Ligand-based design of peptide entry inhibitors targeting the endosomal receptor binding site of filoviruses

Abstract

Filoviruses enter cells through micropinocytosis and trafficking into the endosomes in which they bind to the receptor Niemann-Pick C1 protein (NPC1) for membrane fusion and entry into the cytoplasm. The endosomal receptor-binding is critical step for filovirus entry. Designing inhibitors to block receptor binding will prevent viral entry. Using available binding structural information from the co-crystal structures of the viral GP with the receptor NPC1 or with monoclonal antibodies, we have conducted structure-based design of peptide inhibitors to target the receptor binding site (RBS). The designed peptides were tested for their inhibition activity against pseudo-typed or replication-competent viruses in a cell-based assay. The results indicate that these peptides exhibited strong activities against both Ebola and Marburg virus infection. It is expected that these peptides can be further developed for therapeutic use to treat filovirus infection and combat the outbreaks.

1. Introduction

Ebola virus (EBOV) and Marburg virus (MARV) are well-known members of the filoviruses family and have caused human severe hemorrhagic fever and deadly outbreaks with high mortality rates from 25% to 90% (1). A large EBOV outbreak from 2013 to 2016 in West Africa infected 28,616 and caused 11,310 deaths (2). A recent EBOV outbreak in Kivu, Democratic Republic of the Congo (DRC) in 2018–2020, gave 3317 confirmed cases and 2,280 deaths (WHO). Filoviruses can be transmitted by direct contact with an infected person’s bodily fluids or with contaminated materials. Although MARV and EBOV were first identified in 1967 and 1976, respectively, a lack of approved drug treatments emphasizes the need for new therapeutic strategies (3).

Filoviruses are enveloped single-stranded negative-sense RNA viruses with a genome of about 19 kb which encodes seven structural proteins: the nucleoprotein (NP), the virion protein (VP) 24, VP35, VP30, VP40, the glycoprotein (GP), and the RNA-dependent RNA polymerase (L)(4-6). The glycoprotein (GP) is the sole viral protein on the virion surface and is important for cell entry, the first step in viral infection. The GP is initially synthesized as a single polypeptide (~676 residues), that is later is cleaved by cellular furin protease into disulfide-linked two subunits of GP1 and GP2 (7-9). The GP1 is exposed on the surface of the virion with GP2 serving as a membrane anchor and contains the fusion loop. The glycoprotein spike displays a trimeric structure which consists of three GP molecules including three GP1 and GP2. After the virus is taken into the cell through macropinocytosis and trafficking through endosomes (10, 11). EBOV GP1 is further cleaved by endosomal cysteine protease cathepsins (catL and catB) to remove the glycan cap and glycosylated mucin domain to expose the receptor binding site (RBS) for receptor binding (12-14). For MARV, cathepsin cleavage appears to be not directly needed for activation (15). Although, there are some differences in the GP structures and fusion activation between EBOV and MARV (16), both require the same endosomal receptor, Niemann-Pick C1 protein (NPC1), for release into the cell cytoplasm (17-19). The cleaved EBOV GP (GPcl) is also called primed GP which includes GP1 and GP2 portions to form a domain for interacting with receptor NPC1 domain C (20, 21). The GPcl binding to the cellular receptor NPC1 will cause GP2 conformational change for membrane fusion including the exposure of GP2 fusion loop (FL) that will reach to the cellular membrane to form a pre-hairpin structure. Under the affinity of GP2 N-terminal (NHR) and C-terminal (CHR) heptad repeat regions, a 6-helix bundle will form and pull the two membranes together in proximity for fusion and finally lead to a post-conformation of the hairpin formation, so that the viral capsid enters the cell cytoplasm and the genome can be released for replication (22, 23). It is evident that receptor binding is a key step for triggering the viral entry process. Thus, blocking receptor binding will block viral entry so that the receptor binding site (RBS) become an ideal drug target for preventing viral infection (24). In MARV, monoclonal antibodies binding to the RBS have exhibited strong neutralizing activities against MARV infection such as MR78 (25) and MR191 (26) since the MARV GP RBS has exposed on the virion surface (16). Thus, the MARV RBS has demonstrated to be the best target for antibody neutralization and drug development (18, 21, 27, 28). Therefore, we have used structure-based approaches to design peptide inhibitors to target this site according to the ligand bindings, such as the receptor NPC1 and the monoclonal antibodies. Since this RBS has a large cavity and can be fit by a larger molecule like short peptides, which may be even more effective to block the receptor binding than small molecules in terms of the physical barrier to interfere the receptor binding. The receptor NPC1 based cyclic peptide inhibitors were also reported to show strong activities against pseudotyped Ebola virus in a cell-based assay (29). Peptide based therapies have the advantages for treating infectious viral diseases since their strong selectivity and low toxicity (29, 30). Thus, it could be rapidly developed for clinical use against filovirus outbreaks. Here, we show successful design of peptide inhibitors based on RBS binding ligands that can effectively block viral entry. Our findings demonstrate that in a cell-based system these peptides can efficiently inhibit both pseudotyped and wild type virus infection.

