoxantrone or as a control, with DMSO, and then exposed these cells to SARS2-CoV-2 (USA-WA1/2020) at an MOI of 0.1 for 5 h. We then stained the cells with an antibody specific for the S protein to detect viral particles inside the cells. To demonstrate the antibody specificity, uninfected cells were stained in parallel, which showed only a www.nature.com/scientificreports/ low diffusive background signal (Fig. 4A, top panels). By contrast, control cells infected with SARS-CoV-2 contained many bright spike-positive puncta throughout the cytoplasm (Fig. 4A, middle panels). Strikingly, when Mitoxantrone-treated cells were infected and stained, we detected a significantly lower number of spike-positive signals compared to DMSO-treated cells (Fig. 4A, bottom panels; Fig. 4B). These results suggest that like pseudoviral particles, the infection of authentic SARS-CoV-2 is also inhibited by Mitoxantrone.

Mitoxantrone inhibits SARS-CoV-2 infection in a human 3D
EpiAirway model. We next tested the anti-SARS-CoV-2 activity of Mitoxantrone in a model that mimics viral infection in the human airway. To this end, we chose human lung epithelial cell-derived 3D EpiAirway tissues, which contain three pseudostratified layers including an apical ciliated surface, the underlying mucociliary epithelium and a basal membrane. These tissues were cultured at the air-liquid interface (ALI) and can be infected with SARS-CoV-2 from the api- To evaluate the cytotoxicity of Mitoxantrone, we measured the level of lactate dehydrogenase (LDH) in the basal culture media. LDH is a cytosolic enzyme released to the cell exterior upon cell death. Accordingly, we detected high levels of LDH in the basal media from tissues treated with the apoptosis-inducing drug bleomycin (Fig. 5E). We also observed a small increase in LDH in the basal media collected from SARS-CoV-2-infected cells compared to uninfected control, which was prevented if cells were pretreated with Remdesivir (Fig. 5E). Virusinduced cytotoxicity and LDH release was also inhibited in Mitoxantrone-treated tissues at 24 h (Fig. 5F). At 96 h post-infection, a low concentration of Mitoxantrone continued to lower the LDH level in the basal media compared to DMSO-treated control (Fig. 5G). However, we observed a small increase in LDH level in 1 μM Mitoxantrone-treated samples (Fig. 5G). This is probably due to Mitoxantrone's intrinsic cytotoxicity (see below). Collectively, these data suggest that Mitoxantrone has a potent anti-SARS-CoV-2 activity at concentrations with low cytotoxicity. global impact of Mitoxantrone treatment on cells and to identify target inhibition that is irrelevant to viral entry, we analyzed the mRNA sequencing profile of HEK293T cells exposed to Mitoxantrone. This treatment dramatically changed the expression of many genes (Fig. 6A). Gene ontology (GO) analysis of the 828 genes upregulated by Mitoxantrone by at least twofold with adjusted p-values less than 0.05 in triplicated experiments showed that the enriched processes fall into two categories: they are either associated with nucleosome assembly or disassembly or linked to events at the plasma membranes such as cellular response to stimulus, response to nutrient levels, regulation of cell-cell adhesion, etc. (Fig. 6B). The latter is consistent with the observed interaction of Mitoxantrone with the cell surface HSPG, which presumably induces compensatory gene expression changes to maintain proper cellular responses to extracellular stimuli. On the other hand, the upregulation of histone genes and genes regulating nucleosome assembly (Fig. 6C) is consistent with the reported interaction of Mitoxantrone with nucleic acid and DNA topoisomerase (Fig. 6D), which is thought to account for the cytotoxicity of Mitoxantrone in cancer therapy. Structural analyses show that the di-hydroxy-anthraquinone moiety of Mitoxantrone forms extensive hydrophobic interactions with the DNA/topoisomerase complex (Fig. 6E,F).
Since the anti-SARS-CoV-2 activity has been linked to the two symmetric arms, we proposed that modification

