CORD-19:500c5ea0961b573ba5355d364ca94c00a0a26e4a JSONTXT

The dengue virus non-structural protein 1 (NS1) is secreted from mosquito 2 cells in association with the intracellular cholesterol transporter chaperone Abstract 11 12 Keywords: 13 dengue virus; zika virus; yellow fever virus; flavivirus; NS1; caveolin-1; chaperone caveolin 14 complex; viral protein trafficking; unconventional secretion; mosquito cells 15 16 17 ABSTRACT 18 Dengue virus (DENV) is a mosquito-borne virus of the family Flaviviridae. The RNA viral genome 19 encodes for a polyprotein that is co-translationally processed into three structural proteins and . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 2 20 seven non-structural proteins. The non-structural protein 1 (NS1) is a multifunctional viral protein 21 actively secreted in vertebrate and mosquito cells during DENV infection. In mosquito cells, NS1 22 is secreted in a caveolin-1 (CAV-1) dependent manner by an unconventional pathway. The 23 caveolin chaperone complex (CCC) is a cytoplasmic complex formed by caveolin-1 and the 24 chaperones FKBP52, Cy40 and CyA which is responsible for cholesterol traffic inside the cell. 25 In this work, we demonstrate that in infected mosquito cells, DENV NS1 is secreted by an early 26 and unconventional route that bypasses the Golgi apparatus in close association with the CCC. 27 Treatment of mosquito cells with classic secretion inhibitors such as brefeldin A, golgicide A and 28 Fli-06 showed no effect on NS1 secretion, but significant reductions in recombinant luciferase 29 secretion and virion release. Silencing the expression of CAV1, FKBP52 with siRNAs or the 30 inhibition of CyA by cyclosporine A resulted in significant decrease in NS1 secretion without 31 affecting virion release. Using co-localization, co-inmunoprecipitation and proximity ligation 32 assays, NS1 was found to co-localize and interact with all the protein components of the CCC 33 in mosquito infected cells. In addition, CAV-1 and FKBP52 expression was found augmented in 34 DENV infected cells. Finally, the treatment of ZIKV infected mosquito cells with brefeldin A and 35 golgicide A showed no effect on NS1 secretion, while affecting virion release. ZIKV NS1 was 36 also found to co-localize with CAV-1 in infected mosquito cells. These results suggest that in 37 mosquito cells, ZIKV NS1 follows the same secretory pathway observed for DENV NS1. The 38 association of NS1 with the cholesterol transporter CCC agrees with the lipoprotein nature of 39 secreted hexameric NS1. 40 41 AUTHOR SUMMARY . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 3 42 Dengue protein NS1 is secreted in infected mosquito and vertebrate cells. In humans, secreted 43 NS1 have been associated with pathogenesis. In mosquito cells, NS1 follows an unconventional 44 secretion pathway that is dependent on Caveolin-1. This work shows that in mosquito cells, NS1 45 secretion is associated to the chaperone caveolin complex , a complex formed by caveolin-1 and 46 several chaperones, in charge of cholesterol transport within the cells. Reduction of the 47 expression or the activity of chaperone caveolin complex in mosquito infected cells, diminished 48 the secretion of NS1 without affecting virion release. Direct interaction between NS1 and the 49 chaperone caveolin complex proteins was demonstrated by several assays. Moreover, 50 increased expression of the caveolin-1 and co-chaperone FKBP52 during dengue infection was 51 found, presumably in response to the higher requirements of these proteins during dengue virus 52 infection. Results obtained with ZIKV infected mosquito cells suggest that also ZIKV NS1 is 53 released following an unconventional secretory route in association with the chaperone caveolin 54 complex. The functions of secreted NS1 within mosquito are unclear. However, giving the 55 importance of the soluble NS1 in the vertebrate host, manipulation of the NS1 secretory route 56 may prove a valuable strategy for dengue mosquito control and patient treatment. 57 58 59 4 64 infection and its public health burden are hard to estimate, but at least 100 countries are currently 65 endemic for dengue [2]. DENV has a single stranded RNA genome of positive polarity. The 66 genome of DENV is translated into a single polyprotein and encodes for three structural (C, E, 67 prM/M) and seven non-structural (NS) (NS1, NS2A, NS2B, NS3, NS4A, NS4B, NS5) proteins. 68 The polyprotein is then cleaved by host and viral proteases to release individual viral proteins 69 [3]. The viral genome replication process within the host cell is mainly driven by the NS proteins. 