Silicon nitride: a potent solid-state bioceramic inactivator of ssRNA viruses

Surface inactivation of human microbial pathogens has a long history. The Smith Papyrus (2600 ~ 2200 B.C.) described the use of copper surfaces to sterilize chest wounds and drinking water. Brass and bronze on doorknobs can discourage microbial spread in hospitals, and metal-base surface coatings are used in hygiene-sensitive environments, both as inactivators and modulators of cellular immunity. A limitation of these approaches is that the reactive oxygen radicals (ROS) generated at metal surfaces also damage human cells by oxidizing their proteins and lipids. Silicon nitride (Si3N4) is a non-oxide ceramic compound with known surface bacterial resistance. We show here that off-stoichiometric reactions at Si3N4 surfaces are also capable of inactivating different types of single-stranded RNA (ssRNA) viruses independent of whether their structure presents an envelop or not. The antiviral property of Si3N4 derives from a hydrolysis reaction at its surface and the subsequent formation of reactive nitrogen species (RNS) in doses that could be metabolized by mammalian cells but are lethal to pathogens. Real-time reverse transcription (RT)-polymerase chain reaction (PCR) tests of viral RNA and in situ Raman spectroscopy suggested that the products of Si3N4 hydrolysis directly react with viral proteins and RNA. Si3N4 may have a role in controlling human epidemics related to ssRNA mutant viruses.

www.nature.com/scientificreports/ by plaque assay (Influenza A H1N1) or TCID50 assay (FCV and HEV71). A schematic draft of the inoculation and subsequent testing/characterization procedures is offered in the Supplementary Information. For the plaque assay, a confluent monolayer of MDCK cells in a 6-well plate was washed twice with serumfree DMEM (SF DMEM), followed by infection with 100 μl of virus suspension in a tenfold serial dilution. After incubation at 37 °C for 1 h with tilting every 10 min, unabsorbed inoculum was removed, and infected cells were overlaid with 4 ml of DMEM containing 2.5 μg/ml trypsin (Sigma-Aldrich Co. LLC, Saint Louis, USA) and 0.2% albumin (Wako Pure Chemical Industries, Ltd., Osaka, Japan). The plate was incubated at 37 °C in an atmosphere of 5% CO 2 for 2 days. For plaque counting, cells were fixed with 5% glutaraldehyde solution for 2 h, the agarose medium was removed, and the cells were stained with 1% crystal violet.
After removal of the culture supernatant, a confluent monolayer of CRFK or LLC-MK2 cells in a 96-well plate was inoculated with 50 μl of virus suspension in a tenfold serial dilution. After incubation at 37 °C for 1 h with tilting every 10 min, unabsorbed inoculum was removed, and infected cells were added with 100 μl of DMEM containing 4% FBS, 0.2% albumin and 2.5 μg/ml of trypsin. The plate was incubated at 37 °C in an atmosphere of 5% CO 2 for 6-8 days. Cytopathic effect (CPE) was observed in all wells, and TCID 50 calculated by the Reed-Muench method. Statistical analyses of the results were carried out by means of the unpaired Student's t-test. The statistical significance of the data was evaluated as highly statistically relevant (p < 0.01; labeled with two asterisks), statistically relevant (p < 0.01; labeled with one asterisk), or non-significant (n.s.).
In order to confirm the infectivity of viruses exposed or unexposed to Si 3 N 4 particles, we visualized MDCK cells inoculated with Influenza A virions exposed and non-exposed to Si 3 N 4 powder in the fluorescence microscope after washing the infected cells with TBS (20 mM Tris-HCl pH 7.5, 150 mM NaCl), fixing them with 4% paraformaldehyde for 10 min at RT, and permeabilizing them with 0.1% triton X in TBS for 5 min at RT. Successively, the cells were blocked with 2% skim milk in TBS for 60 min at RT, and stained with mouse antiinfluenza A virus nucleoprotein antibody (red) (AA5H ab20343, Abcam, Cambridge, UK) (Dilution = 1:500) and fluorescence signals from cell F-actin (green) (Thermo Fisher Scientific) (Dilution = 1:500) for 60 min at RT. After washing with a washing buffer, cells were incubated with an Alexa 549 Goat Anti-mouse IgG F(ab')2 (Thermo Fisher Scientific, MA, USA) (Dilution = 1:250) for 60 min at RT in the dark. Three samples each were prepared for Si 3 N 4 exposed and non-exposed tests, respectively. Three visual fields were observed of each sample under a fluorescence microscope (BZX710; Keyence, Osaka, Japan), subsequently the total number of cells and the number of infected cells were calculated using Keyence BZ-X Analyzer. Statistical analyses of the results were carried out by means of the unpaired Student's t-test (p < 0.001; labeled with three asterisks).
The RNA of viruses exposed or unexposed to Si 3 N 4 particles was extracted using a QI Amp Viral RNA Mini Kit (QIAGEN N.V., Hilden, Germany) according to the manufacturer protocol. Reverse transcription was performed using a ReverTra Ace qPCR RT Master Mix (TOYOBO CO., LTD., Osaka, Japan). 8 μl of RNA template was added to 2 μl of 5 X RT Master Mix. The mixture was incubated at 37 °C for 15 min, 50 °C for 5 min, 98 °C for 5 min and 4 °C hold.