Discussion
The rapid development and deployment of COVID-19 vaccines have significantly slowed down the current pandemic. Nevertheless, the evolution of new viral strains combined with the disparity in global vaccine distribution has led to new waves of infected cases and deaths. Thus, an effective and economical therapeutic agent is urgently needed, particularly in developing countries that have limited access to vaccines. Our study suggests that Mitoxantrone derivatives with reduced affinity to DNA-topoisomerase complexes may be promising anti-COVID drug candidates. Mitoxantrone has been approved for treating acute myeloid lymphoma, prostate cancer, and multiple sclerosis. The anti-cancer activity of Mitoxantrone has been attributed to its interactions with DNA-topoisomerase complexes, which inhibit DNA replication 33 . Mitoxantrone also effectively inhibits the entry of spike-pseudotyped viral particles 18 and authentic SARS-CoV-2 virions (this study) but the underlying mechanism has been unclear. Our study showed that Mitoxantrone has a high affinity for heparin/heparan sulfate, but it does not prevent the association of heparan sulfate with spike. Instead, it stabilizes spike in a complex with heparan sulfate, increasing its resistance to protease digestion. The presence of Mitoxantrone in a heparan sulfate-spike complex appears to reduce the fusogenic activity of spike, which in turn diminishes the infectivity of SARS-CoV-2. However, the precise mechanism by which Mitoxantrone inhibits spike-mediated membrane fusion remains to be elucidated. The binding of Mitoxantrone to heparan sulfate also diminishes the endocytosis of SARS-CoV-2 viral particles and many other HSPG cargos 18,21 . Whether these two activities of Mitoxantrone share a common mechanism is also unclear. Because Mitoxantrone targets both DNA topoisomerases and heparan sulfate in mammalian cells, a potential strategy for developing Mitoxantrone into an antiviral agent is to generate derivatives that only bind to heparan sulfate.
Compared to other viral entry inhibitors under development, Mitoxantrone is unique as it is a small molecule targeting a host factor required for viral entry. Thus, it has several advantages. For example, several neutralizing antibodies have been approved for use in patients with mild COVID symptoms. Many studies also reported promising anti-SARS-CoV-2 activities for nanobodies against spike in animal models [34][35][36][37][38] . Although www.nature.com/scientificreports/ antibody-based therapies are generally safe and effective, they require high doses, which makes these therapies quite expensive. Additionally, most neutralizing antibodies recognize the RBD of spike, which is rapidly evolving as the virus continues to spread. Consequently, new variants with mutated spike may not be recognized by existing antibodies. In this regard, a safe, small molecule inhibitor targeting a host factor required for viral entry may offer an alternative therapeutic option that can be used either as a single agent or in combination with spike antibodies or other anti-viral agents. Drugs targeting host factors have been successfully developed to treat other infectious diseases. For example, Maraviroc, an anti-HIV agent targeting the host membrane protein CCR5 has been approved for AIDS treatment 39 . Therefore, it is tempting to generate drugs with similar heparan sulfate-binding activities as Mitoxantrone but improved safety profile. Our structural analyses suggest that the two hydroxyl groups in the anthraquinone moiety of Mitoxantrone may be the best place to introduce modifications, which are expected to reduce its binding to DNA-topoisomerase. Given the broad role of the cell surface HSPG in pathogen entry, Mitoxantrone derivatives may even be used to treat a new viral infection that exploits the cell surface HSPG for viral attachment and entry.

