Cathelicidin-Derived Antimicrobial Peptides Inhibit Zika Virus Through Direct Inactivation and Interferon Pathway

Zika virus (ZIKV) is a neurotrophic flavivirus that is able to infect pregnant women and cause fetal brain abnormalities. Although there is a significant effort in identifying anti-ZIKV strategies, currently no vaccines or specific therapies are available to treat ZIKV infection. Antimicrobial peptides, which are potent host defense molecules in nearly all forms of life, have been found to be effective against several types of viruses such as HIV-1 and influenza A. However, they have not been tested in ZIKV infection. To determine whether antimicrobial peptides have anti-ZIKV effects, we used nine peptides mostly derived from human and bovine cathelicidins. Two peptides, GF-17 and BMAP-18, were found to have strong anti-ZIKV activities and little toxicity at 10 µM in an African green monkey kidney cell line. We further tested GF-17 and BMAP-18 in human fetal astrocytes, a known susceptible cell type for ZIKV, and found that GF-17 and BMAP-18 effectively inhibited ZIKV regardless of whether peptides were added before or after ZIKV infection. Interestingly, inhibition of type-I interferon signaling resulted in higher levels of ZIKV infection as measured by viral RNA production and partially reversed GF-17-mediated viral inhibition. More importantly, pretreatment with GF-17 and BMAP-18 did not affect viral attachment but reduced viral RNA early in the infection course. Direct incubation with GF-17 for 1 to 4 h specifically reduced the number of infectious Zika virions in the inoculum. In conclusion, these findings suggest that cathelicidin-derived antimicrobial peptides inhibit ZIKV through direct inactivation of the virus and via the interferon pathway. Strategies that harness antimicrobial peptides might be useful in halting ZIKV infection.

Zika virus (ZIKV) is a neurotrophic flavivirus that is able to infect pregnant women and cause fetal brain abnormalities. Although there is a significant effort in identifying anti-ZIKV strategies, currently no vaccines or specific therapies are available to treat ZIKV infection. Antimicrobial peptides, which are potent host defense molecules in nearly all forms of life, have been found to be effective against several types of viruses such as HIV-1 and influenza A. However, they have not been tested in ZIKV infection. To determine whether antimicrobial peptides have anti-ZIKV effects, we used nine peptides mostly derived from human and bovine cathelicidins. Two peptides, GF-17 and BMAP-18, were found to have strong anti-ZIKV activities and little toxicity at 10 µM in an African green monkey kidney cell line. We further tested GF-17 and BMAP-18 in human fetal astrocytes, a known susceptible cell type for ZIKV, and found that GF-17 and BMAP-18 effectively inhibited ZIKV regardless of whether peptides were added before or after ZIKV infection. Interestingly, inhibition of type-I interferon signaling resulted in higher levels of ZIKV infection as measured by viral RNA production and partially reversed GF-17-mediated viral inhibition. More importantly, pretreatment with GF-17 and BMAP-18 did not affect viral attachment but reduced viral RNA early in the infection course. Direct incubation with GF-17 for 1 to 4 h specifically reduced the number of infectious Zika virions in the inoculum. In conclusion, these findings suggest that cathelicidin-derived antimicrobial peptides inhibit ZIKV through direct inactivation of the virus and via the interferon pathway. Strategies that harness antimicrobial peptides might be useful in halting ZIKV infection.
Keywords: antimicrobial peptides, Zika virus, innate immunity, cathelicidins, plaque-forming assays inTrODUcTiOn Zika virus (ZIKV) is an enveloped neurotrophic flavivirus mainly transmitted by Aedes mosquitoes (1). Since early 2015, the rapid spread of ZIKV in South America is known to cause increased incidence of intrauterine infection that is associated with fetal microcephaly and other con genital malformations (2)(3)(4)(5)(6)(7). There is also strong evidence to link ZIKV with Guillain-Barré syndrome (GBS), which is an autoimmune disorder that cau ses demyelination in peripheral nerves (8)(9)(10)(11). In addition, meningitis and meningoencephalitis have also been descri bed in adults with history of ZIKV infection (12). Although there are much fewer cases of vectorborne ZIKV infection in the epidemic area since February of 2017, potential for new epidemics beyond currently impacted countries remains high. The epidemics of ZIKV have brought increased urgency to the development of safe and effective antiZIKV drugs. How ever, currently no vaccines or specific therapies are available for the prevention or treatment of ZIKV infection.
Antimicrobial peptides (AMPs) are a group of relatively short (usually less than 50 amino acids), cationic (net charge of + 1 to + 7), and amphipathic peptides that constitute part of innate immune molecules in nearly all forms of life (13)(14)(15). These peptides may be expressed constitutively or induced in response to infectious/inflammatory stimuli, and play an important role in eliminating invading pathogenic microorganisms, including bacteria, fungi, and viruses. Several AMPs also have selective inhi bitory effects on tumor growth (16). Although most AMPs have limited sequence and structure homologies, they are grouped into families based on their sequence and structural similarities (15). In mammals, cathelicidins and defensins are two major families of AMPs. In humans, there are also other AMP families, including lysozyme, dermcidin, and histatin (17).
Cathelicidins are generally located at the Ctermini of a 15-18 kDa precursors that share a highly conserved domain called cathelin (acronym for cathepsin L inhibitor) (18). In humans, LL37, a 37 aminoacid cationic peptide starting with a pair of leucines, is a widely studied peptide derived from the only catheli cidin gene (19). LL37 is well documented in host defense against a variety of microbial infections (20)(21)(22)(23)(24). However, the effect of LL37 has not been tested in ZIKV infection. In the current study, we hypothesize that cathelicidinderived peptides are effective in inhibiting ZIKV infection. We examined nine different pep tides mostly derived from human and bovine cathelicidins and found that several of them have antiZIKV activities in vitro. We elucidated that the mechanisms of action of these cathe licidinderived peptides are through direct inactivation and/or the activation of typeI interferon (IFN) signaling.