We selected primers and probes for Influenza A H1N1 virus based on genomic regions highly conserved in various subtypes and genotypes of the Influenza virus A (matrix protein gene). The fluorogenic probe for the Influenza A virus consisted of oligonucleotides with the 5′ reporter dye 6-carboxyfluorescein (FAM) and the 3′ quencher dye 6-carboxytetramethylrhodamine (TAMRA). A 20-μl PCR was performed using 2 μl of cDNA, 10 μl of KAPA PROBE FAST qPCR Master Mix (KAPA BIOSYSTEMS, Massachusetts, USA) containing ROX, 900 nM each influenza virus A primer, and 100 nM each probe. Amplification and detection were performed with an ABI StepOnePlus Real Time PCR System (under the following conditions: 1 cycle of 20 s at 95 °C, 45 cycles of 1 s at 95 °C and 20 s at 60 °C. During amplification, the System detector monitored real-time PCR amplification by quantitatively analyzing fluorescence emissions. The reporter dye (FAM) signal was measured against the internal reference dye (ROX) signal to normalize for non-PCR-related fluorescence fluctuations occurring from well to well. The threshold cycle represented the refraction cycle number at which a positive amplification reaction was measured and was set at 10 times the standard deviation of the mean baseline emission calculated for PCR cycles 3 to 15.
In situ Raman spectra were collected using a highly sensitive instrument (LabRAM HR800, Horiba/ Jobin-Yvon, Kyoto, Japan) with a 20 × optical lens. The spectroscope operated in microscopic measurement mode with confocal imaging in two dimensions. A holographic notch filter within the optical circuit was used to efficiently achieve high-resolution spectral acquisitions. A spectral resolution of 1.5 cm −1 was obtained using a 532 nm excitation source operating at 10 mW. The Raman emission was monitored by means of a single monochromator connected to an air-cooled charge-coupled device (CCD) detector (Andor DV420-OE322; 1024 × 256 pixel). The acquisition time was fixed at 10 s. Thirty spectra were collected and averaged at each analysis timepoint. Raman spectra were deconvoluted into Gaussian-Lorentzian sub-bands using commercially available software (LabSpec 4.02, Horiba/Jobin-Yvon, Kyoto, Japan).

Investigated ssRNA viruses
Three ssRNA viruses were investigated, which were purposely selected with different structural and isoelectric point (IEP) characteristics, as follows: Influenza A/Puerto Rico/8/1934 H1N1 (-ssRNA; enveloped; IEP = 7.5; referred to as Influenza A H1N1, henceforth), Feline calicivirus (FCV; + ssRNA; non enveloped; IEP = 3.9), and Enterovirus 71 (EV-A71; + ssRNA; non enveloped; IEP = 10.9). These three viral entities were selected because of their different genomic and virion structures, surface charge, and because of their propensity to drift. The individual molecular composition of the capsid, envelope proteins, and RNA of the selected viruses conferred on them distinct specificity and infectivity characteristics. More detailed descriptions of the used viruses with related references are given in the Supplementary Information.

Si 3 N 4 powder and its surface reactions in aqueous solution
A scanning electron microscopy (SEM) image of the Si 3 N 4 powder is shown in Fig. 1a. The observation revealed a relatively homogeneous and roundish morphology of the Si 3 N 4 powder with an average diameter of 6 μm and a standard deviation of ± 1.6 μm. The micrometric Si 3 N 4 particles were in turn composed of smaller elongated crystallites, which imparted to the particles a relatively rough surface in the sub-micrometric scale, thus increasing the specific surface area that potentially interacts with the virions. X-ray photoelectron spectroscopy (XPS) characterizations of the Si 3 N 4 powder are shown in Fig. 1b (Fig. 1c, d, respectively), but the former edge also revealed a minor fraction of N sites bonded to Si vacancies (cf. N-Si (v) sub-band in Fig. 1c). The set of XPS data shows that the solid surface of Si 3 N 4 terminates with amphoteric silanols (Si-OH) and basic secondary amine (Si 2 -NH) groups. An additional set of surface chemistry and stoichiometry characterizations of the micrometer sized Si 3 N 4 powder used in this study is given in the Supplementary Information ( Fig. S-1). Data collected by attenuated total reflection Fourier transform infrared (ATR-FTIR) and cathodoluminescence (CL) spectroscopies basically confirmed the XPS data in showing that the surface of Si 3 N 4 is a zwitterionic one, being mainly composed of amphoteric silanols and basic silylamine surface groups with minor fractions of nitrogen vacancy sites, V N 3+ , Si dangling bonds (N 3 Si), N dangling bonds (Si 2 N), and N 4 + defects associated with the presence of N-N bonds 16 .
In aqueous environment dissociation/protonation reactions take place, according to which dissociation of amine surface sites and the subsequent formation of Si-OH bonds result in the separation of ammonium ions (NH 4 + ) or ammonia (NH 3 ) and in the creation of additional surface silanols 17 . According to ab initio calculations 18,19 , the most energetically favorable path for Si 3 N 4 hydrolysis can be represented, as follows: (1) (1) www.nature.com/scientificreports/ According to Eq. (1), the nitrogen ions that elute from the Si 3 N 4 surface predominantly become NH 4 + and NH 3 at low and high (> 9.25) pH values, respectively. At physiological pH, the release of nitrogen species leaves positively charged N-vacancies at the surface, which partly counterbalance negatively charged deprotonated silanols (Si-O − ). It also generates free electrons, which in turn induce splitting of the surrounding water molecules and contribute to create transient ROS and, consequently, a variety of RNS including hydroxylamine, NH 2 OH, and, ultimately, nitric oxide, NO 12 .