2. Methods and materials

2.1. Peptides, strains, and plasmids

The designed peptides were synthesized from the KE-BioChem and PepMic Companies (China). The viral envelope glycoprotein genes used for peudotyped viruses were the Ebola virus (Zaire ebolavirus, GenBank: AIO11753.1), Marburg virus (Marburg marburgvirus Uganda 02Uga07, GenBank: ACT79201.1) and the VSV-G was from Vesicular stomatitis virus (VSV). The HIV-1 plasmid pSG3ΔEnv was from NIH AIDS Reagent Program.

2.2. Pseudotyping viruses

HIV-1 backbone plasmid pSG3ΔEnv was used for making pseudotyped viurses. The glycoprotein genes (GPs) of Ebola and Marburg viruses were synthesized and cloned into the pCDNA3.1+ expression vector. Both plasmids of pSG3ΔEnv and the GP envelope were co-transfected into 293T cells in a 10-cm plate using transfection reagent polyethyleneimine (PEI). The plates were cultured in a tissue incubator at 37°C and 5% CO₂ for two days, then the medium was harvested and centrifuged at 5000 rpm for 10 min to remove cell debris. The supernatants containing the pseudotyped were made aliquots and stored at −80°C.

2.3. Inhibition assay against pseudotyped filoviruses

Inhibition assay was performed in a 96-well plate using pseudotyped viruses and TZM-bl cells (6000/well) as this cell-line has a Luciferase report gene under the inducible promoter of Tat protein factor. The mixtures of viruses and peptide samples were transferred onto the target cell wells for infection. One-day post infection, the media were removed, the cells were washed once with PBS and incubated in fresh media for one more day. Then the cells were lysed in 1X Passive Lysis Buffer (Promega) and kept at room temperature for 20 minutes. The luciferase activity was measured using luciferin substrate (Promega) in a Veritas Luminometer. All the samples tested in triplicates and the neutralization activities were calculated in comparison with controls of positive (virus only) and negative (cells only).

2.4. Inhibition assay against replication-competent Ebola virus

As previously described (PMID 33705865), HeLa cells were preincubated for 1 h with peptides previously dissolved in water, before being challenged with wild type Ebola virus Zaire at an MOI of 0.1 to 0.3. After 36 h, cells were fixed in formalin and removed from the BSL-4 lab for staining with virus specific monoclonal antibody against GP (4F3, IBT Bioservices, MD, USA) followed by an Alexa 488-labeled anti-mouse secondary antibody (ThermoFisher). Cell nuclei were stained using Hoechst 33342 (ThermoFisher) at 1:10,000. Wells were then imaged using an automated imaging microscope (Cytation 1, Biotek, VT, USA). Images were processed using CellProfiler (Broad Inst, MA, USA) to count infected and total cells using a custom pipeline that is available upon request. Infection efficiency was calculated as the ratio of infected cells to total cells. Experiments were performed with at least 3 replicates.

2.5. Cell viability assay

Cell viability was measured by MTT assay. The TZM-bl cells (3000/well) were seeded in a 96-well plate and incubated for 24h at 37°C. Media was removed and replaced with 100μl of peptide solution in triplicates for two days, and then replace the peptide solution with 100μl complete DMEM for continuingly culturing for one more day. The cultured media were removed, and the cells were washed once with PBS for analysis. A 50μl solution of 5mg/ml MTT [3-4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] (Sigma-Aldrich, MO, USA) was added to each well. The plates were incubated for 3 h at 37°C, and the absorbance was measured at 570nm and with 650nm (background) wavelength.