MaTerials anD MeThODs ethics statement
All experiments for human fetal astrocyte generation were per formed with the approval of the Scientific Research Oversight Committee at the University of Nebraska Medical Center (UNMC). Human fetal brain tissues were obtained from elective aborted specimens (gestational age 12 weeks to 16 weeks) follow ing completion of the abortion procedure through collaborative works with the Birth Defects Research laboratory at University of Washington. The protocol is in compliance with all relevant state and federal regulations and is approved by the University of Washington Institutional Review Board (IRB, Protocol no. 961826A07) and UNMC IRB (Protocol no. 12302FB). Written informed consent was obtained with all subjects using an IRBapproved consent form at the University of Washington. All consenting subjects were donors of fetal tissue that were 19 years of age or older with clear comprehension. The UNMC investigators do not have access to signed consent forms.

cells and Zika Viruses
Human fetal astrocytes were derived from a singlecell isolation process of fetal brain tissues as previously described (25,26). Briefly, dissociated brain tissue was incubated with 0.25% try psin for 30 min, followed by neutralization with 10% fetal bovine serum (FBS), and further dissociated by trituration. The singlecell suspension was cultured as adherent cells in DMEM/F12 (Thermo Fisher Scientific, Waltham, MA, USA), supplemented with 10% FBS, penicillin (50 units/mL), and streptomycin (50 µg/mL) (Thermo Fisher Scientific). This process yields a culture of >95% glial fibrillary acidic protein (GFAP, Dako Corp., Carpinteria, CA, USA) positive astrocytes in immunocytochemical staining. Vero cells (ATCC, CCL81), an African green monkey kidney cell line, were maintained in Dulbecco's Modified Eagle Medium (DMEM) with 5% FBS. Experiments with ZIKV were performed exclusively inside a BioSafety Level 2 (laboratory). All procedures utilized in this study were approved by the Institutional Biosafety Committee (IBC 1605013BL2) and followed biosafety level II practices as shown in National Institutes of health (NIH) Guideline Appendix GIIB. ZIKV strain MR766 (Uganda, 1947) was obtained from ZeptoMetrix Corp., Buffalo, NY, USA, and propagated in Vero cells. ZIKV infection of the Vero cells and fetal astrocytes was previously characterized (27) and all infections in the current studies were at the MOI of 0.5.

