The syntheses of the nanopolymers were performed by thermally co-polymerizing L-lysine and L-arginine in various ratios using boric acid to catalyze the amidation reaction.
3.1. Hyperbranched Nanopolymer Characterizations
TGA-DSC was used to evaluate the compounds’ thermal stability under the reaction conditions used for the syntheses. The analyses additionally provided important information on how temperature affects amino acid reactions. A comparison of the thermal analyses of L-lysine and L-arginine showed that the latter was thermally more stable. On the other hand, L-lysine was among the amino acids most susceptible to thermally induced degradation (
Figure 2).
At 400 °C, L-lysine and L-arginine lost 80% and 55% of their weight, respectively. L-arginine showed two endothermal effects, the first at 219 °C without any mass loss and the second at 236 °C with a 30% mass loss. The first peak of heat flow in
Figure 2a can be attributed to a molecular rearrangement of the guanidinium group. The second, formally involving the loss of 1 mol of NH
3 and 1 mol of water, can be assigned to a double intramolecular cyclization. The first cycle was obtained through the internal nitrogen of the guanidinium group attacking the α-carbon and expelling NH
3, leading to a proline derivative. At this point, the primary amino group of the guanidinium residue causes a nucleophilic reaction that attacks the carboxylic group. The reaction produces a cyclic amide, a lactam, with a loss of 1 mol of water, resulting in a two-ring structure obtained through the combination of creatinine and proline moieties [
18].
The TGA of boric acid exhibited two major endothermic signals that peaked at 130 and 160 °C, respectively, due to dehydration and the production of metaboric acid (HBO2). The first-derivative curves of the TGA revealed that the endothermic events corresponded to the weight loss of the samples. The classic peptide reaction occurred through the condensation of carboxylic groups and amines. The DSC curves showed that the endothermic amidation reaction occurred at the same temperature in the L-lysine and L-arginine at 235 °C.
When boric acid and L-arginine were combined, the reactions occurred at slightly higher temperatures, with the amidation reaction peaking at 242 °C with a small shoulder at 223 °C. Noticeably, with the addition of boric acid, the first endothermic event, peaking at 223 °C, was connected with a weight loss, as indicated by a shoulder in the first-derivative curve, which did not occur in pure L-arginine. This loss can be explained by the interaction between boric acid and L-arginine, which significantly impacted the thermal reaction pathway. While bare L-arginine lost water and ammonia to give an intramolecular cyclic amidation, in combination with boric acid, it formed a cyclic catalytic intermediate instead [
19]. The second endothermic event can thus be related to the formation of the amidation product, in which the L-arginine moiety, activated by the boric acid, reacted with a second residue of L-arginine or L-lysine, affording nanopolymers, whose nature depended upon the Lys:Arg ratio. The same behavior can also be envisaged for the amidation of L-lysine catalyzed by boric acid (see
Figure S1). While the DSC profile of the pure L-lysine showed a single peak at 236 °C, that of the equimolar mixture of L-lysine and boric acid revealed two close endothermic events, peaking at 196 and 214 °C, that can be related to the formation of the boric-acid–lysine intermediate and the resulting amidation product, respectively. This behavior was observed in all the samples prepared with different Lys:Arg ratios catalyzed by boric acid (see
Figure S1). In particular, the samples with ratios equal to 1:0.1 showed a profile similar to that of BA:Lys = 1:1. This is not surprising, since the amount of L-arginine was only 10%. Starting from the 1:0.25 ratio, the first endothermic event was observed in the curve as a weak shoulder. At the same time, the L-arginine amount increased, giving rise to a single band convoluting the two events: the formation of the intermediate and subsequent amidation. The shape of the DSC curve can be explained by considering that, in all the samples, the amount of boric acid, which is a limiting factor for forming the cyclic intermediates, was kept constant. In contrast, the amount of amino-acid moieties progressively rose to a 1:3 ratio in the sample with Lys:Arg = 1:2. This tendency appeared more evident when examining the first-derivative profile of the weight loss (see
Figure S1), where the two peaks related to the weight loss gradually transformed into a single band with a shoulder.