2.6. Molecular docking

The molecular docking was conducted using the Web-based program HDOCK (31-33) for protein-protein docking or HPepDock (33-35) for protein-peptide docking. The specific residues were given for running a local docking in peptide docking, but protein docking, a template-based docking was performed based on the homology structural models. For EBOV-GP receptor-based docking, the input residues are: V79, W86, W88, L111, L122, V141, P146, A148, A152 or for MARV-GP receptor docking: W70, W72, L95, L106, I125, P130, A136. The ligand specific residues were defined by the key residues Phe (F) or Try (Y) for the pocket binding in the receptor binding site. By default, hundred docking models for each docking were generated and the top 10 models were evaluated based on the binding energies and poses.

2.7. Data statistical analysis

Graphpad Prism software was used for all statistical analyses for making the neutralization and cytotoxicity figures and determining average values, standard errors, values of IC₅₀ or CC₅₀.

3. Results

3.1. Structure-based design of peptide inhibitors

The co-crystal structure of NPC1 and EBOV GPlc (PDB 5F1B) has provided the direct contact information for NPC1 binding. There is a large hydrophobic cavity (~1000Å) in the EBOV GPlc to which the two loops, loop1 and loop2 (L1 and L2) of NPC1 are mainly involved in the binding (20). Especially, the major binding L2 has a hydrophobic residue Phenylalanine (F) 131 inserting into the deep pocket in the binding site. In addition, another adjacent F132 is also presented in this binding cavity. The Tyrosine (Y) 51 in the L1 is also involved the binding at the binding site (Fig 1A and B). These contacts indicate more hydrophobic residues such as Phe (F) and Tyr (Y) rich in the N-terminal side of the F131 but there are more charged residues such as Aspartic acid (D) and Lysine (K) residues in the N-terminal side (Fig. 1C). Besides the information from the receptor NPC1 binding in EBOV, there are some more information from monoclonal antibody binding to the MARV GP receptor binding-site (RBS), such as antibodies MR78 (PDB 5UQY)(25) and MR191 (PDB 6BP2) (26) since both of them bind to the receptor binding site (RBS) as the receptor NPC1 does. Thus, their bindings can also provide useful information for the inhibitor design. Although their protein sequences of EBOV GP and MARV GP are quite different (about 60% difference), they are structurally more similar that is why they can use the same receptor NPC1 for entry (16, 17). From the superimposed structures of the receptor NPC1 with the antibodies MR78 and MR191, it is apparent that their heavy chain loop 3 (HCDR3) were superimposed with the NPC1 loop 2 (L2) and the obvious common feature is a Phenylalanine residue (F106) inserting into the deep pocket like the Phe131 of receptor NPC1 (Fig. 1D). This Phenylalanine residue (F*) must has played a critical role in this binding. Additionally, the antibody binding appears to have more tyrosine (Y) involved which is another feature and has a strong indication that the tyrosine is a more fitting residue for the pocket binding. Based on these binding features of the receptor and monoclonal antibody bindings with the RBS, we conducted the design to generate short peptides for testing (Fig. 1C). The receptor binding domains (RBDs) in their GP1 of EBOV and MARV have about 45% identity and 60% similarity that are more similar than the other regions in the GPs. Thus, they can be well superimposed in a RSMD value of 2.2104 in the Cα atom of 140 residues.

Fig. 1.

Ligand structure-based design of peptide inhibitors. The binding residues of receptor NPC1 loops L1 the and L2 at the binding site of GPcl (PDB 5FB1). A. Ribbon models of NPC1 and EBOV-GP. B. Surface (EBOV-GP) and ribbon (NPC1) complex model. C. Upper panel, peptide sequences of designed erp1 to erp5; Lower panel, the HCDR3 loop sequences of antibodies MR78 (PDB 5QUY) and MR191(PDB 6BP2). The Phe (F*) residues in red are aligned in the middle which is correlated with F*131 (NPC1), F106 (MR78 and MR1919), F*7 (erp1 and erp1c) and F*4 (m78abc and m191) and others. D. Superimposed of NPC1 and MR78 and MR191, the Phe (F* in red) are labeled. NPC1, red, MR78, cyan and MR191, pink. The receptor binding domain (RBD) of NPC1 was used in this superimposition. RBS, receptor binding site. m78abc and m1919abc are derived from the antibody MR78 and MR191, respectively. NPC1, Niemann-Pick C1 protein; RBS, receptor-binding site.