70 The NS1 protein act as an scaffolding protein that anchors the replication complex to the ER 71 membrane and interacts physically with NS4B [4]. The NS1 protein is a 352-amino-acid 72 polypeptide with a molecular weight of 46-55 kDa, depending on its glycosylation status. The 73 NS1 protein exists in multiple oligomeric forms and is found in different cellular locations: a cell 74 membrane-bound form in association with virus-induced intracellular vesicular compartments, 75 on the cell surface and as a soluble secreted hexameric lipoparticle [4]. The NS1 monomeric 76 form rapidly dimerizes in the endoplasmic reticulum (ER), then three dimeric forms of NS1 77 arrange to form an hexamer [5]. The hexameric form of NS1 shows an open barrel form filled 78 with lipids and cholesterol which resemble the lipid composition of the HDL particle [6]. 79 Recent studies have shown that the DENV NS1 protein was not only secreted from vertebrate 80 cells, but that was also efficiently secreted from mosquito cells lines [7,8]. The secretion of NS1 81 in vertebrate cells follows the classical Golgi-pathway [9]. However, NS1 secretion in infected 82 mosquito cells is associated to a caveolin-1 (CAV-1) dependent pathway and was found to be 83 Brefeldin A (BFA)-insensitive, suggesting a traffic route that bypasses the Golgi-complex [10]. 84 Caveolae are made up of interlocking heteropolymers of a family of small proteins (caveolins: 1-85 3) and a second family of accessory structural proteins (flotillins and three family of cavins). The 86 caveolar architecture is connected with unstructured cavin filaments by coiled-coil domains into . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 5 87 a polygonal net-like complex. This complex is believed to provide scaffolding for compartmented 88 cellular processes and participates in multiple cellular functions, including endocytosis, 89 transcytosis, membrane homeostasis, inflammation, and signal transduction [11]. Caveolin 1 90 (Cav-1), a 21-24-kDa scaffolding protein, is not only a key structural component of the caveolae 91 organelle but also plays an important in the transport of free-cholesterol inside the cell [12,13]. 92 The chaperone caveolin complex (CCC) is a cytosolic complex reported to transport cholesterol 93 synthetized de novo from the ER to cell membranes or other compartments within the cell. CCC 94 has been described as a complex of CAV-1, Cyclophilin A (CyA), FK506-binding protein 4 or 95 heat shock protein 56 (FKBP52), and Cyclophilin 40 or D (Cy40) [12,14]. Cyclophilin A (CyA), 96 an 18 kDa peptidylprolyl cis-trans isomerase, is a ubiquitous and multifunctional protein. In 97 addition to its role as a host-cell receptor for cyclosporine A, CyA has diverse functions in 98 inflammatory conditions and diseases [15,16]. The 52-kDa FK506-binding protein (FKBP52) an 99 immunophilin belonging to the FKBP family, is a known co-chaperone of the heat shock protein 100 90 (HSP90) and thus may play a role in the intracellular trafficking of hetero-oligomeric forms of 101 the steroid hormone receptors [17,18]. Cyclophilin 40 (Cy40), a member of a family of highly 102 homologous peptidylprolyl cis-trans isomerases (PPIases), is known to play a role in 103 mitochondrial permeability transition (MPT), being an integral constituent of the MPT pore [19]. 104 Given the CAV-1 dependent secretion of NS1 protein in mosquito cells and the lipoprotein nature 105 of the released hexameric form of NS1, it was found plausible to study the association of NS1 106 traffic to the novo cholesterol transport in DENV infected mosquito cells. In this work, data is 107 presented indicating that in infected mosquito cells, DENV NS1 enters the unconventional 108 secretory pathway very early after maturation in the ER and usurps the cholesterol transport 109 between the ER and the plasma membrane, mediated by the CCC, to reach the extracellular . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 6 110 space. In addition, data is presented suggesting that a similar pathway is used for the secretion 111 of Zika Virus NS1 protein in infected mosquito cells. 112 113 RESULTS 114 NS1 secretion is not affected by drugs that disrupt early steps of the classical secretion 115 pathway 116 Golgicide A (GCA) is a powerful inhibitor of the COPI vehicle transport from ER to Golgi [20]. 