A consequence of these transient off-stoichiometric reactions is a robust pH buffering, which stabilizes the local pH in the neighborhood of the Si 3 N 4 surface at around 8.5 [12][13][14] . A direct confirmation of ammonium/ ammonia elution has recently been recorded as a function of pH using colorimetric ammonia assays after placing bulk Si 3 N 4 samples into vials containing water 14 . Silicon elution was also found according to inductively coupled plasma atomic emission spectroscopy. Conversely, the elution of Si ions took several orders of magnitude longer time than that of N. Despite the balance alteration of charged surface groups, silanol solubility proceeds at a relatively slow rate at physiological pH and could be instantaneously promoted only under extremely alkaline pH values (i.e., > 10) 20-22 . In Si 3 N 4 , the relative proportion of each surface group depends on both extent of surface hydrolysis and environmental pH. However, secondary silylamine are generally preponderant among basic sites because they are by far more resistant to hydrolysis than the primary silylamine sites 23,24 . In the present context, the N lone pair at the (stable) secondary silylamine sites adsorbed on the surface of Si 3 N 4 is key in the electrostatic interaction between ssRNA viruses and Si 3 N 4 particles, as explained later.
In an attempt to delve into the electrostatic interaction between Si 3 N 4 powder and virions, we attempted to alter the surface bond population of the Si 3 N 4 particles upon pH treatments in acidic and alkaline environments (210 °C; 48 h under pH = 4, 7, and 12). , respectively, of the Supplementary Information. These sets of XPS and ATR-FTIR analytical data revealed that the balance between silanol and amine bond populations on the particle surface could indeed be altered to some extent in both acidic and alkaline environments. The results of narrow scan at the Si 2p edge (in Fig. 2a) show that it is possible to maximize the population of N-Si-N bonds at the Si 3 N 4 surface at the expenses of the N-Si-O ones upon treating the powder in a strongly alkaline environment. This trend is confirmed by the narrow scan at the N 1s and O 1s edges (in Fig. 2b, c, respectively). On the other hand, the acidic environment was more effective in increasing the population of N-Si-O above N-Si-N bonds than the alkaline one 23 . However, the latter treatment also introduced new types of bonds, which included Si-OH and H-N-H 24 , as shown by the ATR-FTIR data in Fig. S-2 It should be noted that, despite the observed variations in surface bond balance, silanol terminals and secondary amines were always the two dominating adsorbed species. This is best seen in the XPS O 1s narrow scans in Fig. 2c). These results are consistent with previous studies 25, 26 that identified pH < 4 as a threshold value for neutral Si-OH species to become the only surface terminating species. A direct consequence of bond balance alterations is the variation in surface IEP (Fig. 3). The variations in IEP values as a function of treatment in aqueous environment at different pH fluctuated in the range + 0.5/− 0.8 around the value 4.5 recorded for the as-received powder (cf. broken curves in Fig. 3). As a consequence, the net surface charge of the Si 3 N 4 particles remained largely negative relative to homeostatic pH, with only a minor fraction of positively charged sites stemming from both SiNH 3 + species and 3 + charged N-vacancies. Figure 3 also shows the IEPs of the viruses investigated in this study, as reported in literature [27][28][29] .
Previous investigations on bulk Si 3 N 4 samples showed the possibility of modulating charging and wetting properties of the surface 30 . In principle, alterations of the surface bond structure of Si 3 N 4 could be obtained through the concurrent application of temperature and pH environments. However, substantial variations toward a lower IEP value could only occur upon exposing the Si 3 N 4 surface to temperatures > 1000 °C in oxygen atmosphere 30 . In this section, we confirmed that both silanols and secondary silylamine surface groups contribute to Si 3 N 4 surface charging in aqueous medium. At homeostatic (or higher) pH, the former species deprotonates and thus charge negative, while the latter species possess an N lone pair capable of additional carbon bonding. While the balance between silanol and silylamine groups can be altered by pH treatments, the two species always coexist and intermix on the Si 3 N 4 surface. As a result, one would expect that the electrostatic interaction between viruses and Si 3 N 4 surface is only partially affected by the development of acidic and alkaline biological environments.

Immunochemistry characterizations
Influenza A H1N1 virions. The effect of the Si 3 N 4 surface on H1N1 infection was analyzed in Madin-Darby Canine Kidney (MDCK) cell culture. MDCK cells are widely used to examine infectivity of Influenza virus, because they are highly susceptible to this virus. Their infection showed replication kinetics similar to that obtained upon infecting human cells 31 . The MDCK cell line is also a suitable choice for efficiently supporting replication and, consequently, for correctly addressing the receptor binding preference of various viruses [32][33][34] .