Materials and methods
Reagents and plasmids. Spike protein was purchased from Sino Biologicals. Mitoxantrone was purchased from Microsource (Cat #01503278) and stored by the NCATS compound management department. Banoxantrone was purchased from Sigma (Catalog number: SML1854). pcDNA3.1-SARS-CoV-2-Spike plasmid was obtained from BEI resource 7 . pLV-Mcherry was obtained from Addgene (Catalog number: 36084). Hoechst 33342 staining solution and proteinase K were purchased from Thermo Fisher Scientific (Catalog number: 62249 and AM2546).
Heparin Sepharose beads pulldown and protease K digestion. Heparin pulldown assays were performed as previously reported 18 . Briefly, spike 800 ng was incubated with heparin or control Sepharose beads in a buffer (25 mM Tris 7.4, 2 mM magnesium chloride, 2 mM potassium chloride, 0.05% NP40) containing either no additional salt or 150 mM sodium chloride. After incubation at 4° for 20 min, bound materials were collected by centrifugation. After two quick washing with the binding buffer, bound proteins were eluted with the Laemmli buffer. Where indicated, heparin beads were first treated with 50 µM Mitoxantrone to saturate all binding sites before being incubated with spike.
For protease treatment experiment, 1.28 μg spike protein was incubated with either DMSO, Mitoxantrone (25 μM), heparin (25 μM) or Mitoxantrone together with heparin in the presence of 0.1 M of Tris-HCl (pH7.4) in pure water (220 μl of the total volume). Mitoxantrone and heparin were pre-mixed before adding to the S protein to avoid unwanted precipitations. After incubation at room temperature for 15 min, the reactants were aliquoted into 40 μl per tube. 10 μl pre-diluted proteinase K was added to make the final concentrations as 0.05, 0.067, 0.08, or 0.1 μg/ml. The reactions were further incubated for 5 min at room temperature and then quenched by pre-heated 4 × Laemmli buffer and heating. The samples were resolved by SDS-PAGE electrophoresis followed by immunoblotting.
Spike-and ACE2-mediated cell fusion assay. HEK293T target cells expressing spike protein and mCherry were generated by transfecting cells with pcDNA3.1-SARS-CoV-2-Spike and pLV-Mcherry at 1:1 ratio. Effect cells were HEK293T cells stably expressing ACE2-GFP, as described previously 18 , or as a control, HEK293T cells transfected with pEGFP. To initiate cell-cell fusion, effect cells and target cells were mixed at 1:1 ratio in a live cell imaging chamber and placed in the incubator for 90 min. Cells were then fixed by 4% paraformaldehyde in phosphate buffer saline and then stained with a Hoechst 33342 staining solution to label nuclei. Cell were imaged by a Zeiss LSM780 confocal microscopy.

SARS-CoV-2 infection on cells. 7 × 10 4 Vero E6 cells (ATCC #CRL-1586) were seeded into an Millicell
EZ SLIDES chamber slides (Millipore) 24 h before infection. On the day of infection, cells were pre-treated with 200 nM mitoxantrone or as a control with DMSO for 30 min. After 30 min treatment, Vero E6 cells were infected with live SARS-Cov-2 (USA-WA1/2020) (ATCC #NR-52281) at an MOI of 0.1 for 5 h at 37 °C, 5% CO2. Cells were then washed, fixed with 4% paraformaldehyde. The viral spike protein was stained with an in-house developed rabbit anti-spike antibody and goat anti-rabbit IgG (H + L) highly cross-adsorbed antibody with Alexa Fluor 488 (Invitrogen). Images were acquired using FluoView FV10i Confocal Laser Scanning Microscope (Olympus). All live virus infections were performed in a biosafety level 3 (BSL3) laboratory as approved by the Institutional Biosafety Committee (IBC).

