A broad-spectrum virus- and host-targeting antiviral peptide against SARS-CoV-2 and other respiratory viruses


 The 2019 novel respiratory virus (SARS-CoV-2) causes COVID-19 with rapid global socioeconomic disruptions and disease burden to healthcare. The current COVID-19 and previous emerging virus outbreaks highlight the urgent need for broad-spectrum antivirals. Here, we showed that a defensin-like peptide P9R exhibited potent antiviral activity against pH-dependent viruses that require endosomal acidification for virus-host membrane fusion, including the enveloped coronaviruses (SARS-CoV-2, SARS-CoV and MERS-CoV), the pandemic A(H1N1)pdm09 virus, avian influenza A(H7N9) virus, and the non-enveloped rhinovirus. P9R could significantly protect mice from lethal challenge by A(H1N1)pdm09 virus and showed low possibility to cause drug-resistance virus. Mechanistic studies indicated that the antiviral activity of P9R depended on the direct binding to viruses and the inhibition of virus-host endosomal acidification, which provides a new concept that virus-binding alkaline peptides could broadly inhibit pH-dependent viruses. These results suggest that the dual-functional virus- and host-targeting P9R could be a promising candidate for combating pH-dependent respiratory viruses.Authors Hanjun Zhao, Kelvin K. W. To, and Kong-Hung Sze contributed equally to this work.


Introduction
Novel respiratory viruses often cause severe respiratory tract infections and spread quickly due to the lack of pre-existing immunity. In the recent two decades, three highly China 6,7 . Due to the lack of effective antivirals, especially for coronaviruses, these respiratory viruses are associated with significant morbidity and mortality. Furthermore, these emerging respiratory viruses have also caused severe economic and social disturbances.
The COVID-2019 outbreak has clearly illustrated the importance of broad-spectrum antivirals. While an outbreak of unusual pneumonia was reported from Wuhan of China in December 2019, the identity of SARS-CoV-2 was reported on January 8, 2020 by China CDC 8 . Currently, there is no specific drug treatment for this new virus. An effective broadspectrum antiviral will improve patients' outcome and may reduce transmission in the community and hospitals even before the identification of the novel emerging virus and the specific antiviral drug.
In past decades, researchers and our group have devoted to the discovery of broadspectrum antivirals targeting hosts or viruses 9,10 . Most broad-spectrum antivirals are targeting host pathways which are utilized by many viruses in their life cycles for viral replication or by stimulating host antiviral immunity 11,12 . We have previously reported a broad-spectrum antiviral peptide P9 13 . This peptide, derived from mouse β-defensin-4, was found to have antiviral activity against SARS-CoV, MERS-CoV, and influenza viruses by inhibiting endosomal acidification. We have also reported antiviral peptides against influenza virus by delivering defective interfering gene 14  To determine whether the net charge of the peptide affects the inhibition of endosomal acidification, using the endosomal acidification assay, we identified that P9R (+5.6) could more significantly inhibit endosomal acidification in live cells than that of P9 (Fig. 2ab), which are consistent with the stronger antiviral activity of P9R than that of P9. In addition, peptide PA1 with less positive charge (+1.7), which has the same amino acid sequence as P9 except 3 additional acidic amino acid at the C terminal, could not inhibit endosomal acidification (Fig. 2ab) and lost the antiviral activity (Fig. 2c). Hence, the degree of net positive charge was correlated with the degree of inhibition of endosomal acidification and antiviral activity.
