Reprogramming of the Antibacterial Drug Vancomycin Results in Potent Antiviral Agents Devoid of Antibacterial Activity

Influenza A and B viruses are a global threat to human health and increasing resistance to the existing antiviral drugs necessitates new concepts to expand the therapeutic options. Glycopeptide derivatives have emerged as a promising new class of antiviral agents. To avoid potential antibiotic resistance, these antiviral glycopeptides are preferably devoid of antibiotic activity. We prepared six vancomycin aglycone hexapeptide derivatives with the aim of obtaining compounds having anti-influenza virus but no antibacterial activity. Two of them exerted strong and selective inhibition of influenza A and B virus replication, while antibacterial activity was successfully eliminated by removing the critical N-terminal moiety. In addition, these two molecules offered protection against several other viruses, such as herpes simplex virus, yellow fever virus, Zika virus, and human coronavirus, classifying these glycopeptides as broad antiviral molecules with a favorable therapeutic index.


Introduction
Seasonal infections by influenza A and B viruses are each year responsible for significant morbidity and mortality [1]. Besides, zoonotic influenza A viruses occasionally enter the human population to cause serious pandemics with a high number of fatalities [2]. Antiviral drugs are essential for influenza treatment and prevention, including in the context of pandemic preparedness. At the moment, four drug classes are available: The M2 ion channel blockers and neuraminidase inhibitors, approved in all countries [3], and two polymerase inhibitors, recently approved in a few countries [4,5]. For each of these drugs, emergence of resistant mutants is possible; this is particularly problematic when the mutant viruses are fit and human-to-human transmissible [6][7][8][9]. Hence, additional drug classes with a distinct mechanism of action remain essential [10][11][12].
Several recent investigations demonstrated that some antibiotics from bacterial origin display interesting antiviral properties [13] representing a unique example of drug repurposing [14].
To synthesize the first variant, we used literature procedures to prepare vancomycin aglycone (1) and its hexapeptide derivative (2) by the Edman degradation (Scheme 1) [35,36]. Then, by the copper-catalyzed diazotransfer reaction, the N-terminal azido derivative (3) was successfully prepared in analogy to our previous work [24,30].
Pharmaceuticals 2020, 13,139 3 of 15 To synthesize the first variant, we used literature procedures to prepare vancomycin aglycone (1) and its hexapeptide derivative (2) by the Edman degradation (Scheme 1) [35,36]. Then, by the copper-catalyzed diazotransfer reaction, the N-terminal azido derivative (3) was successfully prepared in analogy to our previous work [24,30]. The subsequent copper-catalyzed azide-alkyne click reaction using alkyne compound 4 (Scheme 2) yielded triazole derivative 6. The second compound modified on the N-terminus (7) was prepared by the same method using the already described maleimide derivative 5 [30] as the alkyne compound in the final step. The third derivative (8) in this group was synthesized using derivative 2 and hexanesulfonyl chloride by sulfonamide formation, similar to what we previously published for teicoplanin congeners [30]. The subsequent copper-catalyzed azide-alkyne click reaction using alkyne compound 4 (Scheme 2) yielded triazole derivative 6. The second compound modified on the N-terminus (7) was prepared by the same method using the already described maleimide derivative 5 [30] as the alkyne compound in the final step. The third derivative (8) in this group was synthesized using derivative 2 and hexanesulfonyl chloride by sulfonamide formation, similar to what we previously published for teicoplanin congeners [30].

Scheme 2.
Structures of previously prepared alkynes 4 and 5 used for the synthesis of derivatives 6 and 7.
As for the C-terminal modifications, we decided to prepare two derivatives carrying a 1,2,3-triazole ring, since this moiety often generates bioactive analogues [37]. In order to minimize the structural difference between the side chain in the N-versus C-terminal position, we used 2-azidoethylamine as a small linker for the synthesis of the appropriate C-terminal analogues (Scheme 3). After preparation of the amines, the reaction of compound 2 with amine 9 using the PyBOP reagent gave amide derivative 11 (Scheme 4). Using the same conditions, the reaction between compound 2 and maleimide 10 as the alkyne compound yielded compound 12. In the case of the Cterminal sulfonamide derivative 13 we also used the similar, small linker moiety. In this approach, an amine functionalized sulfonamide was first synthesized from n-hexanesulfonyl chloride and ethylenediamine, then the peptide coupling reaction between this compound and compound 2 using PyBOP yielded compound 13.