The virus was preliminary exposed in solution to a 15 wt.% fraction of Si 3 N 4 powder in solution for different times (between 1 and 30 min; referred to as "virus inactivation times", henceforth) prior to infecting the MDCK cells. Si 3 N 4 particles were removed through filtration and centrifugation, and the MDCK cells were inoculated with the virus samples for 48 h before infectivity evaluation. Viral infectivity was compared to that of unexposed virus samples tested under the same conditions as negative controls (cf. the complete description in the Supplementary Information and a schematic draft of the testing procedure in Fig. S-3). The plaque assay was applied as a standard method in determining virus concentration in terms of infectious dose. www.nature.com/scientificreports/ We counted the number of plaques in cell samples and assessed virus quantities as plaque forming units PFU/100 μL as a function of virus inactivation time. Figure 4a, b show the results (with their related statistical validation) of these experiments (n = 3) plotted in logarithmic scale for inoculation experiments conducted at room temperature and 4 °C, respectively. Comparisons are made with control cell samples inoculated with virions unexposed to Si 3 N 4 (labeled as "control"). The respective plots of reduction rate of infectious virus are also shown in the above figures. At room temperature (Fig. 4a), the PFU counts decreased and the reduction rate increased sharply as compared to control with increasing virus inactivation times. Remarkably, the reduction rate already reached ~ 90% (p > 0.001 in Student's test; n = 3) after exposing the virus for only 1 min to the Si 3 N 4 powder. For exposure times equal to or longer than 10 min reduction rates > 99.8% were recorded for all cell samples inoculated with Si 3 N 4 -exposed virions. A reduction of temperature to 4 °C apparently increased the viral load since a slower decrease rate (~ 84%) was recorded for 1 min exposure to Si 3 N 4 . However, with increasing Si 3 N 4 exposure time to 5 min or longer the reduction rate was always > 99% (Fig. 4b). This set of experiments consistently clarified that the Si 3 N 4 powder substantially inactivated the H1N1 Influenza A virus within exposure times as short as few minutes. Figure 5 shows the results of an additional set of experiments conducted at room temperature under the same conditions as above, but using Si 3 N 4 powder with surface charge conditions altered by pH treatments in either   Fig. 5 show a slightly lowered effectiveness for virions treated with Si 3 N 4 powder exposed to a strongly alkaline environment, although a virus reduction rate > 99% could be reached independent of pH. As previously described in Sect. "Investigated ssRNA viruses", exposing the Si 3 N 4 powder to a strongly alkaline environment enabled maximizing the surface population of  FCV and EV-A71 virions. Experiments conducted with the same protocol as above were then applied to quantitatively assess virus inactivation efficiency upon pre-treating FCV and EV-A71 virions with the Si 3 N 4 powder. The former virus was inoculated into cat kidney cortex epithelial cells (CRFK line). This line of feline kidney cells has extensively been used for viral infectivity assays and for studying the biology of various retroviruses and derived vectors 35 . The EV-A71 virus was tested against rhesus monkey kidney cells (LLC-MK2 line), because these cells were found persistently infected with simian virus and developed into a stable carrier state characterized by extensive viral proliferation without obvious cytopathic effects 36 . Figure 6a, b show the results (with their related statistical validation) of median tissue culture infectious dose (TCID 50 ) and reduction rate of infectious virus as a function of virus inactivation time for the cases of FCV and EV-A71 inoculation in CRFK and LLC-MK2 cells, respectively. Data (n = 3) on CRFK and LLC-MK2 cells were collected at room temperature with inoculating virions previously treated in aqueous solutions containing 30 and 15 wt.% Si 3 N 4 , respectively. A comparison of the inactivation efficiency of Si 3 N 4 toward different viruses (cf. Figs. 4a and 6) revealed that the FCV was the most resistant to inactivation: even with a 30% fraction of Si 3 N 4 powder during pre-treatment of virions, double the time (10 vs. 5 min) was needed to achieve a nearly full inactivation as compared to H1N1 Influenza A virions. On the other hand, the inactivation trend for EV-A71 was very similar to that observed for the H1N1 Influenza A strain (15% Si 3 N 4 powder in the inactivating solution in order to reach virus reduction rate > 99% with 5 min virion exposure to Si 3 N 4 particles).
The above immunochemistry characterizations suggest that a chemical interaction occurs between ssRNA viruses and the Si 3 N 4 surface in aqueous environment. Such a chemical interaction leads to a high degree of virus inactivation. Although the details of inactivation mechanisms might slight differ among the viral species studied, a substantial inactivation effect could commonly be achieved within minutes upon adjusting the fraction of Si 3 N 4 particles in the inactivating medium. To this extent, we shall assume that the observed viral inactivation is commonly related to the phenomenon of Si 3 N 4 hydrolysis in aqueous environment as represented by Eq. (1). The viral inactivation effect persisted after thermally treating the Si 3 N 4 powder in different pH environments, although a relatively small decrease in antiviral efficiency could be found upon exposing the Si 3 N 4 powder to a strongly alkaline environment (cf. Fig. 5). The key to interpret such differences likely relates to the electrical charge of surface groups and the IEP of the Si 3 N 4 surface. This point will be discussed in a later section.

Fluorescence imaging and RT-PCR analyses of gene fragmentation
In order to substantiate the virus inactivation effect as recorded by immunochemistry characterizations, we performed fluorescence assays on inoculated MDCK cells and TaqMan-based real-time reverse transcription (RT)-polymerase chain reaction (PCR) analyses 37 on the Influenza A H1N1 viral strain before and after inactivation upon contact with Si 3 N 4 powder. Figure 7 shows fluorescence micrographs of the MDCK cell population associated with viral RNA. Fluorescent staining was performed 9 h after infection. Cells inoculated with unexposed virus (control sample), but subjected to an otherwise identical procedure, and cells inoculated with  Fig. 7c. In order to obtain information about the viral RNA structure after surface interactions with Si 3 N 4 powder, we performed RT-PCR assay on the viral RNA of Influenza A H1N1 supernatant virions exposed and non-exposed to Si 3 N 4 powder at room temperature. Two viral concentrations (i.e., 1.0 × 10 7 and 5.0 × 10 5 PFU/ml) were tested for a fixed 15 wt.% Si 3 N 4 powder concentration in aqueous solution. Independent of viral concentration, the RT-PCR tests revealed the occurrence of significant RNA fragmentation for the three selected probes (cf. selected primers and probes for TaqMan amplification of cDNA with Influenza A H1N1 virus sequence and quantitative results of RT-PCR tests in Tables 1 and 2, respectively). We interpreted these results as an experimental evidence of RNA fragmentation by hydrolysis-driven reactions at the virion/Si 3 N 4 interface.