cell Viability assay
Cell viability was examined by the CellTiter 96 ® Aqueous One Solution Cell Proliferation Assay (Promega, Madison, WI, USA). Vero cells and astrocytes were seeded into 96well tissue culture plates (Fisher Scientific) and treated with fresh growth medium containing different concentrations of each peptide for 2 or 4 days at 37°C. At the experimental end point, MTS [3(4,5dimethyl thiazol2yl)5(3carboxymethoxyphenyl)2(4sulfophenyl) 2Htetrazolium] was added to the cultures and incubated for 1 h. The absorbance at 490 nm was recorded through BioTek Gen5 data analysis software (Winooski), VT, optical density (OD) values were used as a representation of cell viability.

ZiKV Plaque-Forming assay
Vero cells were plated into 12well plates at 5 (105 cells/well the day before infection). On the day of infection, the mon olayers of Vero cells were inoculated with 100 µL of 10fold serial dilutions of viral stocks and incubated at 37oC for 1 h. After viral inoculation, medium containing ZIKV particles was removed and 1 ml overlay containing 0.6% molecu lar bio logy grade agarose (Agarose Unlimited, Alachua, FL, USA) in Modified Eagle Medium (Gibco) with 2% FBS. Cells were maintained at 37°C in 5% CO2 for 4 days. On day 5, cells were fixed with 4% paraformaldehyde solution in PBS and stained with 1% crystal violet solution in 20% methanol in water. Viral plaques were photographed using a CanonScan 9950F scanner and each plaque was counted as a plaqueforming unit (PFU). Viral titer was calculated as PFU/[volume virus (mL) × (dilu tion factor)].

real-Time rT-Pcr
Total mRNA was isolated with TRIzol Reagent (Thermo Fisher Scientific) and RNeasy Mini Kit (QIAGEN Inc., Valencia, CA, USA) following the manufacturer's recommendations. Two realtime RTPCR methods were used for the current studies. First, SYBR Greenbased RTPCR assay was used to determine intracellular ZIKV RNA in Vero cells. For this assay, reverse transcription was performed using Verso cDNA synthesis Kit (Thermo Fisher Scientific) and 1 µg of total RNA in a 15µL final volume. The RTPCR analyses were performed using 7.5µL SYBR Green PCR Master Mix (Thermo Fisher Scientific) with 0.5µL cDNA, 1.5µL H2O, 5.5µL oligonucleotide primer pairs at 10 µM. Primers used were ZIKV RNA: forward sequence 5TGGGAGGTTTGAAGAGGCTG3, reverse sequence 5TCT CAACATGGCAGCAAGATCT3; GAPDH: forward sequence 5GGAGCGAGATCCCTCCAAAAT3, reverse sequence 5GG CTGT TGTCATACTTCTCAT GG3. PCR program: 1, 50°C for 2 min; 2, 95°C for 2 min; 3, 95°C for 15 s; 4, specific annealing temperature for 15 s; 5, 72°C for 1 min. Steps 2-4 were repeated 40 times. All samples were amplified in triplicate for analysis. Relative ZIKV RNA levels were determined and standardized with a GAPDH internal control using comparative ΔΔCT method (35). Second, TaqManbased RTPCR assay was used to determine intracellular ZIKV RNA and other gene expressions in Figures 3-6 and extracellular ZIKV RNA in Figure 3. The TaqMan assay was performed in a StepOne TM RealTime PCR system (Thermo Fisher Scientific). Primers used for TaqMan realtime RTPCR were all from the Thermo Fisher Scientific, which included ZIKV (forward sequence: 5TTGGTCATGATACTGCTGATTG C3, reverse sequence: 5 CCTTCCACAAAGTCCCTATTGC3, and probe sequence: 5′CGGCATACAGCATCAGGTGCATAGG AG3), IFNα2 (Hs00265051_s1), IFNβ1 (Hs01077958_s1), eukaryotic 18s rRNA (Hs99999901_s1), βactin (Hs99999903_ m1), and GAPDH (Catalog number: 4310884E). Relative ZIKV mRNA levels were determined and standardized with a GAPDH or 18s rRNA endogenous control using comparative ΔΔCT method (35). All reference genes, including GAPDH, βactin, and 18s rRNA, detected identical changes of genes ( Figure S1 in Supplementary Material). Therefore, typically one was chosen as the endogenous control for gene expressions in the current studies. For ZIKV in supernatants, RNA was extracted from the supernatants through TRIzol™ LS Reagent (Thermo Fisher Scientific, 10296028) following the manufacturer's manual. Since reference genes were not expressed in the supernatant RNA, an equal volume of extracted RNA was used in realtime RTPCR. Quantitative genomic RNA from ZIKV (NR1838DQ, ATCC) was used as a standard to calculate viral copies. All primers used in the study were tested for amplification efficiencies and the results were similar to each other.