Scheme 2.
Structures of previously prepared alkynes 4 and 5 used for the synthesis of derivatives 6 and 7.
As for the C-terminal modifications, we decided to prepare two derivatives carrying a 1,2,3-triazole ring, since this moiety often generates bioactive analogues [37]. In order to minimize the structural difference between the side chain in the N-versus C-terminal position, we used 2-azidoethylamine as a small linker for the synthesis of the appropriate C-terminal analogues (Scheme 3). As for the C-terminal modifications, we decided to prepare two derivatives carrying a 1,2,3-triazole ring, since this moiety often generates bioactive analogues [37]. In order to minimize the structural difference between the side chain in the N-versus C-terminal position, we used 2-azidoethylamine as a small linker for the synthesis of the appropriate C-terminal analogues (Scheme 3). After preparation of the amines, the reaction of compound 2 with amine 9 using the PyBOP reagent gave amide derivative 11 (Scheme 4). Using the same conditions, the reaction between compound 2 and maleimide 10 as the alkyne compound yielded compound 12. In the case of the Cterminal sulfonamide derivative 13 we also used the similar, small linker moiety. In this approach, an amine functionalized sulfonamide was first synthesized from n-hexanesulfonyl chloride and ethylenediamine, then the peptide coupling reaction between this compound and compound 2 using PyBOP yielded compound 13. Scheme 3. Synthesis of amines 9-10 for the C-terminal modifications.
After preparation of the amines, the reaction of compound 2 with amine 9 using the PyBOP reagent gave amide derivative 11 (Scheme 4). Using the same conditions, the reaction between compound 2 and maleimide 10 as the alkyne compound yielded compound 12. In the case of the C-terminal sulfonamide derivative 13 we also used the similar, small linker moiety. In this approach, an amine functionalized sulfonamide was first synthesized from n-hexanesulfonyl chloride and ethylenediamine, then the peptide coupling reaction between this compound and compound 2 using PyBOP yielded compound 13.

Antibacterial Evaluation
Antibacterial tests were carried out by the broth microdilution method on a panel of eight Grampositive bacterial strains, using vancomycin and teicoplanin as reference compounds. Neither of the new compounds exhibited significant activity against any bacterium, proving successful elimination of antibacterial activity, as anticipated (Table 1).

Antibacterial Evaluation
Antibacterial tests were carried out by the broth microdilution method on a panel of eight Gram-positive bacterial strains, using vancomycin and teicoplanin as reference compounds. Neither of the new compounds exhibited significant activity against any bacterium, proving successful elimination of antibacterial activity, as anticipated (Table 1).

Antiviral Evaluation
With regard to antiviral activity, the two N-terminal triazole derivatives 6 and 7 displayed robust activity against the three influenza A or B viruses tested. Upon microscopic inspection, no virus-induced cytopathic effect (CPE) was observed in virus-infected cells treated with 6.25 µM of compound 6 or compound 7 ( Figure 1A). The quantitative antiviral efficacy (EC 50 ) and cytotoxicity (CC 50 ) values, both determined by the MTS cell viability assay, are summarized in Table 2. With EC 50 values of~3 µM and a CC 50 value of 41 µM (compound 6) and 18 µM (compound 7), the molecules had a selectivity index (ratio of CC 50 to EC 50 ) of 14 and 6, respectively. Both molecules exhibited clear inhibition of influenza virus replication, since they strongly reduced the virus yield in the supernatant ( Figure 1B), giving EC 99 values of 3.5 µM (compound 6) and 4.6 µM (compound 7), which is 5-to 6-fold lower than the EC 99 for ribavirin (23 µM). At these concentrations, the compounds were devoid of cytotoxicity, as assessed by MTS cell viability assay in mock-infected cells ( Figure 1C). The N-terminal n-hexanesulfonyl derivative 8 and C-terminally modified compound 11 proved inactive. For compound 8, this was somewhat surprising since the analogous teicoplanin pseudoaglycone derivative showed high activity [30]. On the other hand, this result is in line with our previous findings on ristocetin and teicoplanin aglycone derivatives, indicating that even minor structural differences in the peptide core can lead to significantly different anti-influenza virus activity [27]. The C-terminally modified compounds 12 and 13 were only slightly active against one or both influenza A virus strains. Compared to compounds 6 and 7, compound 13 displayed an 8-fold higher antiviral EC 50 value by MTS assay ( Table 2); its lower potency was also evident in the virus yield reduction assay ( Figure 1B). This points to the importance of the modification site, since the C-terminally modified compounds were clearly inferior to the N-terminal analogues.