Raman spectroscopy of Si 3 N 4 -exposed and unexposed virions
In situ Raman experiments were performed on Influenza A H1N1 virions before and after 10 min exposure to 15 wt.% Si 3 N 4 powder in aqueous solution. Figure 8a shows a Raman spectrum in the frequency range 600 ~ 800 cm −1 , which was collected in situ on the unexposed population of Influenza A-H1N1 virions (control sample). According to general classifications 38 , the examined Raman spectral region in this range mainly contained vibrational features from C-S bonds and from ring vibrations in specific RNA nucleotides. In this spectral zone, we detected 12 main Raman bands in the control sample of H1N1 virions (cf. Fig. 8a and Table S-I for vibrational frequencies, physical origins of the 12 labeled bands, and related references). In Fig. 8b, the Raman www.nature.com/scientificreports/ spectrum of the unexposed virion population is shown in the frequency interval 800 ~ 1010 cm −1 . In this zone, additional 11 bands were detected and labeled with respect to their frequency, physical origin, and related references as given in Table S-II. The Raman signals in this zone mainly included S-H bending modes in addition to ring vibrations from nucleotides and O-P-O stretching vibrations in RNA backbone. In the present context, a relevant vibrational signal was represented by Band 18, which is mainly related to S-H in-plane bending of homocysteine 39 . The same vibrational mode also contributed Band 22, although in strong overlap with Amide III signals from proteins 40 . Substantial morphological differences could be found by comparing spectra of the same viral strain before and after entering in contact with the Si 3 N 4 powder. A more extensive description and discussion of the complete spectrum of H1N1 virions and MDCK cells have been given in a previously published paper including a Raman follow-up of the viral inoculation process 41 . In this paper, we only discuss two fundamental spectral features related to structural alterations of methionine and RNA in the H1N1 virions after entering in contact the with the Si 3 N 4 particles. Bands 3, 4, 5, 7, and 8 (at 640, 650, 669, 698, and 716 cm −1 , respectively) in the Raman spectrum of untreated virions (Fig. 8a) are related to C-S bond stretching in the methionine residue 39,40,[42][43][44][45][46] , which is the first standard amino acid residue at the N-terminal position of hemagglutinin (HA). More specifically, Bands 3 and 4 represent C-S bond vibrations on the CH 2 side for the molecule in the gauche configuration, while Band 5 arises from the same C-S bond on the CH 2 side for the molecule in the trans configuration. Conversely, Bands  www.nature.com/scientificreports/  www.nature.com/scientificreports/ 7 and 8 relate to C-S stretching on the CH 3 carboxyl side in gauche and trans configurations, respectively 42 . Upon comparing the Raman spectra of H1N1 virions before and after 10 min. treatment with 15 wt.% Si 3 N 4 in water solution (in Fig. 8a, c, respectively), the most striking differences were related to C-S bond vibrations and consisted in a significant intensity reduction of Bands 7 and 8, while a substantial enhancement could be found in the relative intensity of Band 4 and a less pronounced one for Band 5. Concurrently, two new bands appeared, which were shifted toward higher frequencies with respect to Bands 4 and 8 and were labeled as Bands 4* and 8*, respectively (cf. Fig. 8c). The origin of these two new bands can be related to modifications of the methionine structure. More precisely, Bands 4* and 8* could be attributed to C-C-S stretching on the CH 2 side and C-S-C stretching on the CH 3 side of methionine in thioether configuration 39 . This point will be discussed in more details later. However, based on these observations, we hypothesize that the several concomitant spectral modifications detected for C-S bond-related vibrations suggest that the methionine residue of HA promptly interacted with the Si 3 N 4 surface. In turn, such interaction induced fundamental modifications of its thioether group and terminal structure. Comparing higher frequency spectra in Fig. 8b, d, additional differences could be located between the unexposed (control) viruses and the Si 3 N 4 -treated ones, as follows: (i) An increased signal for Bands 18, 19, and 22 (located at 913, 929, and 985 cm −1 , respectively) for Si 3 N 4 -exposed virions (Fig. 8d). All these bands are contributed by signals from homocysteine: Bands 18 and 22 correspond to the in-plane S-H bending mode, while Band 19 arises from C-C stretching 39 . Note, however, that no "pure" signal for homocysteine S-H vibration is available, since Band 18 overlaps a ring deformation mode of RNA adenine 46  The meaning of the spectral modifications recorded after exposing the virions to Si 3 N 4 powder is discussed in the remainder of this section based on basic Raman experiments on individual methionine polymorphs in aqueous solution after interactions with the Si 3 N 4 powder.