Viral attachment assay
Viral attachment assay was adapted from a previous publica tion (36) with modifications. Briefly, primary human fetal astrocytes were preincubated at 4°C for half an hour and then treated with AMPs along with ZIKV infection at the MOI of 0.5. After infection at 4°C for 2 h, cultures were washed with fresh medium for three times and RNA was isolated from whole cells. ZIKV RNA was determined through realtime RTPCR. Data were normalized to 18s rRNA and presented as fold change compared with the ZIKV group. One experimental group was subjected to trypsin digestion to remove any attached virions. This group served as a positive control for the viral attachment experiment.

Direct inactivation of ZiKV by aMPs
Zika virus viral stocks were diluted to 10 6 PFU/mL with serumfree medium and peptides were added at the indicated final concentrations. The mixtures were incubated at 37°C for 1-4 h. At each time point, 200 µL of the mixtures was removed for virus yield determination. Tenfold serial dilutions of the mixtures were prepared and number of infectious Zika virions in the inoculum was determined by the aforementioned plaque forming assay on Vero cell monolayers. . Amino-acid sequences of the AMPs were compared with the Vector NTI software suite (Thermo Fisher Scientific). (a) Working relationships between the nine AMPs derived from human and bovine cathelicidins are shown in the dendrogram based on the amino-acid sequence homology. The length of each tree branch is proportional to the number of amino-acid differences between a particular sequence and its theoretical ancestral sequence (presented as parentheses for each sequence). (B) Amino-acid sequence alignment. Color codes indicate degree of similarity: yellow, font red: identical; cyan, font blue: conservative; green, font black: similar; no color, font green: weakly similar, no color, font black: non-similar.

statistical analysis
Statistical analyses were performed using GraphPad Prism 7.00 and IBM SPSS Statistics Version 22. The data were presented as means ± SD unless specified otherwise. The EC50 values of cytotoxicity data were calculated using GraphPad Prism 7.00. Differences between groups were compared using the oneway ANOVA with Bonferroni posttest for multiple comparison. For the data that are not normally distributed due to varying levels of ZIKV inhibition, we applied the log transformation to meet normality of data distribution before statistical tests. p < 0.05 was considered as significant. All assays were performed at least three times in triplicate or quadruplicate.