117 Thus, the cytotoxicity of GCA in the mosquito cell lines (C6/36, Aag2) and the vertebrate cell line 118 BHK-21, used for comparisons, was measured using the reduction of tetrazolium salts to 119 examine proliferation in cells treated with serial dilutions of GCA. No significant cytotoxicity was 120 observed under 30 M of GCA in any of the three cell types (Fig 1A) . Fli-06 is a novel drug which 121 inhibits the diffusion of ER synthetized proteins to the ER-exit sites [21,22]. Fli-06 cytotoxicity 122 was determined in C6/36 and Aag2 cells also using also tetrazolium salt reduction. No significant 123 cytotoxicity was observed under 100 M of Fli-06 in the mosquito cells lines (Fig 1B) . Thus, 124 concentration of 27 M and 100 M were used for GCA and Fli-06, respectively, since these 125 concentrations proved non-toxic, yet effective in causing Golgi disruption (Fig 1C) . Figures 1D 126 and 1E show that the GCA-treatment of DENV infected C6/36 and Aag2 cells did not cause any 127 change in NS1 secretion. In contrast, NS1 secretion was reduced over 70% in infected BHK-21 128 cell treated either with GCA or BFA ( Fig 1F) . As expected, virion release was reduced in 129 mosquito and vertebrate cells after GCA or BFA treatment (Fig 1G, 1H and 1I ). To further explore 130 the traffic route of NS1 in DENV infected mosquito cells, cells were also treated with Fli-06. 131 Treatment of infected C6/36 or Aag2 cells with non-toxic concentrations of Fli-06 shows no effect . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 7 132 in NS1 secretion, while significantly affecting virion release (Fig 1G, 1H) . Experiments with Fli-133 06 were not carried out in BHK-21 due to high drug toxicity. The results with GCA and Fli-06 134 indicate NS1 is not secreted by a classical secretory pathway and that newly synthetized DENV 135 NS1 do not reach the ER-exit sites and leaves the ER compartment very early after synthesis. 136 Since the release of mature virions, which are multimolecular complexes, may not be equivalent 137 to the release of a single protein, an additional control for protein trafficking through the classical 138 secretory pathway was included. To this aim, a vector, pAc5-mCherry-GLuc-neo, constitutively 139 expressing Gaussia Luciferase in C6/36 cell was constructed (S1 Fig). To evaluate the effect of 140 GCA treatment on GLuc secretion, cells were treated with GCA for 24 h and the activity of GLuc 141 present in the cell supernatants analyzed using a luminescence assay. A decrease in luciferase 142 activity, in relation to the control (DMSO), was detected upon treatment of C6/36 with GCA, and 143 in BHK-21 cells transfected with ptk-GLuc vector, run in parallel as control (Fig.1J). These results 144 indicated that the luciferase reporter is secreted in both mosquito and vertebrate cells following 145 a classical secretory route and reinforce the results indicating that in mosquito cells DENV NS1 146 is secreted by an unconventional secretory route that bypasses the Golgi (Fig 1K). 147 148 NS1 secretion is dependent on the CCC 149 The CCC is formed by the association of CAV-1 with the chaperones CyA, FKBP52, Cy40 150 [12,13]. The CCC is responsible for cholesterol transport inside the cell. Previous data by us [10] 151 indicated that NS1 secretion was dependent on CAV-1. Thus, to fully evaluate the participation 152 of the CCC in DENV NS1 secretion from mosquito cells, CyA was pharmacologically inhibited 153 by treatment with Cyclosporin A (CsA) and the expression of the chaperones FKBP52 and Cy40 . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 8 154 knocked down using siRNAs. Previously, the non-cytotoxic concentration for CsA in mosquito 155 and vertebrate cells were determined (Fig 2A) . DENV2 or DENV4-infected C6/36 or Aag2 cells 156 were treated with 9 M CsA for 24 hours; NS1 secretion was reduced in about 30% and 40% in 157 C6/36 and Aag2 cells, respectively (Fig 2B, 2E) . However, virion release in the treated mosquito 158 cells was not significantly affected (Fig 2C and 2F ). In contrast, the inhibition of CyA in BHK-21 159 cells did not have any effect in either NS1 or virion secretion (Figs 2H and 2I). To further 160 corroborate the participation of CyA in cholesterol and lipid traffic, the lipid droplet count in CsA 161 treated versus untreated cells was determined. In all cases, C6/36, Aag2 and BHK-21 cells 162 treated with CsA showed a diminished amount of lipid droplets per cell as compared with un-163 treated cells (Fig 2D, 2G and 2J). 