Figure 3a shows the FTIR absorption spectra of L-lysine and L-arginine in the 1800–1500 cm
−1 range (the full spectra are reported in
Figure S2). The spectra were collected using the as-purchased amino acid powders. The absorption spectrum of L-lysine was characterized by a main band peaking at 1581 cm
−1 assigned to -COO
- antisymmetric stretching, a shoulder at 1615 cm
−1 due to the antisymmetric stretching of -NH
3+, and the rocking band of the same group at 1515 cm
−1. The L-arginine spectrum showed four absorption bands at 1722 cm
−1 (ν
s C=O), 1728 cm
−1 (bending NH
2), 1623 cm
−1 (ν
s C=N in the guanidinium group), and 1555 cm
−1 (bending N-H). The observation of these bands upon reaction can give some clues about the changes in the structure (
Figure 3b). The reaction catalyzed by boric acid induced polymerization, as shown by the rise in the amide I band around 1650 cm
−1. Previous works have indicated that a reaction between L-lysine and boric acid forms a hyperbranched polymer [
10,
12], HBPL (
Figure 3b). The addition of a small amount of L-arginine, sample LBA0.1, did not change the FTIR spectrum for the reasons we have explained before (vide supra). At higher L-arginine concentrations, the spectra showed a distinct amino acid signature that overlapped the HBPL spectrum, indicating that not all the amino acid functional groups reacted. The full FTIR absorption spectra of the Lys-Arg samples are reported in
Figure S3.
Figure 4 shows the NMR spectra in the 0.4–4.6 ppm range of the samples prepared with increasing amounts of L-arginine. The reference spectra of pure L-lysine and L-arginine are reported in
Figure S4. Significant NMR signal broadening suggests the formation of slowly reorienting macromolecular structures. The intensity of the signals at 1.4 and 3.0 ppm, assigned to the γ- and ε-CH
2 protons of L-lysine, respectively, decreased with increasing the amount of L-arginine in the mixture. The signal from ε-CH
2 protons, also observed in commercial poly-L-lysine (
Figure S4), showed a downfield shift as a function of increasing amounts of L-arginine, suggesting a progressive amide bond formation until the Lys:Arg ratio reached the 1:1 value. The sample with the lowest content of L-arginine (LBA0.1) showed a signal at 3.75 ppm that progressively shifted to 3.91 ppm in LBA0.25 and to 3.96 ppm in LBA0.5 and LBA1 (See
Figure S5). This trend suggests that the signal was due to the α-CH
2 groups next to the amide bonds of L-lysine. In particular, the rise in the signals at 4.15 and 4.25 ppm indicates the formation of a hyperbranched polymeric structure via amide bonding, which was in agreement with FTIR analysis and previously published data [
20,
21].
The signal at 4.38 ppm corresponded to α-CH protons in the dendritic structure formed by the reaction of the carboxyl group with the secondary amino group of the guanidinium star moiety in L-arginine. This signal also rose in intensity with the increase in the amount of arginine, confirming this possible attribution. The NMR data suggest that increasing the L-arginine content in the mixture caused the formation of a more branched and less linear structure.
With the increase in the amount of L-arginine co-reacted with L-lysine, the NMR spectra in the same range were dominated by signals that were due to polymerized L-arginine, as can be observed by comparing the spectra with that of pure L-arginine. Furthermore, the NMR data suggest that the reaction of L-lysine with L-arginine at low concentrations formed an interconnected structure. L-arginine became, in this case, a modifier of the main L-lysine hyperbranched polymeric structure. On the other hand, at high concentrations, L-arginine formed its own interconnected network, and the polymer was intrinsically more disordered with fewer linear structures.
Figure 5 shows a two-dimensional NMR spectrum that displays the homonuclear (
1H-
1H COSY) correlations for the sample with the lowest concentration of L-arginine (LBA0.1), confirming the previous assumptions. In particular, the signals of the α-CH proton and their β-CH
2 siblings (3.81 and 1.88 ppm, respectively), usually correlated with peptide formation, were shifted downfield (4.23 and 1.94 ppm), indicating the formation of hyperbranched polymers (HBP), as previously reported for hyperbranched poly-L-lysine (HBPL) [
10]. The formation of a hyperbranched structure is not surprising since the Lys:Arg molar ratio was only 1:0.1, and, thus, adding a small amount of L-arginine did not significantly alter the hyperbranched nature of the nanopolymer. This hyperbranched character was also confirmed by another two-dimensional NMR spectrum reporting a proton–carbon heteronuclear correlation (
1H-
13C HSQC) (see
Figure S6). Here, the signals of the α-C carbon and the correlated α-CH proton of the amino acids involved in the peptide chains were downshifted at 56.44 (α-C carbon) and 4.22 ppm (α-CH proton), respectively, for the L-lysine moiety and to 56.23 and 4.28 ppm for L-arginine.