Antiviral Evaluation
With regard to antiviral activity, the two N-terminal triazole derivatives 6 and 7 displayed robust activity against the three influenza A or B viruses tested. Upon microscopic inspection, no virusinduced cytopathic effect (CPE) was observed in virus-infected cells treated with 6.25 µM of compound 6 or compound 7 ( Figure 1A). The quantitative antiviral efficacy (EC50) and cytotoxicity (CC50) values, both determined by the MTS cell viability assay, are summarized in Table 2. With EC50 values of ~3 µM and a CC50 value of 41 µM (compound 6) and 18 µM (compound 7), the molecules had a selectivity index (ratio of CC50 to EC50) of 14 and 6, respectively. Both molecules exhibited clear inhibition of influenza virus replication, since they strongly reduced the virus yield in the supernatant ( Figure 1B), giving EC99 values of 3.5 µM (compound 6) and 4.6 µM (compound 7), which is 5-to 6fold lower than the EC99 for ribavirin (23 µM). At these concentrations, the compounds were devoid of cytotoxicity, as assessed by MTS cell viability assay in mock-infected cells ( Figure 1C). The Nterminal n-hexanesulfonyl derivative 8 and C-terminally modified compound 11 proved inactive. For compound 8, this was somewhat surprising since the analogous teicoplanin pseudoaglycone derivative showed high activity [30]. On the other hand, this result is in line with our previous findings on ristocetin and teicoplanin aglycone derivatives, indicating that even minor structural differences in the peptide core can lead to significantly different anti-influenza virus activity [27]. The C-terminally modified compounds 12 and 13 were only slightly active against one or both influenza A virus strains. Compared to compounds 6 and 7, compound 13 displayed an 8-fold higher antiviral EC50 value by MTS assay ( Table 2); its lower potency was also evident in the virus yield reduction assay ( Figure 1B). This points to the importance of the modification site, since the C-terminally modified compounds were clearly inferior to the N-terminal analogues.   Encouraged by the promising anti-influenza virus activity of compounds 6 and 7, we tested the two compounds against a range of DNA-and RNA-viruses evaluated in human embryonic lung (HEL) fibroblast, HeLa or Vero cells. For each virus, appropriate reference compounds were included. Protection against virus-induced cytopathicity as well as compound cytotoxicity were determined by the MTS cell viability assay. As shown in Table 3, the two compounds exhibited broad protection against a large variety of viruses, including herpesvirus types 1 and 2 and vaccinia virus. They retained full effectivity against a thymidine kinase deficient form of HSV-1, which was 61-fold (acyclovir) and 89-fold (ganciclovir) resistant to antiherpetic drugs. Moreover, the compounds proved effective against two emerging pathogens for which no therapy is currently approved, i.e., coronavirus (inhibited by compounds 6 and 7) and Zika virus (inhibited by compound 7). At a non-toxic concentration of 25 µM of compound 6, no coronavirus 229E-induced cytopathicity could be observed microscopically (Figure 2A), which agrees with an EC 50 value of 11 µM as determined by MTS cell viability assay (Table 3). In addition, treatment of infected cells with 25 µM of compound 6 resulted in a 1000-fold reduction of the viral RNA copy number in the supernatant, yielding an EC 99 value of 20 µM for reduction of virus yield ( Figure 2B). Hence, we established, by virus yield assays, that compound 6 suppresses the replication of influenza virus and coronavirus, and for the other viruses, activity was indicated by the protection against viral CPE. This broad activity against distinct viruses fits with our hypothesis that these molecules may act by disrupting the viral endocytosis process, similarly to what we reported for a glycopeptide active against influenza virus [23] and what was described for Ebola virus and MERS and SARS coronaviruses [15,16]. This should become clear from mechanistic work ongoing in our laboratory.