3.2. Evaluation of peptide Inhibitors against pseudotyped viruses

Pesudotyped filoviruses have been documented to be comparable to the infectious viruses for testing inhibition of viral entry which is able to be conducted at the BSL-2 laboratories (27, 36, 37). TZM-bl cells used for HIV assay were applied for the testing since the HIV-backbone was used for making these pseudotyped viruses. The TZM-bl cells have an integrated tat driven Luciferase gene so the Luciferase activity can be easily measured by a Luminometer. The peptide samples were tested in different concentrations. The data for EBOV are shown in the Fig. 2. All IC₅₀ values were calculated and showed on their figures. It is indicated that peptides erp1 and erp5 exhibited better inhibitions with the IC₅₀ values of 21.7μM and 27.3μM, respectively. The erp1 and erp5 also showed activities, but weaker against MARV with the IC₅₀ values of 58.7μM and 69.4μM respectively (Fig. 3). The erp1c is a cyclic form of erp1 and showed increased activities against pseudotyped Ebola with a lower IC₅₀ of 12.3μM and MARV of 28.5μM (Fig. 4A and B). For MARV antibody-based peptides m78abc and m191abc, both peptides showed strong activities against MARV with the IC₅₀ values of 12.5μM and 8.5μM, respectively (Fig. 4D and F). Peptide m78abc seems potent for both MARV and EBOV (Fig. 4C and D), peptide m191abc was less potent for EBOV with the IC₅₀ of 37.2 μM (Fig. 4E). These results are suggested that there are some differences in their receptor binding-site (RBS) residues between MARV and EBOV.

Fig. 2.

Inhibition of peptide inhibitors against pesudotyped Ebola virus (EBOV). The TZM-bl cells and serial concentrations of peptides were used for the experiments and each sample tested was in triplicates. The IC₅₀ were statistically determined by the method of Plaque assay using the GraphPad Prism. Peptides: erp1 (A), erp3 (B), erp4 (C) and erp5 (D).

Fig. 3.

Inhibition of peptide inhibitors against pesudotyped Marburg virus (MARV). The TZM-bl cells and serial concentrations of peptides were used for the experiments and each sample tested was in triplicates. Peptides: erp1 (A) and erp5 (B).

Fig. 4.

Inhibition of cyclic peptides against pesudotyped Ebola virus (EBOV) and Marburg virus (MARV). The TZM-bl cells and serial concentrations of peptides were used for the tests and each sample tested was in triplicates. Cyclic peptide inhibitors: erp1c/EBOV (A), erp1c/MARV (B) m78abc/EBOV (C), m78abc/MARV (D), m191abc/EBOV (E), m191abc/MARV (F).

3.3. Evaluation of peptide Inhibitors against replication-competent EBOV

To confirm these inhibition activities from the pseudotyped virus test, a replication-competent Ebola virus was used for testing at BSL-4 containment at the National Emerging Infectious Diseases Laboratories (NEIDL), Boston University. Three peptide samples erp1, erp1c and erp5 were tested and the results were comparable to those from pseudotyped viruses (Fig. 5, Table 1). The erp1 and erp1c had the values of IC₅₀ 16.7μM and 10.54μM, respectively (Fig. 5A and B). The cyclic form erp1c showed better activity than the linear form erp1. The erp5 had an IC₅₀ of 26.6μM (Fig. 5C) which is very close to the 27.3μM of IC₅₀ from pseudotyped virus test. It was suggested that the data from the pseudotyped system is trustworthy and close to those from wild-type viruses.

Fig. 5.

Inhibition of peptide inhibitors against replication competent Ebola viruses. Hela cells were treated with the indicates amount of each peptide and performed in triplicates for erp1 peptide (A), erp1c peptide (B) and erp5 peptide (C).

Table 1.