In order to confirm the above hypothesis of interaction between charged surface groups on the Si 3 N 4 surface and the HA methionine terminus of Influenza A H1N1 virions, we performed basic Raman experiments on L-and DL-methionine in aqueous solution containing 15 wt.% Si 3 N 4 powder. The Raman spectra of the two above methionine polymorphs before and after interaction with the Si 3 N 4 particles are given in Fig. 9a for the spectral zone 600 ~ 800 cm −1 (cf. labels). Drafts of the HA terminus of the methionine molecule in its gauche and  www.nature.com/scientificreports/ trans configurations is depicted in Fig. 9b. The labels in the spectra of Fig. 9a, which refer to the deconvoluted virion spectra in Fig. 8, help to visualize important similarities between the spectra of individual methionine polymorphs and that of Influenza A virions, as follows: (i) A substantial rearrangement of the low-frequency spectral region of the L-methionine polymorph containing Bands 3 ~ 5 (cf. upper spectrum in Fig. 9a), with a significant enhancement of Band 4 and the appearance of a new Band 4* in the metastable DL-methionine polymorph after treatment with Si 3 N 4 in aqueous solution (cf. lower spectrum in Fig. 9a). (ii) A substantial rearrangement of the spectral region containing Bands 7 and 8, which belonged to DL-and L-methionine, respectively (cf. both Fig. 9a, b), with the appearance of a new Band 8* in the metastable DL-methionine polymorph at a frequency corresponding to that observed the Raman spectrum of virions in Fig. 8c.
A likely interpretation of the above spectral features attributes the significant reduction in intensity of Bands 7 and 8 upon methionine interaction with Si 3 N 4 powder to the formation of thioether groups in the methionine residue, which is thus deprived of the C-S bond on the CH 3 side to turn into homocysteine (as schematically depicted in Fig. 9b). Such chemical circumstance in turn enhances the vibrational activity of C-S bonds on the CH 2 side, which can explain the observed enhancement in the intensity of Band 4. This interpretation is supported by the observation of the new Band 20* at around 959 cm −1 , which corresponds to S-H in-plane bending vibration of homocysteine, and by the increased signal for Bands 18, 19, and 22 also contributed by vibrational modes of homocysteine in the Si 3 N 4 -exposed virions (cf. Fig. 8d and Table S-II). Note also that Band 18 is possibly contributed by the stretching mode of C-COO − deprotonated carboxyl group, which is in line with a pH buffering effect induced by Si 3 N 4 in the aqueous solution 14 .
The basic experiments on pure methionine polymorphs in Si 3 N 4 -added aqueous solution suggest that the most probable scenario behind the interaction between Si 3 N 4 surface and the envelope of Influenza A virions involves deprotonated silanol groups at the surface of Si 3 N 4 exerting a strong electrostatic attraction toward the C-COOH terminus of the methionine residue (Fig. 10a). On the other side of the methionine molecule, thioether www.nature.com/scientificreports/ cleavage can be triggered by protonated amino groups, which act as hydrogen bond donors and form strong hydrogen bonds with hydrogen bond acceptor silylamine sites on the Si 3 N 4 surface (Fig. 10b) 48 . The potential attitude of secondary silylamines to bond to carbon likely involves a direct link with the methionine methyl group CH 3 to form a quaternary amine with positive charge (Fig. 10b). We thus hypothesize that, once the viral capsid is mechanically forced to detach from the Si 3 N 4 particle by centrifugation or filtration, the thioether group of methionine residue might cleave while its C-COOH group deprotonates, thus transforming methionine residues into a zwitterionic homocysteine form (Fig. 10c), as detected by Raman spectroscopy. In addition to S-related bond vibrations, the low-frequency region of the Raman spectrum of virions contains fingerprints of the secondary structure of RNA 49,50 . In this study, bands in this zone exhibited significant changes assignable to conformational perturbations of the ribosylphosphate backbone 51 . For example, the observed reduction in intensity of Band 13 (at ~ 818 cm −1 ), which is related to the O-P-O stretching in RNA backbone 52 (cf. Fig. 8b, d), points to a trend to phosphodiester cleavage 51 . One possible scenario, which is also supported by RT-PCR analyses showing substantial RNA fragmentation after interaction with Si 3 N 4 (cf. Table 2), is alkaline transesterification of the RNA backbone through hydrolysis of its phosphodiester bond. In case of alkaline transesterification, the process of genome degradation 53 starts with the deprotonation of the 2′-hydroxyl group by ammonia base (Fig. 10d), it is followed by the formation of a transient pentaphosphate unit (Fig. 10e) and, upon further interaction with acidic ammonium, ends with RNA fragmentation due to self-cleavage (Fig. 10f). Additional damages to the virion RNA structure could be hypothesized from the complete disappearance of Bands 15, 17, and 21, which belong to ring structures in RNA guanine nucleotide (cf. Fig. 8d), and from the missing Band 2 related to ring vibrations of RNA uracil nucleotide (cf. Fig. 8c).
Finally, note that, unlike the mechanism of methionine degradation (peculiar to Influenza A viruses), the processes of ammonia-induced RNA fragmentation could be common across a number of different ssRNA viruses, independent of their enveloped or non-enveloped structures. Despite significant differences in protein structure among different virions, highly volatile NH 3 and NO radicals can penetrate the virions and damage their RNA, independent of the capsid structure of the virion 53 . In a previous in situ study on the effect of Si 3 N 4 hydrolysis on living gram-positive bacteria 13 , Raman spectroscopy similarly revealed damages to the DNA structure by disappearance of Raman band assigned to vibrations in deoxyadenosine triphosphate, ring breathing of guanine, and phosphodiester stretching. This effect was attributed to oxidation of ammonia into hydroxylamine along with the concurrent presence of superoxides and highly volatile radical species such as nitric oxide and peroxynitrite. Additional experiments on different Influenza viral strains of the A and B type are ongoing in order to validate the present interpretation of Raman data.