Chemistry
Vancomycin hydrochloride was a gift from TEVA Pharmaceutical Industries Ltd. (Debrecen, Hungary). Vancomycin aglycone hexapeptide, trifluoromethanesulfonyl azide, compounds 4 and 5 were prepared as described elsewhere [24,30]. TLC was performed on Kieselgel 60 F254 (Merck) with detection either by immersing into ammonium molybdate-sulfuric acid solution followed by heating

Chemistry
Vancomycin hydrochloride was a gift from TEVA Pharmaceutical Industries Ltd. (Debrecen, Hungary). Vancomycin aglycone hexapeptide, trifluoromethanesulfonyl azide, compounds 4 and 5 were prepared as described elsewhere [24,30]. TLC was performed on Kieselgel 60 F 254 (Merck) with detection either by immersing into ammonium molybdate-sulfuric acid solution followed by heating or by using Pauly's reagent for detection. Flash column chromatography was performed using Silica gel 60 (Merck 0.040-0.063 mm) and Silica gel 60 silanized (0.063-0.200 mm). The 1 H NMR (500MHz, 400 MHz) 13 C NMR (125 MHz, 100 MHz) and 2D NMR spectra were recorded with a Bruker DRX-400 and Bruker Avance II 500 spectrometers at 300K. Chemical shifts are referenced to Me 4 Si and to the solvent residual signals. MALDI-TOF MS analysis of the compounds was carried out in the positive reflectron mode using a BIFLEX III mass spectrometer (Bruker, Bremen, Germany) equipped with delayed-ion extraction. 2,5-Dihydroxybenzoic acid (DHB) was used as matrix and CF 3 COONa as cationizing agent in DMF. Elemental analysis (C, H, N, S) was performed on an Elementar Vario MicroCube instrument.

Synthesis
Synthesis of azido vancomycin aglycone hexapeptide (3): 350 mg (0.34 mmol) vancomycin aglycone hexapeptide (2) was obtained from 750 mg (0.5 mmol) vancomycin hydrochloride (1) by Edman degradation as described in the literature [35,36]. Sodium azide (65 mg, 1.0 mmol) was added to dry pyridine (1.5 mL) cooled to 0-5 • C. Tf 2 O (0.8 mmol, 134 µL) was added dropwise over the course of about 30 min. The reaction mixture was stirred for another 2 h at 0-5 • C. Then, 350 mg (0.34 mmol) of 2 was dissolved in 15 mL pyridine, then 95 µL (2.0 equiv., 0.68 mmol) Et 3 N was added followed by the solution (1.5 mL) of the freshly prepared trifluoromethanesulfonyl azide, and finally 800 µL of a 10 mg/mL CuSO 4 · 5H 2 O solution. The solution was stirred at room temperature overnight, then the solvents were evaporated. The crude product was dissolved in dilute NH 4 OH, then the pH was set to 1-2 with 1N HCl, the resulting cloudy mixture was extracted with n-BuOH three times, the butanolic phase was washed with water, then evaporated and purified by flash column chromatography using step gradient elution (MeCN:H 2 O = 100:0→97:3→94:6→92:8). The title compound was obtained in 250 mg yield (70%) as an off-white powder. NMR: see Synthesis of compound 6: 132 mg (0.127 mmol) of 2 was dissolved in 2.0 mL of DMF:H 2 O 3:1 mixture. Next, 72 mg (1.25 equiv., 0.16 mmol) of alkyne 4 was added, followed by 6 mg CuSO 4 · 5H 2 O. The reaction mixture was stirred at room temperature for 12 h. By this time TLC (cellulose, n-PrOH: cc. NH 4 OH = 6:4) indicated good conversion. The solvents were evaporated until a syrupy residue was obtained. Ether was added, and the product was filtered off after precipitation and washed with additional ether to remove the excess alkyne. Purification was carried out by C 18 reverse phase column chromatography (H 2 O:MeCN = 70:30→60:40→55:45) followed by gel chromatography using Sephadex LH-20 in MeOH. The title compound was obtained as a white powder in 69 mg yield (37%). NMR: see Table S2 in supporting information. Elemental analysis: see Table S8  After 12 h stirring at room temperature, TLC indicated good conversion. The reaction mixture was worked up and purified by C 18 reverse phase column chromatography as described above. The title compound was obtained as a yellow powder in 52 mg yield (38%). NMR: see Table S3 in supporting information. Elemental analysis: see Table S8 in supporting information. MALDI-MS m/z calcd. for C 69 H 74 Cl 2 N 10 O 22 S 2 + Na + [M + Na] + : 1551.369. Found: 1551.367.