SARS-CoV-2 infection in a 3D EpiAirway model. Infection in the 3D EpiAirway model was carried
out by University of Louisiana as a contracted service. In general, human tracheobronchial epithelial cells (Epi-Airway™ from MatTek) were culture on inserts at ALI in 6-well plates. Prior to addition of drugs or virus, accumulated mucus from the tissue surface were removed by gently rinsing the apical surface twice with 400 μl TEER buffer. All fluids from the tissue surface were carefully removed to leave the apical surface exposed to the air. Mitoxantrone and control compounds were diluted into the assay medium (AIR-ASY-100). Tissues were pretreated with compounds for 1 h on the apical and basal sides. After compound pretreatment, liquid was removed from the apical side, and virus (2019-nCoV/USA-WA1/2020 at MOI of 0.1) was inoculated in 0.15 mL assay medium onto the apical layer for 1 h. Following 1 h inoculation, virus-containing media was removed from apical layer and the apical side was washed with 400 μl TEER buffer. The basolateral medium was replaced TCID 50 (50% tissue culture infectious dose) assay. Human bronchial epithelial cells (HBEC's 3D-Epi-Airway™) were seeded into culture inserts for 6-well plates one day before viral infection. Prior to addition of drugs or virus, accumulated mucus from the tissue surface were removed by gently rinsing the apical surface twice with 400 μl TEER buffer. All fluids from the tissue surface were carefully removed to leave the apical surface exposed to the air. Mitoxantrone was diluted into the assay medium and placed at room temperature before co-treatment with virus (MOI: 0.1) onto apical layer and basal layer for 1 h. Following 1 h treatment, virus was removed from apical layer and basolateral medium was replaced with fresh maintenance medium and compound at 24 h, 48 h and 72 h post-infection. At 24 h and 96 h post-infection, the apical layer was washed with 0.4 mL of TEER buffer (PBS with Mg 2+ and Ca 2+ ) and aliquoted to separate microfuge tubes (1.5 mL). Eight-fold serial dilutions of apical layer supernatant sample concentrations were added to 96-well assay plates containing Vero E6 cells (20,000/well). The plates were incubated at 37 °C, 5% CO 2 and 95% relative humidity. Following 3 days (72 ± 4 h) incubation, the plates were stained with crystal violet to measure cytopathic effect (CPE). Virus titers were calculated using the method of Reed and Muench (Reed et al., 1938). The TCID 50 values were determined from duplicate samples. LDH assay. Medium from the basolateral layer of the tissue culture inserts was removed 24-and 96-h postinfection and diluted in LDH Storage Buffer as per the manufacturer's instructions (LDH-Glo Cytotoxicity Assay, Promega). Samples (5 μl) were further diluted with LDH Buffer (95 μl) and incubated with an equal volume of LDH Detection Reagent. Luminescence was recorded after 60 min incubation at room temperature. A no cell control was included as a negative control to determine culture medium background and bleomycin included as a positive cytotoxic control.

Gene expression and structural analyses.
To study the Mitoxantrone's effect on gene expression in HEK293T cells, we analyzed the mRNA sequencing data from our previous study, which is stored in the NCBI sequence read archive with accession ID: PRJNA645209. We used the web-based String protein interaction network program to analyze significantly up-or down-regulated genes. The identified networks were exported into Cytoscape for graph-making. The complex of Mitoxantrone with DNA-topoisomerase was analyzed by LigPlot 40 to reveal the interaction between Mitoxantrone and the enzyme.
To determine how Mitoxantrone interacts with DNA and topoisomerase, we used LigPlot to analyze the molecular interactions present in the published structure (PDB: 4g0v) 33 .
Image processing and statistical analyses. Confocal images were processed using the Zeiss Zen software. To measure fluorescence intensity, we used the open-source Fiji software. Images were converted to individual channels and regions of interest were drawn for measurement. Statistical analyses were performed using either Excel or GraphPad Prism 9. Data are presented as means ± SEM, which was calculated by GraphPad Prism 9. P values were calculated by Student's t-test using Excel. Nonlinear curve fitting and IC 50 calculation was done with GraphPad Prism 9 using the inhibitor response three variable model or the exponential decay model. Images were prepared with Adobe Photoshop and assembled in Adobe Illustrator. Data processing and reporting are adherent to the community standards. Uncropped gel and immunoblot images are shown in supplementary Figs

Funding
Open Access funding provided by the National Institutes of Health (NIH).