Inhibition of host endosomal acidification alone is not sufficient for positively charged peptide inhibiting virus replication To determine whether the antiviral activity solely relied on the positive charge of peptide, we designed a peptide P9RS (+5.6) which had the same positive charge as P9R (+5.6), but P9RS differed from P9R by 11 of 30 amino acids. P9RS efficiently inhibited host endosomal acidification to the similar degree as P9R in live cells (Fig. 2ab). However, in the plaque reduction assay, there was no significant reduction of plaque numbers for SARS-CoV-2 and A(H1N1)pdm09 virus when viruses were treated by P9RS even at 25 μg ml -1 (Fig. 2c).
To investigate why P9RS failed to inhibit viral replication despite potent inhibition of host endosomal acidification, we studied the binding between the peptide and virus. Using ELISA-RT-qPCR assay, P9R and PA1 could efficiently bind to SARS-CoV-2 and A(H1N1)pdm09 virus but P9RS did not bind to SARS-CoV-2 and A(H1N1)pdm09 virus (Fig.   2d). The observation of P9R but not P9RS binding to virus was further confirmed by confocal microscopy in H1N1-infected cells (Fig. 2e). Thus, the direct interaction of peptide with virus was required for the antiviral activity of positively charged peptide P9R.
In contrast, P9RS without the ability of binding to virus could not inhibit viral replication even though it carries the same positive charge as P9R and inhibits host endosomal acidification.
The broad-spectrum antiviral activity of P9R relies on targeting viruses to inhibit virus-host endosome acidification In the above experiments, we had demonstrated that P9R and P9RS can inhibit no-virus endosomal acidification (Fig. 2ab). However, without binding to virus, P9RS could not inhibit viral replication. To illustrate this result, we showed that P9R and bafilomycin A1 could efficiently inhibit the virus-host endosomal acidification in infected live cells, but P9RS could not inhibit the virus-host endosomal acidification in infected live cells (Fig.   3a), even though both of P9R and P9RS could inhibit the endosomal acidification of novirus endosomes (Fig. 2a). The efficient inhibition of P9R on virus-host endosomal acidification could be due to the binding of P9R to virus (Fig. 2de) and then inhibiting the virus-host endosomal acidification (Fig. 3a). Lacking the binding ability to viruses ( were empty spaces in no-virus endosomes to allow P9RS freely entering endosomes to prevent endosomal acidification (Fig. 2a). It should be noted that PA1 with a similar sequence as P9R could efficiently bind to SARS-CoV-2 and A(H1N1)pdm09 virus (Fig. 2d), but it significantly lost the antiviral activity against SARS-CoV-2 and A(H1N1)pdm09 virus (Fig. 2c). The binding of P9R to SARS-CoV-2 and A(H1N1)pdm09 virus could be significantly reduced when viruses were pretreated by PA1 (Fig. 3b). This indicated that PA1 had the same binding sites on viral particles as P9R but only peptide binding to virus alone could not account for the antiviral activity. P9R binding to virus was the first step to exert the antiviral activity.
To further confirm that broad-spectrum antiviral activity of P9R was due to the broadly bindings of P9R to different viruses and viral proteins, we demonstrated that P9R but not Next, we tried to determine the structure of P9R using NMR spectroscopy. The results indicated that the solution structure of P9R was flexible with short variable helical patches and with positively charged peptide surface (Fig. 3g). We hypothesized that P9R could broadly bind to different viruses because these short α-helical patches with flexible linkages may allow it to adapt its structure to fit the binding pockets of different viral proteins. More co-binding structure analysis will be needed for identifying the broadly binding mechanism of P9R with different viral proteins in future. In conclusion, here we demonstrated the novel antiviral mechanism that positively charged P9R needs targeting viruses and then prevents virus-host endosomal acidification to inhibit pH-dependent virus