Peptide Inhibitory activities against Ebola and Marburg viruses

Peptide	Sequence	Form	Pseudo-EBOV IC50 (μM)	EBOV IC50 (μM)	Pseudo-MARV IC50 (μM)	CC50 (μM) (TZM-bl cells)
erp1	HKKGDDFFVYADY	linear	21.7	16.7	58.7	>100.0
erp1c	HCKGDDFFVYACY	cyclic	12.3	10.5	28.5	82.6
erp3	KGDDFFPYYHS	linear	47.5	-	>80.0	>100.0
erp4	SRYYPFFGDS	linear	52.6	-	>80.0	>100.0
erp5	SQVISNPYYPFYYS	linear	27.3	26.6	69.4	>100.0
m78abc	CGTFYYYC	cyclic	16.1	-	12.5	126.3
m191abc	CSGFVEWC	cyclic	37.2	-	8.5	75.3

-, not determined.

3.4. Specificity of the peptide inhibitory activity

To examine whether the inhibitions of peptides are specific for the filoviruses Ebola and Marburg, the pseudotyped Vesicular stomatitis virus (VSV) was chosen for testing as VSV is also an enveloped, negative-sense RNA virus which can infect a wide range of mammalian cells. Five potent peptide inhibitors were tested (erp1, erp5, erp1c, m78abc and m191abc) including the linear and cyclized peptides clearly showed no activities against the pseudotyped VSV virus but having the activities against EBOV and MARV from the side-by-side assays (Fig. 6). It is suggested that these peptide inhibitors are specific for the filoviruses EBOV and MARV. It is implied that these peptide inhibitors specifically target the receptor NPC1 binding site of the viruses.

Fig. 6.

Specificity assay of peptide inhibitors against different pseudotyped viruses. EBOV, Ebola virus (blue); MARV, Marburg virus (red); VSV, Vesicular stomatitis virus (green). Peptide inhibitors: erp1 (A), erp5 (B), erp1c (C), m78abc (D), and m191abc (E).

3.5. Cytotoxicity analysis

Cytotoxicity of peptides is an important factor on evaluating the inhibition of the qualified inhibitors. The cell viability assay method (MTT) was used for evaluation of these peptides to the TZM-bl cells which were used for the inhibition assay. All the cell viability data were presented in the Figure 7A (bars) and 7B (curves). There is no significant cytotoxicity observed when the peptide sample concentrations tested up to 100μM (Fig. 7, Table 1), although the cell viabilities for some samples had some decreases at the 100μM concentration. It is suggested that the peptides are basically safe for further pre-clinical research in vivo.

Fig. 7.

Cell viability assays at different concentrations of peptide inhibitors. The TZM-bl cells were used for the testing. Each sample tested was in triplicates. PBS was used as the negative control. A. Data in bar graphs of linear peptides: erp1 blue, erp3 red; erp4 cyan, erp5 purple. B. Data in curves od cyclic peptide: erp1c, m78abc and m191abc.

3.6. Molecular docking

To analyze the interactions between these peptide inhibitors and Ebola virus GP, molecular docking was conducted. The erp1 is a linear peptide derived from the Loop 2 of NPC1 receptor, but the erp1c was the circle form of erp1. We would like to see the binding difference between the linear and circle forms since there are some inhibitory differences against the Ebola virus infection. The two key Phe (F) residues in the docking models are shown for the binding comparison along with the NPC1 receptor (Fig. 8A and B). It is obvious that there are some similar from their phenyl rings in the binding, but the circle form seems more like the NPC1. The docking energy scores for the linear and circle forms of the peptides were −170.525 (kcal/mol) and −186.045 (kcal/mol), respectively (Fig. 8C and D). For the antibody-based peptides, we analyzed the peptide m78abc and compared with the antibody MR87 heavy chain (MR78-H). Since the two neighbor residues Phe 106 (F*106) and Try 107 (Y107) of the antibody MR78 or Phe 4 (F*4) and Try 5 (Y5) of the peptide m78abc are corresponding to the two Phe (F) residues F*131 and F132 of NPC1 receptor in the pocket binding (Fig. 1), we illustrated these two key residues in the docking models (Fig. 9A). The MR78 heavy-chain (MR78-H) was docked to the EBOV-GP1 and the F*106 and Y107 of HCDR3 loop were showed in the binding pocket (Fig. 9B). The peptide m78abc was exhibited a similar binding model with the MR78-H, and the residues F*4 and Y5 were apparently presented in the receptor binding pocket with a docking energy score of −180.578 (kcal/mol) (Fig. 9C). Taken together, the docking results are suggested that the two Phe residues (F*7 and F8) in the peptides erp1 and erp1c or the two residues Phen and Try (F*4 and Y5) in the peptide m78abc are the pocked binding residues which must play a major role for the binding.