resUlTs aMPs inhibit ZiKV infection in Vero cells
To determine whether AMPs have antiZIKV effects in vitro, we selected nine peptides, including eight active AMPs and one nonantimicrobial control peptide--RI10. Most of these peptides are closely related to human cathelicidin LL37. LL37 is aligned with GI20 (32), GF17 (28), and RI10 (29), which corresponds to residues 13-32, 17-32, and 19-28 of LL37, respectively, while another two peptides, 17BIPHE2 and merecidin, are designed based on GF17 to gain stability to proteases such as chymotrypsin (Figures 1A,B). In contrast, bovine cathelicidinderived BMAP18 and databaseobtained DASamP2 are more distantly related to LL37 compared with those aforementioned AMPs (30,32). BMAP18 and GF17 share a similar amphipathic helical pattern, where four residues are identical and seven residues are semiconserved (Figure 1).
We first evaluated the cytotoxicity of Vero cells after treatment of AMPs through a cell viability MTS assay. Vero cells were exposed to each of the nine peptides with doses ranging from 0 to 50 µM for 48 h. Among the AMPs, five showed cell viability of more than 50% compared with untreated controls (peptide con centrations at 0 µM) at the highest dose tested (EC50 > 50 µM). These five AMPs include RI10, DASamP2, GF17, BMAP18, and GI20 (Figure 2A). In contrast, LL37, GI20Dform, 17BPIHE2, and merecidin exhibited varying levels of cytotoxic effects with EC50 ranging from 4 to 20 µM. To test the antiZIKV effects of the AMPs, we chose the doses of 0.4, 2, and 10 µM on the basis of cytotoxicity and early work that showed their effec tive treatment concentrations (32). ZIKV infection was tested at 48h postinfection because our earlier work showed that infection causes significant cytotoxicity beyond this time point (27). All AMPs except RI10 showed dosedependent decrease of ZIKV RNA compared with untreated controls (Figure 2B). Specifically, Merecidin, LL37, GI20, and GI20 Dform sig nificantly decreased ZIKV RNA at 0.4 µM. DASamP2, GF17, LL37, BMAP18, GI20, and GI20 Dform significantly decreased ZIKV RNA at 2 µM. All peptides except RI10 significantly decreased ZIKV RNA at 10 µM. Taken together, these data demonstrate that eight out of the nine selected AMPs inhibit ZIKV infection in Vero cells.

cathelicidin-Derived aMPs inhibiting ZiKV infection in Primary human Fetal astrocytes
Since significant cytotoxicity could interfere with the interpreta tion of the antiZIKV activity by AMPs. We treated Vero cells for 2 days ( Figure S2A in Supplementary Material) and 4 days ( Figure  S2B in Supplementary Material) with the 9 AMPs at 10 µM and evaluated the cytotoxicity through the MTS assay. Consistent with Figure 2A, GF17, RI10, and BMAP18 caused minimal cytotoxicity to the Vero cells. Based on these cytotoxicity data, we excluded 17BIPHE2, Merecidin, LL37, GI20, and GI20D form from further testing. GF17 and BMAP18 both belong to the family of cathelicidin. Therefore, we decided to investigate how cathelicidinderived GF17 and BMAP18 impact ZIKV infection. The inactive peptide RI10 was included as a negative control. Since human fetal astrocytes are a natural host of ZIKV, we tested the effects of GF17 and BMAP18 treatment in this cell host. Cytotoxic effects of BMAP18, GF17, and RI10 on human astrocytes were evaluated and no obvious cytotoxicity was observed over 2 days (EC50 > 50 µM, Figure 3A; Figure  S3A in Supplementary Material) and 4 days ( Figure S3B in Supplementary Material) of exposure to them. Therefore, consist ent with the data in Vero cells, BMAP18, GF17, and RI10 are not cytotoxic to human fetal astrocytes. Next, we tested antiviral effects of RI10, BMAP18, and GF17 in ZIKVinfected human fetal astrocytes that we previously char acterized (27). Pretreatments with GF17 and BMAP18, but not RI10, decreased intracellular ZIKV RNA in a dosedependent manner (Figure 3B). Both GF17 and BMAP18 showed most significant inhibitory effect (p < 0.001) at the concentration of 10 µM. At 2 µM, GF17 and BMAP18 reduced intracellular ZIKV  RNA to 46 and 15% of the untreated ZIKV group, respectively. At 10 µM, GF17 and BMAP18 reduced intracellular ZIKV RNA to 1 and 4% of the untreated ZIKV group, respectively. The inhibitory effects of GF17 and BMAP18 were also demon strated on extracellular ZIKV RNA in the culture supernatants. For BMAP18, the level of ZIKV RNA was reduced to 5.6% at 2 µM (p < 0.01) and 1.3% at 10 µM (p < 0.01). For GF17, the ZIKV yield was reduced to 16.7% at 2 µM (p < 0.05) and 0.1% at 10 µM (p < 0.01) ( Figure 3C). Consistent with the ZIKV RNA data, ZIKV virions were reduced by GF17 and BMAP18, but not RI10, in a dosedependent manner (Figures 3D,E). Taken together, these data demonstrate that GF17 and BMAP18 inhibit ZIKV infection in primary human fetal astrocytes.
To test the therapeutic potential of cathelicidinderived AMPs against ZIKV, we changed the treatment of AMPs from preinfection treatment to 24h postinfection treatment. ZIKV RNA was determined at 24 h after AMP treatment ( Figure 4A). PostZIKV treatment with GF17 for 24 h significantly decreased ZIKV RNA compared with the untreated ZIKV group [p < 0.001; Figure 4B)]. ZIKV RNA in BMAP18treated group also trended downward; however, the difference was not significant (p = 0.063). Also, at 24 h after AMP treatment, Zika virions were quantified through plaqueforming assay; fewer virions were found in BMAP18 and GF17treated groups, compared with those of untreated ZIKV group (Figures 4C,D), suggesting that similar to preinfection treatment, postinfection treatment with BMAP18 or GF17 also inhibits ZIKV infection.