Discussion and outlook
Molecular targets of solid-state viral inactivation. The trimeric surface protein hemagglutinin (HA) plays a critical role in two consecutive steps of viral replication: (i) binding to sialic acid on the cell membrane (as a preliminary step toward endocytosis); and, (ii) inducing membrane fusion through a conformational change under the lowered pH environment of the virion-containing acidified endosome 54 . Some 16 distinct subtypes of HA exist 55 , and the identification of a multivalent antiviral remedy effective against multiple HA subtypes yet represents a challenge. In the context of the above item (i), a compound that mimics sialic acid in HA binding could prove effective in attracting the virus independent of HA subtype. Successful examples of the socalled "competitive binding" approach have been reported in the literature in anti-influenza drug design 56,57 , but reports on solid-state viral inactivators based on this approach are yet hardly found in the published literature. Concerning the above item (ii), alternative approaches to HA inactivation have targeted the fusion function of membrane proteins. For example, tert-butylhydroquinone is a compound that binds to three identical sites on the pre-fusion HA trimer with the effect of stabilizing the proteins, preventing their conformational changes in acidic pH environment, and inhibiting viral/endosome membrane fusion 56 . Targeting drug-binding sites with proteins is a well-known research strategy in virus inactivation 57 . Again, however, viral inactivation mechanisms based on the interactions between the HA trimer and a solid-state compound has so far been an unusual research task.
The RNA polymerase of the Influenza virus consists of a trimer of viral proteins PB1, PB2, and PA it is divided into several domains with well differentiated functions, including polymerase, RNA binding, cap-binding, endonuclease, etc. Its primary function is to synthesize (+) sense viral mRNAs and to replicate the (-) sense viral genome. One additional function consists in capping the synthesized mRNAs with a "snatched" 5′ methyl guanosine cap cleaved and then "stolen" from cellular mRNAs 54 . New compounds, such as marchantin E 58 , have proved capable of inhibiting viral RNA polymerase endonuclease activity, thus preventing proteins from cap snatching and viral RNA from being translated. Direct approaches targeting viral RNA have also been pursued 59 . In this latter context, inactivation of Influenza A viral particles has been achieved upon reaction with artificial ribonucleases 60 . These chemical compounds are capable to directly cleave the viral RNA molecules while causing minimal damage to the structure of surface epitopes. The exploitation of RNA cleavage mechanisms holds general validity against infections caused by a wide variety of RNA viruses.
Si 3 N 4 was effective as a solid-state virus inactivator, merging the effects of competitive HA binding, thioether cleavage of methionine residues, RNA phosphodiester bond cleavage and nucleotide damage. The presented data suggested that Si 3 N 4 could provide charged sites on contact, which degrade the structure of HA, and elute NH 4 + / NH 3 molecules that fragment the RNA of virions. However, differences in inactivation kinetics among different ssRNA viruses were detected. Such differences may arise from: (i) different levels of electrical attraction of virions toward the Si 3 N 4 particles; and, (ii) differences in the genomic structure of different viruses.
Regarding the above item (i), the reported IEP of FCV virus is very close to that of Si 3 N 4 (cf. Fig. 3). Accordingly, FCV virions possess an electrical charge very close to that of the Si 3 N 4 surface over a wide range of pH and, www.nature.com/scientificreports/ thus, they experience a lower probability of encountering a Si 3 N 4 particle in suspension due to weak electrical attraction. Conversely, Influenza A H1N1 and EV-71 virions possess IEPs significantly greater than that of Si 3 N 4 . Translated in terms of electrical attraction, these latter virions are subjected to attractive forces and experience a higher probability of being "caught" by the Si 3 N 4 particles. Once in contact with the Si 3 N 4 surface, the virions are "poisoned" by eluted ammonia molecules. Such a two-steps virion/surface interaction has been found very effective also in the case of SARS-CoV-2 virions, which could be inactivated up to > 99% within exposure times as short as 1 min. This composite antiviral mechanism was branded as the "catch-and-kill" effect 15 . Regarding possible differences in genomic structures (i.e., the above item (ii)), the heterogeneity observed here among ssRNA viruses toward Si 3 N 4 inactivation is likely also contributed by differences in sequence and length of the genome 61 . The greater lengths of the FCV genomes (7683 nucleotides 62 ) and EV-A71 genomes (7407 nucleotides 63 ) compared to that of each genome segment of Influenza A H1N1 (the longest RNA segment consisting of 2341 nucleotides 64 ) enhances cleavage probability and may render the former two virions more susceptible to cleavage than the latter one. This factor might partly compensate for a weaker electrical attraction of the FCV toward the Si 3 N 4 surface. In summary, the EV-A71 virus is most efficiently inactivated because of two concurrent circumstances: a high electrical attraction to the Si 3 N 4 surface (due to its high IEP) and a high probability of RNA transesterification due to its greater genome length.
A schematic draft of the "catch-and-kill" antiviral effect developed at the interface between Influenza A HIN1 virions and the Si 3 N 4 particles in aqueous suspension is depicted in Fig. 11. According to this schematic draft, negatively charged silanol groups attract the virions while secondary silylamines "snatch" the C-S bond of the methionine residue at HA terminus. Concurrently, highly volatile NH 3 molecules (directly eluted by the Si 3 N 4 surface) swiftly penetrate the envelop of the virions and damage their RNA backbone and nucleotides. This effect has been consistently demonstrated by both in situ Raman spectroscopic assessments and RT-PCR analyses of viral genome.
