Sulfated endospermic nanocellulose crystals prevent the transmission of SARS-CoV-2 and HIV-1

Biomaterials with antimicrobial activity are gaining attention due to their biodegradability and efficacy in interacting with a wide range of microorganisms. A new cellulose nano-biomaterial, endospermic nanocellulose crystals (ENC) obtained from parenchymal tissue of ivory nut endosperm, has a natural capacity as a universal binder. This feature is enhanced when it is chemically functionalized, and can be exploited in the fight against microbes. We tested the ability of sulfated ENC in aqueous suspension to encapsulate viruses through a crosslinking reaction mediated by cations. 0.25% w/v ENC suspensions efficiently encapsulated spike (S) protein, preventing its interaction with ACE2 receptor. ENC was further able to encapsulate SARS-CoV-2 pseudoviruses and prevent infection of 293T-hsACE2 cells. ENC also suppressed infection of MT-4 cells with HIV-1LAI.04. This antiviral activity of sulfated ENC is due to the irreversible interaction of ENC with viral particles mediated by crosslinking, as antiviral activity was less effective in the absence of cations. Additionally, ENC was used as a matrix to immobilize recombinant ACE2 receptors and anti-S IgG, creating molecular lures that efficiently inhibited SARS-CoV-2 infections in vitro. These results show that sulfated ENC from ivory nuts can be used as an efficient antiviral material.

The primary mode of SARS-CoV-2 transmission is inhalation of respiratory droplets, thus the use of masks is recommended to limit transmission. Numerous epidemiologic studies have shown that masking reduced transmission 1 . Air filtration by HEPA filters also reduced virus in the air by 65-90% 2 . One method to further reduce viral transmission would be to develop materials that combine filtration with antimicrobial properties.
Towards this goal we proposed the use of cellulose nanoparticles to encapsulate viral particles thus preventing their passage. These nanocellulose particles were designed with dense sulfate charges obtained from ivory nuts from a palm tree from the Arecaceae family that grows in tropical regions of South America. Due to its derivation from the endosperm of the palm´s seed, it was designated endospermic nanocellulose (ENC). The unique primary cell walls yielded more homogeneous and thinner cellulose nanoparticles compared to cell walls from other vegetal tissues. The thinner the nanoparticles, the higher the surface area of the suspended material to interact with the viruses. This material could be used as a film for masks and filters to bind viral particles, increasing the effectiveness of these filtration devices.
Here, we demonstrated two different methods of blocking transmission of virus using sulfonated ENC. One was non-specific that consisted of the direct binding (nanoencapsulation) of the SARS-CoV-2 pseudovirus particles, or encapsulation of 293T-hsACE2 target cells. ENC also efficiently encapsulated HIV-1 virions and prevented transmission to MT-4 target cells. The second method was highly specific using molecular lures, anti-SARS-CoV-2 spike (S) protein antibodies and soluble ACE2 receptors, immobilized in nanocellulose scaffolding to specifically bind viral particles.

Results
ENC encapsulation of S protein prevented binding to ACE2 in ELISA tests. To determine whether ENC can encapsulate SARS-CoV-2 S protein thus preventing interaction with ACE2, and to establish an optimal working concentration of ENC, we used the quantitative SARS-CoV-2 neutralizing antibody ELISA with modifications. Spike protein labeled with horse radish peroxidase (HRP) was preincubated with ENC suspension prior to its addition in ELISA wells then combined with biotinylated ACE2 receptor, which attaches to avidin coated plate wells. HRP-labeled S protein that binds ACE2 within the well was then measured after addition of HRP substrate. HRP enzymatic activity was directly proportional to the amount of S protein that escapes ENC. A diagram of the assay is depicted in Fig. 1a.
When S protein was preincubated in 0.27% w/v ENC before addition to the assay, the absorbance units (AU) obtained were below the limit of detection (0.142 ± 0.008 AU compared to lowest standard measurement of 0.291 ± 0.053 AU, n = 3). This indicated that S protein was irreversibly bound by ENC and was not able to react with ACE2. Validity of the assay was confirmed by use of two positive controls containing anti-SARS-CoV-2 S IgG provided with the kit and each gave measurements in the expected range provided with the assay instructions.
To determine the optimal ENC concentration for nanoencapsulating S protein, one experiment was run using a series of dilutions of ENC (0.5, 0.4, 0.33, 0.27, 0.16, and 0.1)% w/v with the addition of 0.9% w/v NaCl solution. The lower the absorbance, the more effective the S protein nanoencapsulation. According to the absorbance curve, we determined that 0.27% w/v ENC suspension was the optimal concentration (n = 3, Fig. 1b).