The thermal polymerization of L-lysine and L-arginine led to the formation of nanopolymers with branched peptidic chains. The protein-like nature was governed by the pH-dependent protonation states of the L-lysine- and L-arginine-derived samples [
20]. As shown in
Figure 6, the two amino acids could assume four protonation–deprotonation states with different charges and pKa values.
The thermally induced amidation gave rise to polypeptide nanopolymers bearing amino acid side-chains that control the overall surface charge of the nanoparticle and thus have an important effect in its interaction with viruses. In particular,
Figure 6 suggests that the amino acid side-chains, at physiological pHs, should be mostly in the +1 states, Lys
+ and Arg
+, respectively.
The sizes of the different nanoparticles were assessed by TEM (
Figure 7). The analysis of the samples at various lysine/arginine ratios showed that the nanopolymers were made of a wide range of structures featuring different electronic densities, which were related to different polymerization degrees. The smaller structures had an average size in the range of 20–50 nm and appeared to be more polymerized with respect to the larger particles, which showed a size ranging from 80 to 200 nm. AFM (see
Supplementary Information, Figure S7) confirmed the presence of harder nanoparticles with a size in the range of 40–60 nm, which appeared embedded into a softer structure, compatible with that of not completely polymerized particles.
DLS was therefore used to evaluate the stability in solution of the nanopolymers (
Figure 8). Regardless of the complex morphology revealed by TEM, the hydrodynamic radius, as measured by DLS, was in the range of hundreds of nanometres, indicating that the smaller nanoparticles tended to form aggregates when dissolved in water. In more detail, the hydrodynamic radius of the nanopolymers was in the range between 150 and 300, similar to the average size of SARS-CoV-2 virions, i.e., 60–140 nm [
22,
23], except for sample LBA2, which was out of the range of interest.
The absolute ζ potential value indicates the relative stability of nanoparticles. A positive ζ potential generally indicates that the surface of a particle has a positive charge and is a measure of the electrostatic repulsion between particles or surfaces in a liquid medium. BAA and LBA2 showed the lowest ζ potential (
Table S2), while the other particles had a positive ζ potential close to +20 mV. The particles with a smaller ζ potential tended to aggregate or precipitate. This was observed in the particles with the highest content of L-arginine, which makes them not suitable for biological applications.
It is important to stress that the ζ potential is not the only factor that determines the surface charge of a particle or surface. The measurement can also be affected by different variables, such as the pH, the ionic strength, and the presence of charged molecules or ions in the surrounding fluid. Therefore, a direct experimental evaluation of the surface charge and surface state is not simple. In the present case, the values obtained when measuring the ζ potential were not taken in a cell culture medium but in an aqueous solution, which does not represent the biological medium where a viral infection occurs. This choice was dictated by the need for avoiding interference due to the heterogeneity of the medium that could have affected the measurements. Also, in the case of SARS-CoV-2 strains, the experimental assessment of the surface properties is complicated. Therefore, only theoretical calculations based on the amino acid residues in the spike are generally reported in the literature. Theoretical calculations, in fact, do not take into account the role of water and ions in bridging the interaction between the virus spike and the cell receptors.
3.3. Antiviral Activity
An antiviral assay was performed in parallel with the cytotoxicity experiments, in a prevention and post infection model, using the SARS-CoV-2-permissive cell line Vero E6 from the same culture. In the prevention model, the cells were first seeded into 96-well plates and exposed for 1 h to different nanopolymer concentrations. Afterwards, the cells were infected with the SARS-CoV-2 original, Delta, and Omicron strains (with a multiplicity of infection, m.o.i., of 0.01) and finally cultured for 72 h.