aMPs inhibiting ZiKV at 2-h Post-infection
To determine the mechanism(s) of AMPmediated ZIKV inhibi tion, we first investigated the ZIKV attachment and viral entry. For viral attachment assay, astrocytes were cultured at 4°C for half an hour before the addition of ZIKV and peptides, and the total RNA was harvested after 2 h at 4°C. An extra group of astrocytes was trypsinized to remove any attached virions. This group serves as a positive control for the assay. BMAP18, GF17, and RI10 treatment did not alter the ZIKV RNA levels in astrocytes at 4°C (Figure 5A), suggesting that these AMPs do not affect ZIKV attachment. To determine whether AMPs affect ZIKV entry to the astrocytes, we treated astrocytes with AMPs for 2 h at 37°C before ZIKV infection and collected total RNA at 2h postinfection ( Figure 5B). GF17 and BMAP18, but not RI10, significantly decreased ZIKV RNA (p < 0.01, Figure 5B) compared with those of untreated ZIKV group. Taken together, these data indicate that, although GF17 and BMAP18 do not affect viral attachment, they impact ZIKV entry at 2h postinfection.

interferon signaling is associated With ZiKV infection and aMP effects
To further explore the mechanism of AMPmediated ZIKV inhi bition, the levels of IFNα2 and IFNβ1 during individual AMP treatment with or without ZIKV infection, were determined through realtime RTPCR (Figures 6A-D). In uninfected fetal astrocytes, control peptide RI10 did not affect the expression levels of IFNα2 ( Figure 6A). In contrast, GF17 treatment significantly increased IFNα2 expression (p < 0.0001). Similarly, BMAP18 also increased IFNα2 by 7 folds though the difference was not statistically significant ( Figure 6A). In ZIKVinfected fetal astrocytes, both GF17 and BMAP18, but not RI10, significantly increased IFNα2 expression levels in a dosedependent manner ( Figure 6A). Furthermore, the level of IFNα2 is negatively correlated with ZIKV RNA level in infected cells ( Figure 6B). Together, these data suggest that AMPs induce the IFNα2 expression in fetal astrocytes and the AMPinduction of IFNα2 likely has a negative impact on ZIKV infection in the cultures. The analysis of IFNβ1 reveals that in uninfected fetal astrocytes, AMP treatment did not affect IFN1 expression levels. Surprisingly, in ZIKVinfected fetal astrocytes, AMP treatment significantly decreased IFNβ1 expression levels ( Figure 6C) and the levels of IFNβ1 were positively correlated  with ZIKV RNA level (p < 0.01) ( Figure 6D). Therefore, these data suggest that IFNβ1 is likely more of a reflection on the ZIKV infection levels and does not play a role in AMPmediated ZIKV inhibition.
To determine the role of IFNs during AMPmediated ZIKV inhibition, we first treated astrocytes with recombinant human IFNα before ZIKV infection. As expected, IFNα treatment significantly inhibited ZIKV infection ( Figure 6E). Next, we used a neutralizing antibody for IFNα and a small molecule fludarabine known to be an inhibitor for IFN signaling (37). The neutralizing antibody for IFNα neutralizes multiple subtypes of human IFNα, including IFNα2, IFNα8, and IFNα21. Inhibition of typeI IFN signaling through neutralizing IFNα antibody ( Figure 6E) or fludarabine ( Figure 6F) dampened but did not completely block the GF17 and BMAP18mediated ZIKV inhibition, suggesting that GF17 and BMAP18 induced IFN signaling is limiting ZIKV infection in the fetal astrocytes. The modest effect of neutralizing IFNα antibody also indicates that typeI IFNindependent ZIKV inhi bition pathway exists.  To determine whether GF17 and BMAP18 have a direct effect on ZIKV viral particles, plaque assay was performed on ZIKV incubated with AMPs. GF17, BMAP18, and RI10 (10 µM) was individually mixed with ZIKV and incubated at 37°C for 1-4 h before plaque assay was performed ( Figure 7A). These treatment time points were chosen based on a prior report on direct inactivation of vaccinia virions by LL37 after 2h incubation (38). After quantification of ZIKV plaques, we found that the effects of GF17 to inhibit infectious ZIKV virions in the inoculum were both dose and timedependent. GF17 inactivated Zika virions by >95% after 1h incubation and the effect increased to >99% after 2h incubation and 99% after 4h incubation ( Figure 7B). Direct incubation of GF17 and ZIKV at 10 µM and 4 h caused the most virions reduction