Anti-Influenza Virus Activity
For antiviral testing, 25 mM compound stocks were prepared in 100% DMSO and stored at 4 • C. The compounds were fully soluble under these conditions. In the antiviral tests, the highest concentration tested was 100 µM, corresponding to a non-toxic DMSO content of 0.4%.
The virus strains (A/H1N1: A/Ned/378/05 and A/PR/8/34; A/H3N2: A/Victoria/361/11; and B/Ned/537/05) were propagated in embryonated hen eggs. The antiviral procedure was published elsewhere [38,39]. Madin-Darby canine kidney (MDCK) cells were seeded at 7500 cells per well into 96-well plates, using infection medium (UltraMDCK medium (Lonza) with 225 mg/L sodium bicarbonate, 2 mM L-glutamine, and 2 µg/mL N-tosyl-L-phenylalanine chloromethyl ketone (TPCK)-treated trypsin). One day later, virus (MOI: 0.001) was added together with 1:5 serial compound dilutions, to reach a total volume of 200 µL per well. Besides the test compounds, two references were included, i.e., zanamivir and ribavirin (positive controls) plus a condition receiving medium instead of compound (negative control). In parallel, the compound dilutions were also added to a mock-infected plate (in which medium was added instead of virus), to determine compound cytotoxicity. Each plate contained two wells in which all reagents yet no cells were added, to serve as blanks in the MTS calculations. After three days incubation at 35 • C, viral CPE was first monitored by microscopy. Then, the supernatants were replaced by MTS reagent (CellTiter 96 ® AQ ueous MTS Reagent from Promega) diluted 1:10 in PBS, and 4 h later, absorbance at 490 nm was measured in a plate reader. To monitor the inhibitory effect of the compounds on virus replication, MDCK cells were seeded, infected (with A/PR/8/34 virus; MOI: 0.001) and treated with 1:2 serial compound dilutions. The plate contained three virus controls (receiving no compound) and two cell controls (receiving no virus and no compound). At day 3 p.i., supernatants were collected and frozen at -80 • C, to quantify the virus yield by one-step qRT-PCR analysis of viral copy number [40]. Two µl supernatant was mixed with 10 µL resuspension buffer and 1 µl lysis reagent (CellsDirect One-Step RT-qPCR kit; Invitrogen) and heated during 10 min at 75 • C. Next, 10 µL lysate was transferred to a qPCR plate containing the qRT-PCR enzymes and buffer (CellsDirect One-Step RT-qPCR kit; Invitrogen), and influenza virus M1-specific primers and probe [40]. The program consisted of: 15 min at 50 • C; 2 min at 95 • C; and 45 cycles of 15 s at 95 • C followed by 90 s at 60 • C. Absolute quantification of vRNA copies was performed by including an M1-plasmid standard. The EC 99 values were calculated by interpolation and defined as the compound concentration causing 100-fold reduction in vRNA copy number, as compared to the virus control receiving no compound. It was ascertained that the cell controls showed no detectable qPCR signal.