AFM and fluorescence confocal microscopy demonstrated encapsulation of S protein in ENC.
To demonstrate at the nanoscopic scale the interactions and crosslinking of ENC particles around the S protein clumps we used atomic force microscopy (AFM). This technique revealed intricate networks of ENC that were denser (bright spots) when attached to S protein clumps (Fig. 2a). This is due to a higher affinity of ENC to protein clumps than to other ENC particles caused by anionic interactions between the cellulose nanoparticles and protein positive moieties.
Confocal microscopy was employed to demonstrate at the micro scale the agglutination of S protein in ENC matrix, and to verify the identity of the presumed S protein clumps seen by AFM. The protein interaction with ENC was evident even after a series of washing cycles, suggesting the nanoencapsulation of protein clumps is irreversible. To demonstrate that, hydrogels encapsulating Alexa Fluor 488 (AF488) labeled S protein were crosslinked, washed in triplicate, and dried followed by confocal microscopy analysis. Figure 2b shows the same

SARS-CoV-2 S pseudovirus cell entry was inhibited by ENC.
To test whether ENC can prevent virus transmission, 293T-hsACE2 cells were inoculated with SARS-CoV-2 pseudovirus (pv) with a green fluorescent protein (GFP) reporter in conditions with and without ENC and analyzed by flow cytometry three days later. SARS-CoV-2 pv entrapped in ENC or captured by recombinant ACE2 protein (ACE2) or anti-SARS-CoV-2 S IgG (IgG) entrapped in ENC was not able to infect cells: 0.003 ± 0.003%, 0.007 ± 0.004%, and 0.022 ± 0.010% of infected cells respectively (p < 0.0001), (n = 16, Fig. 3a and dot plots in Suppl. Fig. 1). Meanwhile, for controls in which SARS-CoV-2 pv was added without ENC, the number of cells infected was 1.76 ± 0.08% (n = 16). Preincubation of SARS-CoV-2 pv with either ACE2 or IgG did not significantly change virus transmission (1.94 ± 0.19% and 1.28 ± 0.10% infected cells, respectively, Fig. 3a). The low level of infection with SARS-CoV-2 pv entrapped by ENC or captured by ACE2 or IgG in ENC was not different than background in uninfected controls (n = 9, p = 0.388-0.993). These results were confirmed by fluorescent microscopy as well, where a similar pattern of GFP + cells was observed (Suppl. Fig. 2a-f).
We next evaluated whether cells could be protected from infection by entrapping the cells in ENC. Cells were first embedded in ENC and then inoculated with SARS-CoV-2 pv in culture medium or SARS-CoV-2 pv in ENC. SARS-CoV-2 pv infection was evaluated by fluorescent microscopy. Cells were cultured as above then the gels were transferred to microscope slides at day 3 post-inoculation, stained with propidium iodide, cover-slipped and The assay was modified to measure the ability of ENC to block the interaction of S protein with ACE2, rather than the ability of anti-S neutralizing antibody to block this interaction; (b) The modified ELISA was used to determine optimum % of ENC suspension (expressed in w/v) to efficiently encapsulate S protein; lower absorbance equals higher S protein encapsulation (n = 3).  Long-term direct cell exposure to ENC had a transitory effect on cell viability and proliferation. To confirm that cells were viable, and that lack of infection in ENC conditions was not due to massive cell death, we evaluated cell viability in the above experiments by flow cytometry using AlexaFluor350 dye, which is taken up by dead cells. This analysis revealed small but significant decreases in cell viability in ENC conditions. The percentage of viable cells in positive controls (cells + SARS-CoV-2 pv) was 98.29 ± 0.23% (n = 16, Fig. 4a) and cells in ACE2 + SARS-CoV-2 pv, IgG + SARS-CoV-2 pv, and ENC + IgG + SARS-CoV-2 pv conditions were similar to positive controls with 96.76 ± 0.88%, 98.11 ± 0.31% and 94.08 ± 1.49% viable cells, respectively. The percentages of viable cells in ENC + SARS-CoV-2 pv and ENC + ACE2 + SARS-CoV-2 pv conditions were statistically different from positive controls with 94.41 ± 1.12% (p = 0.010), 92.44 ± 1.66% (p = 0.006) viable cells, respectively.