In the post-infection model, on the contrary, the cells were first cultured with the SARS-CoV-2 strains for 72 h and then exposed to increasing nanopolymer concentrations for 1 h. In both models, the viral replication was assessed by ELISA to quantify the SARS-CoV-2 nucleoprotein.
The antiviral efficacy data (i.e., SARS-CoV-2 nucleocapsid protein expressed as % of control, mean, and standard deviation) obtained in the ELISA assay for all the nanopolymers are shown in
Figure 10 for the different viral strains, i.e., original, Delta, and Omicron (a, b, and c panels refer to the prevention model, while c, d, and e refer to the post-infection model).
The nanopolymers were highly effective against the original strain, and at the greatest concentration, they could entirely block viral propagation. The data obtained from the prevention model show that all the L-arginine-modified nanopolymers had a stronger antiviral activity compared to HBPL at concentrations equal to or higher than 20 μg mL−1, except for LBA0.5, which still overcame the HBPL performance at a concentration of 100 μg mL−1. This result indicates that the co-polymerization with arginine moieties is an effective method to enhance the antiviral effect. In the case of the Delta virus, the antiviral activity was low and became visible only at 100 μg mL−1, with HBPL being the most effective. The lack of a linear trend in the antiviral efficacy as a function of the L-lysine/L-arginine ratio can be explained by considering that the ELISA results depended on both the antiviral properties of the nanoparticles and their cytotoxicity. As an extreme case, by using an ELISA essay, a theoretical nanoparticle concentration achieving a 0% cell viability produced the same effect as a nanoparticle concentration that is completely effective against virus replication. Indeed, both were capable of minimizing the amount of SARS-CoV-2 nucleoprotein, reducing the probability of viral replication to zero. In intermediate cases, as with the LBA nanopolymers, both the reduction in cell viability (cytotoxicity) and antiviral efficacy contributed to the ELISA, producing a non-linear trend in the data sets.
The nanopolymers were generally highly active towards the Omicron strain, and the effect appeared to be correlated with the L-arginine content, with HBPL being the least effective. These results suggest that L-arginine-modified nanopolymers are able to increase the biological effect of HBPL towards the original and Omicron strains of Sars-CoV-2. The reason behind this behavior should be attributed to the differences in the way the nanoparticles interact with the virus as competitors to cellular receptors depending on the differences in the structural composition and surface charge. Further studies would be needed to verify this theory and understand the resistance displayed by the Delta variant.
The ELISA results obtained from the post-treatment method (
Figure 10d–f) generally show that the antiviral efficacy of the nanopolymers was strongly reduced when the particles were incubated with the cells after viral infection. This indicates that the main effect of the nanopolymers was to reduce the viral replication by inhibiting the virus entry into the cells. This result is in agreement with previous findings on poly-L-lysine hyperbranched nanoparticles. Time-of-addiction experiments have in fact shown a strong reduction in the antiviral activity when the particles were added post infection with respect to a pre-infection incubation [
10].
The efficacy of the different activities of the nanosystems should, however, be evaluated by also taking in consideration the cytotoxicity data. The antiviral efficacy data taken from the prevention model were used together with the cytotoxicity data to generate a therapeutic index value:
CC = 50% cytotoxic concentration and IC
50 = 50% inhibitory concentration, i.e., a measure of the balance between the compound’s efficacy and safety. The TI values for the tested nanopolymers are listed in
Table 1.
While all the nanopolymers exhibited a low TI when tested with the Delta variant, the original and Omicron strain samples exhibited the highest TI values (high efficacy, low cytotoxicity) with the Lys-Arg copolymers containing the least amount of L-arginine (LBA0.1 and LBA0.25). It is finally worth underlining that, in comparison to the hyperbranched nano-homopolymer, HBPL, the addition of a small amount of L-arginine increased the therapeutic index and the antiviral activity of the resulting copolymers by twelve-fold, while the addition of bigger amounts was accompanied by an increase in the cytotoxic effect. Based on these results, we assume that L-arginine-modified nanopolymers might be suitable as preventive tools to be used in a hypothetical nasal spray device to hinder the infection of Sars-CoV-2 and its variants [
28]. Further studies will be needed to evaluate the best formulation for in vivo experiments and exploit the spectrum of activity.