Other Antiviral Procedures
The viruses were propagated and evaluated in the following cell lines: Human embryonic lung (HEL) fibroblast cells, used for human coronavirus 229E [41], herpes simplex virus type 1 (HSV-1 strain KOS, including a thymidine kinase deficient HSV-1/TKmutant), herpes simplex virus type 2 (HSV-2, strain G) and vaccinia virus (strain Lederle); human cervixcarcinoma HeLa cells, used for respiratory syncytial virus (RSV, strain Long), and African Green Monkey kidney Vero cells, used for yellow fever virus (vaccine strain 17D) and Zika virus (strain MR766). The medium used for virus infection was Dulbecco's Modified Eagle's Medium supplemented with 2% fetal calf serum. To prepare virus stocks, confluent cell cultures in 75-cm 2 flasks were infected with the virus and frozen after 3 to 5 days incubation at 37 • C (or 35 • C in case of human coronavirus 229E), when full-blown CPE was visible. After freeze-thawing and centrifugation, the clarified lysates were stored at −80 • C. For the antiviral experiments, the cells were grown in 96-well plates until confluent. Virus was added (MOI: 0.001) together with 1:5 serial dilutions of the compounds. For each virus, appropriate reference compounds were included. The compound dilutions were also added to a mock-infected plate, to determine compound cytotoxicity. When manifest CPE was reached, i.e., after 3 to 5 days incubation at 37 • C (or 35 • C in case of human coronavirus 229E), CPE and compound cytotoxicity were quantified by the MTS assay, and EC 50 and CC 50 values were calculated as explained above for influenza virus.
To assess inhibition of human coronavirus 229E replication, HEL cells were infected and treated with 1:2 serial compound dilutions. The plate contained three virus controls (receiving no compound) and two cell controls (receiving no virus and no compound). At day 4 p.i., supernatants were frozen at −80 • C to determine the virus yield by one-step RT-qPCR assay. Two microliters supernatant was mixed with 11 µl lysis mix containing lysis enhancer and resuspension buffer at a 1:10 ratio (CellsDirect One-Step RT-qPCR kit; Invitrogen), and heated for 10 min at 75 • C. Five microliters of lysate was transferred to a PCR plate containing 9.75 µl RT-qPCR mix (CellsDirect One-Step RT-qPCR) and 0.25 µl Superscript III RT/Platinum Taq enzyme, and coronavirus-229E N-gene specific primers and probe (forward primer 5 -TTAGAGAGCGTGTTGAAGGTG-3 ; reverse primer 5 -GTTCTGAATTCTTGCGCCTAAC-3 ; probe 5 -FAM-TCTGGGTTG-ZEN-CTGTTGATGGTGCTA-IBFQ-3 ). The RT-qPCR program consisted of 15 min at 50 • C, 2 min at 95 • C, and 50 cycles of 15 s at 95 • C and 45 s at 60 • C. An N-gene plasmid standard was included for absolute quantification. Compound activity was expressed as the EC 99 value, i.e., the concentration causing 100-fold reduction in vRNA copy number, as compared to the virus control receiving no compound. It was ascertained that the cell controls showed no detectable qPCR signal.

Conclusions
Starting from vancomycin, we have successfully prepared two derivatives with strong activity against influenza virus. The modifications that we introduced were based on our previous work on the glycopeptide antibiotic teicoplanin. Interestingly, some of these modifications yielded compounds 6 and 7 having the same antiviral potency as the analogous teicoplanin pseudoaglycon derivatives [24,30] but lacking antibacterial activity. The short work described here validates that the glycopeptide scaffold is an underexplored entity to conceive new antivirals with a broad activity spectrum, that besides influenza virus includes emerging pathogens like coronavirus and Zika virus.