Fig. 8.

Molecular docking of antibody MR78 heavy-chain (MR78-H) and the antibody derived cyclic peptide m78abc onto the EBOV-GP1. A. The complex docking model of MR78H and EBOV-GP1. B. Close view of MR78H binding from the docking model (F*106 and Y107). C. The cyclic peptide m78abc docking model (F*4 and Y5). Only Phe (F*) and Try (Y) residues in the binding deep pockets are shown and labeled. RBS, receptor binding site.

Fig. 9.

Molecular docking of NPC1 receptor derived ligands onto the receptor binding site (RBS) of EBOV-GP1. Only the two Phe (F) residues in the deep pockets are showed and labeled. A. NPC1 and EBOV-GP1 structural complex model (from PDB 5F1B). B. Close view of the NPC1 and EBOV-GP1 binding site (F*131 and F132). C. The linear peptide erp1 binding (F*7 and F8). D. The cyclic peptide erp1c binding (F*7 and F8). NPC1, EBOV receptor; RBS, receptor-binding site.

4. Discussion

The design of peptide inhibitors in this report is based on the ligand-binding to the endosomal receptor binding site (RBS) which include the receptor NPC1 and MARV monoclonal antibodies MR78 and MR191. The results have demonstrated the endosomal receptor binding site (RBS) as a target for peptide-based inhibitors (Table 1). The two MARV antibodies of MR78 and MR191 were reported to have no activities against EBOV since the receptor binding site (RBS) has not exposed prior to enter the endosomes (16, 25, 26), but the derived peptides did show certain activities for EBOV. It is suggested that peptides may have advantages being small size which are more accessible than the antibodies to reach endosomes.

Molecular docking analysis from monoclonal antibody MR78 derived peptide inhibitor m78abc are also demonstrated this possibility. The small size peptide has the flexibility to access the deep pocket for achieving the maximum binding strength (Fig. 9). The two major binding residues Phe F* and Y showed some differences from the phenyl rings.

The cyclic peptide appears to be more effective in antiviral infection than the linear peptides from this study. The erp1c is the cyclic form of erp1 peptide. It was apparent that the cyclic form had better activity against EBOV and MARV. Furthermore, the other two cyclic peptides (m78abc and m191abc) also showed potent activities (Table 1). It is probably that cyclic peptides are more fit than the linear form for this receptor binding site (RBS). The existence of a deep cavity in the binding site for NPC1 receptor Phe131 contact may create the binding advantages for the cyclic form of peptides. In fact, it is a mimicry to the major loop contacts from those ligands including the Loop 2 of NPC1 receptor, and the heavy chain HCDR3 loops of MR78 and MR191 antibodies. The cyclic peptides based on the receptor NPC1 loop 2 were reported to have a micro molar rage IC₅₀ against pseudotyped Ebola viruses (38). To compare those peptides, we have used the peptide 3.3 as the control, it was found that our peptides had comparable activities to this peptide.

Molecular docking analysis from the NPCI loop 2 derived peptide erp1 (linear form) and erp1c (circle form) showed that the cyclopeptide showed better binding energy than the linear form (Fig. 8). The two major Phe (F) binding residues exhibited different binding postures from the linear peptide and the receptor NPC1 loop 2.

In comparison to the fusion peptide inhibitors targeting the GP2 NHR, the C-peptides based inhibitions were also reported with the IC₅₀ are in the 40μM range (39). It was found that lipids or cholesterol-conjugated C-peptides can enhance the potency (40). This is consistent with the C-peptides for inhibition HIV infection by which the T-20 based lipopeptides have shown largely increased potency as well as the half-life time (41, 42). Since the C-peptide is targeting the fusion process of 6-helix bundle which locates close to the membrane.

In summary, this research has shown the promising data for targeting the viral GP receptor binding site (RBS) by using peptide-based inhibitors in the endosomes. The IC₅₀ values are now in the lower side of micro-molar range from the ligand-based designs, it is hoped to further improve from more sophisticated structure-based designs. Furthermore, the peptide modifications will be also included for further designs to enhance the inhibition potency and stability such as adding cell membrane penetrating motif (CMT) for easily penetrate the cell membrane and enter the endosomes; or conjugated lipids or cholesterol for enriching the cell or endosomal membrane association to enhance the peptide binding to the membrane born spikes.