ENC prevented transmission of HIV-1 into target cells. To test whether ENC prevented infection
of other viruses, we next assessed its ability to entrap HIV-1 LAI.04 . We infected MT-4 cells with HIV-1 LAI.04 in the presence of ENC. HIV-1 was diluted in NaCl and added either to culture medium or to ENC, and then applied to MT-4 cells for 4 h, washed three times in PBS, and returned to culture. HIV-1 infection was evaluated 4 days post-infection by flow cytometry measurement of p24 + cells. In control experiments without ENC, HIV-1 infected 27.27 ± 2.62% of cells (n = 16, Fig. 6). In contrast, 0.05 ± 0.01% of cells were infected when HIV-1 was entrapped in ENC which was significantly less than in the controls (p < 0.0001) and not significantly above background staining of uninfected cells (0.04 ± 0.01% infected cells, p = 0.704).
The ability of ENC to completely entrap SARS-CoV-2 pv was mediated by sodium chloride. We next examined the mechanism of ENC virus entrapment. It is known that sulfonation of nanocellulose results However, the percentage of infected cells was higher than in the above experiments where NaCl was used for crosslinking the ENC and infection was almost completely suppressed.

Discussion
The current study reports on a nanocellulose that is capable of efficiently preventing viral infection. Nanocellulose derived from ivory nut endosperm yields small nanoparticles (1-5 nm) which provide high surface area for entrapment of microbes. Here, we demonstrate ENC as a universal binder that prevents viral infection of cells. ENC was shown to nanoencapsulate S protein of SARS-CoV-2, as well as whole SARS-CoV-2 pv and HIV-1 virions. Moreover, ENC demonstrated the ability to immobilize IgG and ACE2 recombinant protein that were utilized with high specificity as molecular lures to bind and immobilize SARS-CoV-2 pv. ENC suspensions, when crosslinked, form hydrogels that result from the entanglement of nanoparticles mediated by electrostatic stabilization associated with positively charged molecules such as sodium, calcium and potassium ions. This crosslinking causes sol-to-gel transitions that are irreversible. The ENC particle aggregation that occurs during sol-to-gel transitions is a phenomenon ubiquitous in colloidal and aerosol systems, including nanocellulose aqueous suspensions. Upon dispersion, particles collide and often irreversibly stick together to form larger clusters of a fractal-like morphology 19 . The crosslink density of the gels is dependent on the concentration of nanoparticles, while the pore size of the matrix is an inverse function of the concentration 20 . Different chemical routes of surface functionalization of the nanocellulose can provide ionic surfaces via phosphorylation, carboxymethylation, oxidation, and sulfonation, or promote hydrophobic interactions through acetylation, etherification, silylation, urethanization, and amidation 21 . Here we used sulfonated ENC which allows electrostatic interactions between the cellulose's active groups and charged moieties in proteins such as those on the surface of viruses.
The feasibility of ENC as an inhibitor of viral transmission was first verified by ELISA, which demonstrated effective blocking of SARS-CoV-2 S protein interaction with ACE2 in the presence of ENC, and allowed us to identify optimal working concentrations of ENC. AFM and confocal microscopy validated these results, showing entrapment of S protein in ENC. Previous studies have shown that nanocelluloses are effective membrane filters to remove viruses but they have not been employed to demonstrate prevention of viral infection 22 . Antiviral activity of alginate-based 5 or sialic acid 6 materials have been shown, but the modes of action and applications are different from those of ENC. Our study demonstrates the ability to block infection whether SARS-CoV-2 pv was directly entrapped in ENC, or whether molecular lures (ACE2 or anti-SARS-CoV-2 IgG) were first entrapped and then allowed to bind SARS-CoV-2 pv. Molecular lures have been used to trap both viruses in general or specific viruses [23][24][25] , but the ability to design lures in ENC described here provides an alternate biocompatible method of entrapment. ACE2 and anti-SARS-CoV-2 IgG added directly to the culture medium were not effective at preventing infection. This may be due to the fact that these experiments were performed with conditions that maximize the chance of infection; therefore, the design did not allow for optimal inhibition with ACE2 or IgG, which typically would require a short incubation time between cells and virus followed by removal of virus.