DiscUssiOn
The ZIKV epidemic in South America in 2015 and 2016 is known to cause severe congenital central nervous system (CNS) malformations such as microcephaly, cerebral atrophy, ven triculomegaly, ocular anomalies, and visual/auditory impair ments (39). ZIKV infection is also linked to adult GBS and  incidence of laboratory confirmation of a ZIKV infection continues to increase among GBS cases (11,40,41). Although several therapeutic approaches have been suggested to be effective against ZIKV, including FDAapproved drugs, typeI IFNs, and liveattenuated vaccine (42)(43)(44)(45)(46), currently no vac cination or treatment is available for human use, especially for pregnant women and fetuses (47). In the present study, we demonstrated for the first time that AMPs derived from human and bovine cathelicidins have potent antiviral activity against ZIKV. Humans have only one cathelicidin, which is naturally cleaved to form LL37. We have further identified that GF17, which corresponds to the major antimicrobial region (residues 17-32) of LL37 (28,48), remains potent against ZIKV and has less cytotoxicity compared with LL37. Inhibition of IFN sign aling partially reversed GF17mediated viral inhibition. More importantly, direct incubation with GF17 specifically reduced the number of infectious Zika virions. Therefore, cathelicidin derived GF17 potently inhibits ZIKV through direct inacti vation as well as via the IFN pathway. Antimicrobial peptides are able to stimulate antiviral effects through multiple mechanism of action. For example, AMPs with similar secondary structures may have diverging effects on viral replication (49)(50)(51). Furthermore, the same AMPs may exert antiviral activity on different viruses through distinct mechanisms (52,53). Human AMP LL37 was previously shown to have antiviral activity against HIV1, influenza A virus, Respiratory Syncytial Virus, and Vaccinia Virus (38,(54)(55)(56)(57). However, LL37 is limited for therapeutic use because of its length of 37 residues, making it costly to chemically synthesize. In addition, pathogens have evolved to compromise the efficacy of LL37. For instance, pandemic influenza A virus H1N1 2009 (Cal09) is known to resist to LL37 but not a fragment of LL37 (54). In the current study, we mainly focused on two cathelicidinderived peptides: GF17 and BMAP18 because of their higher specificity to ZIKV. Our data demonstrate that both GF17 and BMAP18 have strong antiviral properties against ZIKV in vitro and the antiviral activities can be achieved at low micromolar concentrations. Bovine cathelicidins BMAP 18 was used to show the specificity and uniqueness of human cathelicidins. Our comparison of human and bovine cathelici dins reveals that human cathelicidin might have achieved an evolutional significance to protect humans from viral infection because it works by directly inactivating virus, whereas the bovine cathelicidinderived peptide is unable to work.
Inhibition of IFN signaling pathway partially blocks the anti ZIKV effect of BMAP18 and GF17. This is consistent with the data that suggest IFN signaling as an important factor regulating ZIKV susceptibility (58,59). It is unclear how ZIKV continues to replicate despite the apparent induction of typeI IFN by AMPs in the astrocytes. ZIKV may use NS5, a nonstructural protein of ZIKV, to antagonize host IFN response by preventing JAKSTAT signaling (60).