Significance of the present findings and future developments. Our previous in vitro studies of bacteria 13 and mesenchymal cells 65 , as well as retrieval analyses of explanted Si 3 N 4 spinal inserts 66 have revealed a dual attitude for Si 3 N 4 ; namely, this bioceramic being capable of lysing bacteria while supporting cell proliferation and functionality. This dualism is the result of a cascade of off-stoichiometric reactions that take place at the surface of Si 3 N 4 , and lead to the formation of silanol complexes and elution of NH 4 + /NH 3 species 12,14 . At physiological pH, the main species eluted is NH 4 + , which enters the cytoplasmic space of cells in controlled concentrations through specific transporters 67 . Ammonium ions represent a nutrient, which is used by the cells to synthesize building-block proteins for both enzymes and genetic compounds as needed to sustain cell differentiation and proliferation 68 . Conversely, highly volatile ammonia molecules freely penetrate the external membrane and directly target the stability of DNA/RNA structures in both mammalian and bacterial cells 69 . However, at the low concentration and slow kinetics of NH 3 elution from Si 3 N 4 at physiological pH, eukaryotic cells receive no damage because their mitochondria can easily neutralize small NH 3 infiltrations through cytochrome c oxidase enzymatic catalysis. This is a reaction chain that produces free electrons and reactive intermediates, which ultimately oxidize NH 3 into hydroxylamine NH 2 OH (ammonia monooxygenase) and boost NO formation 70 . Conversely, bacteria, which lack mitochondria, cannot catalyze NH 3 and face chemical destabilization of their DNA/RNA structure 13,14 .
We probed here whether a surface chemistry mechanism similar to DNA/RNA destabilization could be operative versus ssRNA viruses and discovered the two following (combined) antiviral actions for Si 3 N 4 : (i) competitive Figure 11. Schematic draft of the dual effect of "catch-and-kill" viral inactivation at the interface between Influenza A virions and Si 3 N 4 : negatively charged silanol groups attract the virions at the HA methionine terminus, while highly volatile NH 3 (eluted by the Si 3 N 4 surface) and NO (formed by reactions with ROS from split water) molecules penetrate the virions and damage their RNA sequence by alkaline transesterification and uracil/guanine ring disruption. www.nature.com/scientificreports/ HA binding; (ii) RNA phosphodiester bond cleavage; and, (iii) RNA nucleotide ring damage (specifically in uracil and guanine). The former mechanism is related to electrostatic attraction exerted by deprotonated surface silanols at the Si 3 N 4 surface, while both the latter two ones arise from molecular infiltrations of eluted NH 3 . The present study is in line with our previously published data [12][13][14][15] , and it specifically confirms our recently published data on the inactivation of SARS-CoV-2 by nitride ceramics 15 . The efficacy of Si 3 N 4 as a solid-state virus inactivator relies on RNS rather than ROS species [12][13][14] . Antiviral activities of insoluble solid-state materials have so far been reported for copper, aluminium, and silver compounds 71 . Solid-state cuprous oxide (Cu 2 O) is known to inactivate both enveloped and non-enveloped viruses. Cu 2 O markedly reduces the HA activity by disrupting host cell recognition through denaturing protein structures on viral surfaces, independent of the presence of a viral envelope. However, although solid Cu 2 O shows superior antiviral activity, this compound is quite toxic to cells. A high level of toxicity for copper particles toward VeroE6/TMPRSS2 was also recorded in our recent study on SARS-CoV-2 15 . In aqueous solution, soluble oxygen oxidizes Cu + ions to Cu 2+ through reactions that produce ROS 72 . ROS then readily react with lipids, proteins, and nucleic acids, and result in significant damage to cell structures by quickly cumulating into a situation of oxidative stress 73 . Cu 2 O crystals were found to promote endothelial cell death via autophagy, to elevate the level of ROS such as superoxides, which subsequently activate AMP-activated protein kinase 74 . Unlike the enhancement of ROS caused by metal contact, cells easily metabolize the RNS developed at the interface with Si 3 N 4 for a similar if not even superior antiviral effect. It should be noted that electron microscopy observations of the Si 3 N 4 particles used in this study (cf. Fig. 1a) revealed that they are composed of many smaller crystallites imparting to the particles a quite rough surface. The dimension of viruses is much smaller than the Si 3 N 4 particles and comparable with the nanostructures observed on the surface of the particles. This circumstance could provide an additional scavenging effect on the viruses, which likely adds upon the chemical effect and could also be common to other types of ceramic microparticles showing a similar surface structure.
A recent study 75 revealed high antibacterial efficacy also for Si 3 N 4 particles embedded in a polymeric matrix despite only a minor fraction of particles was added. Two recent studies on 3D-additive Si 3 N 4 coating by laser cladding similarly revealed the possibility to translate antipathogenic properties to metallic 76 and polymeric 77 surfaces. The results obtained on polymer/ceramic composites and coatings suggest that the antibacterial/antiviral properties of Si 3 N 4 could directly be translated into virus-resistant surfaces for use in hospitals, schools, and other environments where contact-based viral disease transmission is a concern.

Conclusion
We described a novel Si 3 N 4 ceramic technology that was effective in inactivating ssRNA viruses independent of virions being of enveloped or non-enveloped types, or their RNA possessing positive or negative polarity. The favorable surface chemistry of Si 3 N 4 can control viral transmission without the toxicity associated with other anti-viral strategies. Virus inactivation by Si 3 N 4 ceramics represents a new, broad-spectrum approach that is safe toward mammalian cells. Its unique combination of surface properties could help control the surface-mediated rapid transmission of viral epidemics. The molecular mechanisms described in our report may be generally effective against other known viruses, and be useful in combatting disease outbreaks. www.nature.com/scientificreports/