Also, infection was prevented when cells were embedded in ENC. This also points to the universal binding property of ENC, as it can entrap not only microorganisms, but also eukaryotic cells. This expands the possibilities for other applications of ENC.
Exposure of cells to ENC for three days resulted in small decreases in cell viability and of proliferation indexes. However, these effects were transient as demonstrated by cell transfer assays. If cells were removed from ENC conditions after 4 h or overnight incubation with ENC, their proliferation indexes returned to normal. Although www.nature.com/scientificreports/ we did not address the mechanism for these decreases, we believe it may be related to insufficient gas exchange or nutrient transfer through the relatively thick ENC gels on top of the cells. In future applications of this technology, it is unlikely that cells will be directly exposed to thick layers of ENC for long periods of time. In general, nanocellulose based biomaterials have high biocompatibility and low toxicity thus they are being developed for many biomedical applications 26 . The ability of ENC to prevent viral transmission was not unique to SARS-CoV-2, as experiments with infectious HIV-1 demonstrated a similar effect. HIV-1 infection was almost completely suppressed when virus was entrapped in ENC. Thus, with a second virus, we showed the ability to block infection of cells by entrapping virus in ENC. Previous studies have shown the ability of various types of nanocellulose to efficiently remove viruses using membrane filtration 22 but here we not only trap virus in ENC, but also demonstrate ENC biological effectiveness.
Taken altogether, these results demonstrate wide potential applicability of ENC that can bind individual proteins (S protein, ACE2, IgG), different virus particles (SARS-CoV-2 pv and HIV-1) and whole cells. The electrostatic interactions between the functionalized ENC and proteins form irreversible crosslinked hydrogels. Nanocellulose is being explored for many biomedical applications due to its ability to immbolize proteins while preserving their structural integrity and enhancing their bioactivity and stability 26 . To demonstrate the mechanism of entrapment by ENC, we established the importance of positively charged molecules, in this case NaCl, in crosslinking the ENC to permanently entrap particles. When NaCl was omitted from ENC preparations, some viral particles were able to escape the ENC scaffold and infect cells.
Our study does have several limitations. We assessed only a few proteins and viruses, and entrapment was assessed only over a relatively short period of time. Also, for future applications, more elaborate characterization of ENC will be required. Potential risks associated with the use of nanoparticles are a concern for bioaccumulation and other potential subcellular interactions. For instance, in in vivo experiments with respirable nanocellulose and cellulose dust, one month of exposure resulted in granulomatous inflammation, fibrosis, and albeobronchiolitis 27 . In contrast, the ENC particles utilized in this work, due to their dense negative charge provided by SO 3 groups from sulfuric acid synthesis, instantaneously crosslink in the presence of positive ions and molecules, changing from nanostructured materials to macrostructured scaffolds which are unable to penetrate tissues or be internalized by phagocytes. Rather, they can embed cells, viruses, and molecules. ENC particles act as building blocks that are assembled together into large crosslinked structures.
ENC is particularly well suited as a universal binder due to its high surface area for entrapment 7 . The ability of ENC to encapsulate viral particles thus preventing their ability to infect target cells, as well as to also entrap proteins and cells, demonstrate the usefulness of this biomaterial for development of new strategies to prevent viral transmission. This material has potential applications as a film for masks and filters, to bind viral particles and increasing the effectiveness of these filtration devices, or to create adsorbent surfaces to capture or inactivate viruses. The concentrations of ENC used in these experiments will lend themselves easily to aerosilization making www.nature.com/scientificreports/ them ideal for such purposes. Nanoparticles that are deemed safe for in vivo use also have potential to enhance bioavailablility of natural compounds, or for drug delivery.