One possible mechanism of action by GF17 is to damage pathogen membrane envelope and interrupt the viral membrane integrity, as previously reported as a carpetbased mechanism (38,61). Analysis of the sequences of LL37, GF17, GI20, and RI10 together with their antiZIKV activity suggests that resi dues 1-12 and 33-37 of LL37 are not essential for the antiZIKV effect. In contrast, residues 18 and 29-32 appear to be critical in that antiviral effect are lost if these residues are not present. In addition, residues 13-16 appear to be important in that antiviral effect are modestly weakened if these residues are removed. Analysis of the sequence of 17BIPHE2, Merecidin, and their antiviral activity also suggests that residue R23 is critical since its substitution significantly reduces the antiviral effect. This is consistent with our previous observation that substitution of lysines with arginines increased antiHIV effects (62). Therefore, these data suggest that residues 17-32 of LL37 are critical for the optimal antiZIKV activities of LL37.
Few drug candidates are available to safely and effectively reduce viral load and prevent the development of disease after infection (42,44,45,(63)(64)(65)(66)(67)(68)(69)(70)(71)(72). Compared with the reported drug candidates, AMPs have multiple advantages. First, treatment with GF17 and BMAP18 may be a safer choice compared with other drug candidates since the LL37 is already present in the human body as a component of innate immune system. Second, the antiviral effect of AMPs is rapid, potent, and concentration dependent. At 10 µM, the reduction of ZIKV RNA by GF17 and BMAP18 starts at 2h postpeptide treatment and ZIKV RNA levels are reduced by over 99% and 95% at 48 h in astrocytes, respectively. Third, the AMPs are effective when applied either before or after ZIKV infection, which indicates the potential use of the peptides as an antiZIKV drug both prophylactically and therapeutically. An additional advantage of AMP is that they function through multiple mechanisms. In the event when IFN signaling is compromised by the virus, peptides are still expected to work against the virus via a direct inactivation mechanism. However, it is unclear whether GF17 and BMAP18 can cross placental and blood-brain barrier. Future studies are needed to evaluate the efficacy of GF17 and BMAP18 in vivo and how to deliver the peptides to the CNS. cOnclUsiOn Cathelicidinderived AMPs potently inhibit ZIKV infection in Vero cells and human fetal astrocytes. GF17 mediates anti ZIKV effect through both a direct inactivation of viral particle and partially through typeI IFN signaling, whereas BMAP18 inhibits ZIKV through typeI IFN signaling. Strategies based on the peptides documented herein might be useful in halting ZIKV infection both prophylactically and therapeutically.

eThics sTaTeMenT
All experiments for human fetal astrocyte generation were performed with the approval of the Scientific Research Over sight Committee at the University of Nebraska Medical Center (UNMC). Human fetal brain tissues were obtained from elective aborted specimens (gestational age 12 weeks to 16 weeks) follow ing completion of the abortion procedure through collaborative works with the Birth Defects Research laboratory at University of Washington. The protocol is in compliance with all relevant state and federal regulations and is approved by the University of Washington Institutional Review Board (IRB, Protocol no. 961826A07) and UNMC IRB (Protocol no. 12302FB). Written informed consent was obtained with all subjects using an IRBapproved consent form at the University of Washington. All consenting subjects were donors of fetal tissue that were 19 years of age or older with clear comprehension. The UNMC investigators do not have access to signed consent forms. aUThOr cOnTriBUTiOns MH, YH, GW, and JZ conceived and designed the experiments. MH, HZ, YL, and JZ performed the experiments. MH, HZ, YH, GW, JHZ, and JZ analyzed the data. GW and BT contributed reagents/materials/analysis tools. MH, YH, GW, and JZ wrote the paper. All authors read and approved the final manuscript.