replication. The efficacy of P9R treatment in vivo
We had demonstrated that the efficient antiviral activity of P9R in vitro is reliant on binding to viruses and the positive charge of P9R to inhibit virus-host endosomal acidification. To further investigate the antiviral activity of P9R in vivo, we treated A(H1N1)pdm09-infected mice at 6 h post infection with additional two doses in the following one day. In this model, 80% of P9R-treated mice survived, which was significantly better than PBS-treated group and PA1-treated group (Fig. 4a). The protection of P9R on infected mice was the same as that in the zanamivir-treated group (80%) and was better than P9. From day 4 to day 10 post infection, there was significantly less body weight loss in P9R group than that in PBS-treated group and PA1-treated group (Fig. 4b). The low dose protection of P9R (25 μg/dose and 12.5 μg/dose) on infected mice and reducing body weight loss further demonstrated that P9R could significantly protect mice when compared with PBS-treated group (Fig. 4cd). The antiviral activity of P9R in vivo was better than that of P9 (Fig. 4c, P<0.05 for 12.5μg/dose), which was consistent with the significantly better antiviral activity of P9R than P9 in vitro.

No emergence of resistant viruses against P9R after serial passages of virus in the presence of P9R
Emergence of resistant mutants occur from time to time 14  The 'one bug-one drug' approach to antiviral drug is successful for HIV, hepatitis C virus and influenza virus 9 . However, there is an urgent need for broad-spectrum antivirals for combating emerging and re-emerging new virus outbreaks, such as the SARS-CoV-2, before the new virus is identified or specific antiviral drug is available.
Endosomal acidification is a key step in the life cycle of many pH-dependent viruses, which is one of the broad-spectrum antiviral targets 9 . In this study, with the increased positive charge in P9R, it could more efficiently inhibit pH-dependent viruses than that of P9. The more positive charge in P9R allowed the peptide to more efficiently neutralize protons inside endosomes, and thereby inhibiting the endosomal acidification. In previous studies, the clinically approved anti-malarial drug chloroquine with activity of inhibiting endosomal acidification had been demonstrated to inhibit enterovirus-A7 30  The antiviral activity of P9R required both of binding to viruses and inhibiting endosomal acidification. PA1 with less positive charge could not inhibit SARS-CoV-2 and H1N1 virus even thought it had the similar binding ability and binding sites to viruses as P9R (Fig.   3b). When we made multiple substitutions on P9R to generate P9RS, P9RS lost the binding ability and antiviral activity to all tested viruses even though P9RS had the same positive charge as P9R and efficiently inhibited host endosomal acidification. The broadly binding mechanism of P9R to different viral proteins may be due to the flexible structure of P9R with positively charged surface (Fig. 3h). The flexible structure may allow P9R to change its structure to fit targeting proteins for broad-specificity bindings 37, 38 , and the positive charge of P9R may play roles for binding to viruses with negatively charged surface 39, 40 .
The five cysteines in P9R may also affect the structure-based binding because previous studies indicated that cysteine substitutions could affect defensin-peptide structure and activity 41, 42 .
In addition, comparing with zanamivir which caused significant drug resistance virus after 10-virus passages in the presence of zanamivir, P9R showed very low risk to cause drugresistance virus even A(H1N1)pdm09 virus was passaged in the presence of P9R for 40 passages. The low risk of resistance induction by P9R may be partially due to the dual targeting ability to broadly bind to viruses and target host endosomes. More co-binding structure analysis will be needed to illustrate the broadly binding mechanism and help to understand the low drug resistance mechanism in future study.
In summary, most highly pathogenic emerging viruses are endosomal pH-dependent viruses. The emerging and re-emerging virus outbreaks remind us of the urgent need of broad-spectrum antivirals. Continual studies on this kind of antivirals, which can broadly bind to different viruses and inhibit pH-dependent viruses by preventing virus-host endosomal acidification with low possibility of causing drug resistance, will give us more armamentarium to combat novel emerging virus outbreaks in future. Plaque reduction assay Antiviral activity of peptides was measured using a plaque reduction assay as we described previously 14

Cytotoxicity assay
Cytotoxicity of peptides was determined by the detection of 50% cytotoxic concentration (CC 50 ) using a tetrazolium-based colorimetric MTT assay as we described previously 13 .
Briefly, cells were seeded in 96-well cell culture plate at an initial density of 2 × 10 4 cells per well in MEM or DMEM supplemented with 10% FBS and incubated for overnight. Cell culture media were removed and then DMEM supplemented with various concentrations of peptides and 1% FBS were added to each well. After 24 h incubation at 37 °C, MTT solution (5 mg ml − 1 , 10 μl per well) was added to each well for incubation at 37 °C for 4 h.
Then, 100 μl of 10% SDS in 0.01M HCl was added to each well. After further incubation at room temperature with shaking overnight, the plates were read at OD570 using VictorTM X3 Multilabel Reader (PerkinElmer, USA). Cell culture wells without peptides were used as the experiment control and medium only served as a blank control.  14 . Extracted RNA was reverse transcribed to cDNA using PrimeScript II 1 st Strand cDNA synthesis Kit (Takara, Cat # 6210A) using GeneAmp® PCR system 9700 (Applied Biosystems, USA). The cDNA was then amplified using specific primers (Table S1) for detecting SARS-CoV-2, MERS-CoV, SARS-CoV, H1N1, H7N9, and rhinovirus using LightCycle® 480 SYBR Green I Master (Roach, USA). For quantitation, 10-fold serial dilutions of standard plasmid equivalent to 10 1 to 10 6 copies per reaction were prepared to generate the calibration curve. Real-time qPCR experiments were performed using LightCycler® 96 system (Roche, USA).

Endosomal acidification assay
Endosomal acidification was detected with a pH-sensitive dye (pHrodo Red dextran, Invitrogen, Cat # P10361) according to the manufacturer's instructions as previously described but with slight modification 14   Nucleoprotein (NP) immunofluorescence assay.
NP staining was carried out as described previously 14  spectra were recorded for resonance assignments. Inter-proton distance restraints were derived from 2D NOESY spectrum with mixing times of 300 ms and 500 ms using automated NOE assignment strategy followed by a manual check. NOE intensities and chemical shifts were extracted using CCPNMR Analysis 2.4.2 46 and served as inputs for the Aria program. Dihedral angle is predicted from the chemical shifts using the program DANGLE 47 . The NMR solution structure of P9R was calculated iteratively using Aria 2.3 program 48 . One hundred random conformers were annealed using distance restraints in each of the eight iteratively cycles of the combined automated NOE assignments and structure calculation algorithm. The final upper limit distance constraints output from the last iteration cycle were subjected to a thorough manual cross-checking and final water solvent structural refinement cycle. The 10 lowest energy conformers were retained from these refined 100 structures for statistical analysis. The convergence of the calculated structures was evaluated using root-mean-square deviations (RMSDs) analyses. The distributions of the backbone dihedral angles (φ, ψ) of the final converged structures were evaluated by representation of the Ramachandran dihedral pattern using PROCHECK-