Materials and methods
Nanocellulose synthesis. ENC was synthesized from ivory nut endosperm by grinding Tagua seeds to a particle size of < 200 µm then subjecting the flour (20 g) to acid hydrolysis in 8 M sulfuric acid at 1:10 ratio (grams of flour: grams of sulfuric acid) at 60 °C for 2.5 h at 300 rpm. The reaction was stopped with an equal volume of deionized water. The suspension was collected in a 50 ml tube, centrifuged at 3000 rpm for 10 min, supernatant discarded, and the pellet resuspended in deionized water and dialyzed for one week, until reaching pH 5.
Adjustment of ENC w/v percentage. ENC percentage w/v in the hydrogel was determined on three samples of 10 ml of hydrogel dried at 80 °C for 48 h. The dried ENC was weighed to calculate % w/v then diluted to 0.25% w/v and stored at 4 °C.
Titration of SO3groups. Conductimetric titration was used to measure -SO 3 Hgroups of ENC that result from acid hydrolysis according to previously published methods 28 . The sample was prepared as follows: 2.5 g of sample suspension containing 0.803% w/v ENC was suspended in water. The sample was diluted in 120 ml of 1 mM NaCl solution, and neutralized with NaOH (2 mM) added in 0.1 ml aliquots at 30 s intervals 29 . The sulfate group's content was determined after three subsequent measurements 30 . The average value of volume of NaOH and conductometry were taken in order to plot the titration curve. The ester sulfate content of the samples was calculated using the equation: X = C t .V 1 /m where: X is the sulfate group's content in μmol/g; C t is the concentration of the NaOH solution, in μmol/l; V 1 is the volume of the NaOH solution consumed at the first intersection point in litres, and m is the oven-dry weight of sample in grams.  . NaCl was critical for crosslinking and capture of virus in ENC. 293T-hsACE2 cells were inoculated with SARS-CoV-2 pseudovirus (pv) with a GFP reporter in conditions with and without endospermic nanocellulose (ENC) and molecular lures (ACE2 or anti-SARS-CoV-2 S IgG (IgG)) and incubated for 3 days, as in previous experiments except that NaCl was replaced with culture medium in preparation of reagents. Cells were harvested for flow cytometry analysis and % of GFP infected cells was displayed for each condition (n = 16, *p < 0.05, **p < 0.01, ***p < 0.001, ANOVA, mean ± SEM). ENC reduced but did not completely inhibit infection as in experiments where ENC was crosslinked by NaCl. Atomic force microscopy. 20 µl of ENC was deposited on a clean glass slide and 5 µl of different concentrations of SARS-CoV-2 S protein were dropped on top. The mix was left to interact then dry in ambient conditions. Atomic force microscopy (AFM) imaging was performed in air on a Bioscope Catalyst AFM (Brukernano, Santa Barbara, CA), in the "tapping" mode, using silicon cantilevers with nominal stiffness of 2.8 N/m and resonance frequency of 75 kHz (FESP cantilevers by Bruker-nano). Acquired images were post-processed using the instrument software (Nanoscope Analysis, v.2.0, Bruker-nano).

ELISA tests.
Fluorescence confocal microscopy. Images of dry membranes resulting from pre-washed and dried ENC hydrogels encapsulating fluorescent S protein were acquired with an Olympus FV3000RS laser scanning confocal microscope using Peltier-cooled GaAsP detectors. Detector emission was automatically optimized for AlexaFluor 488 using a spectral system comprised of a volume phase holographic spectral transmission grating and adjustable slit. Images were collected with 30X silicon immersion objective (UPLSAPO30XS; NA 1.05; Olympus Tokyo). Spots on the bottom of a 35 mm glass bottom dish were acquired using multi-area time-lapse (MATL). Image fields were stitched with Olympus Fluoview software (FV31S-SW, Tokyo, Japan). Pixels were sampled by the Nyquist criterion. Solutions were incubated for 30 min at room temperature, added to cells and incubated for three days at 37 °C, 5% CO 2 . Pseudoviral entry was observed by fluorescent microscopy and flow cytometry. Identical experiments were performed except that NaCl in all solutions was replaced with culture medium.

Cell lines and pseudoviruses/viruses. 293T-hsACE2 cells (Integral
Cells to be embedded in ENC were resuspensed in NaCl with 10 μl cells at 4 × 10 6 cells/ml added to 90 μl ENC for each well, and pv was added either as free pv or ENC encapsulated pv (10 μl pv in NaCl in 90 μl of medium or ENC).
For transfer experiments, cells were harvested after 4 h or overnight incubation by trypsinization with 0.25% Trypsin-EDTA and centrifugation at 400xg for 5 min. Cells were washed twice with PBS, centrifuged, and cultured for the remainder of the 3 days in new plates.

Fluorescence microscopy and flow cytometry. Pseudovirus infection was monitored by visualization
of GFP positive cells after three days on a fluorescent microscope (Keyence BZ-X810, Itasca, IL) equipped with filter cubes for GFP, Texas Red and DAPI. For cell embedding conditions, gels were transferred to microscope slides, stained with propidium iodide (Thermo Fisher, Waltham, MA, 1 μg/ml in PBS), coverslipped and imaged.
Following imaging, cells were harvested by trypsinization, washed and labeled with Alexa Fluor 350 dye (ThermoFisher, 1 μg/ml) for viability, and acquired on a Symphony A5 flow cytometer (BD Biosciences, Franklin Lakes, NJ). Flow cytometry analysis included size gating on FSC vs SSC plots, single cells on FSC-A vs FSC-H plots, live cells by viability dye exclusion with BUV450 vs SSC, and BB515 staining for percent of GFP + cells. Analysis was performed in FlowJo v10.8 (BD Biosciences).
For proliferation, cells were prelabeled with Celltrace™ Violet (ThermoFisher) by suspension in 1 ml of PBS with 1 μl of dye, incubated for 20 min at 37 °C, and washed, with an aliquot fixed immediately. Cells were analyzed by flow cytometry with gating as above, followed by BV421 histograms for the proliferation modeling tool in FlowJo. Proliferation index (PI) was based on dye loss and was calculated as the total number of divisions divided by the number of cells that went into division. www.nature.com/scientificreports/ For determining HIV infection, cells were labeled with viability dye, fixed and permeabilized (Fix and Perm™ cell permeability kit, ThermoFisher), then stained with mouse anti-p24 (KD57-RD1, Beckman Coulter, Brea, CA), washed and fixed in 2% paraformaldehyde. Cells were acquired on flow cytometer and p24 + infected cells were identified. P24 measurement by Luminex. SARS-CoV-2 pseudotyped lentiviral stocks and HIV-1 LAI.04 were quantified using a p24 cytometric bead assay 31 .
Statistical evaluation. Data normality was tested using D' Agostino and Pearson test or Shapiro-Wilk test when n was small. Normal data were analyzed by one-way analysis of variance and presented as mean ± standard error of the mean (SEM). Data not normally distributed were analyzed by Kruskal-Wallis Rank test and presented as median with interquartile range. GraphPad Prism v9.2.0 (GraphPad Software, San Diego, CA) was used for statistics with p < 0.05 considered statistically significant.

Data availability
All data generated or analyzed during this study are included in this published article (and its Supplementary  Information files).