164 The expression of FKBP52 in C6/36 cells was knock down by siRNA transfection. Optimal gene 165 knock down was achieved after 48h post transfection (Fig 3A) . The effect of FKPB52 reduced 166 expression in cholesterol and lipid metabolism was evaluated measuring the lipid droplet 167 counting per cell and a reduction of about 50% in the number of lipid droplets was observed after 168 the treatment (Fig 3B) . In cells treated with siRNA FKBP52, a significant reduction in NS1 of 169 about 20% secretion was observed ( Fig 3C) . Viral release was not affected by FKBP52-knock 170 down (Fig 3D) . Given that FKBP52 is a co-chaperone of HSP90 [17,23], we wanted to discard if 171 the observed inhibitory effect on NS1 secretion might be related to a reduction in HSP90 activity. 172 Thus, the effect of the HSP90 inhibitor geldanamycin (GA) in NS1 release from DENV infected 173 C6/36 cells was tested. As shown in Figure 3E , NS1 secretion was not affected after GA 174 treatment. The expected reduction in virus yield observed in GA treated cell (Fig 3F) show that 175 effective GA doses were used. These results indicate that the effect in NS1 secretion observed 176 after FKBP52 silencing is not related with HSP90 activity and was rather associated with the The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 9 177 function of the CCC. Finally, the expression of Cy40 was significantly knock down at 48h post 178 siRNA transfection (Fig 4A) . A reduction in about 60% of lipid droplets counting per cell was 179 observed in transfected cells ( Fig 4B) . Yet, neither NS1 secretion nor virus release was affected 180 by Cy40 silencing (Fig 4C and 4D ). This result suggests that while Cy40 participates in 181 cholesterol traffic it may be dispensable for NS1 secretion. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 13 266 were detected in BHK-21 cells with ZIKV and YFV (data not shown). The secretion of YFV NS1 267 was found to be sensitive to BFA treatment in both mosquito and vertebrate cells (Fig. 9A) . In 268 contrast, the secretion of ZIKV NS1 was found to be insensitive to BFA and GCA treatment in 269 infected mosquito, but not in infected vertebrate cells (Fig 9B) . In addition, the secretion of ZIKV 270 NS1 was reduced in mosquito, but not in vertebrate cells, treated with the CyA inhibitor CsA (Fig 271 9C ). Co-localizations experiments carried out in C6/36, Aag2 and Vero-E6 cells infected with 272 ZIKV and YFV, showed a stronger, and a significantly different, co-localization between ZIKV 273 NS1 and CAV-1 than between YFV NS1 and CAV-1, in both mosquito cell lines. No co-274 localization of ZIKV or YFV NS1 with CAV-1 was observed in vertebrate cell line ( Fig 9D and 275 9E). Of note, the co-localization data support the predictions derived from the presence of a 276 conserved caveolin binding (CBD) domain in ZIKV (and DENV) but not in YFV NS1 (Fig 9F) . The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 14 289 10.2210/pdb2a65/pdb) with a p-value 2.69e-03. Molecular docking predicts a favorable 290 interaction between the NS1 hydrophobic domain including the grease fingers, where the CBD 291 is located, with a pocket in the CAV-1 scaffolding domain (CSD) with distances less 8Ǻ ( Fig 9A 292 and 9B). The molecular docking results obtained with CAV-1 and NS1 were used as a control to 293 carry on additional docking experiments between the DENV NS1 or CAV-1 and the CCC 294 chaperones FKBP52, Cy40 and CyA. Calculations of the energy binding of several docking 295 clusters showed that the DENV NS1 has the lowest and more favorable energy for CAV-1, 296 compared to the energy binding obtained for FKBP52, Cy40 or CyA. In addition, lower energies 297 were obtained for the binding of CAV-1 with the other chaperones of the CCC, than for the 298 binding of NS1 with those same chaperones ( Fig 9C) . Based in the free energies data, we predict 299 a model of interaction of NS1 with the CCC proteins in which FKBP52, Cy40 and CyA are directly 300 bound to CAV-1, and NS1 is bound to the CCC through the CAV-1 scaffolding domain ( Fig 9D) . The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 15 303 DISCUSSION 304 Our previous work demonstrated that DENV NS1 secretion in mosquito cell lines follows a BFA-305 insensitivity secretion pathway that depends on CAV-1; moreover, NS1 was found to interact 306 directly with CAV-1 [10]. Therefore, in the present work, we explored the association of NS1 with 307 proteins involved, together with CAV-1, in the intracellular transport of cholesterol and the 308 dependence of NS1 secretion on such association. The results revealed that NS1 is associated 309 to the chaperone caveolin complex (CAV-1, Cy40, FKBP52 and CyA) and that the integrity of 310 this complex is necessary for the secretion of NS1 in mosquito infected cells. Additional data 311 suggest that in virus infected mosquito cells, ZIKV NS1, but not yellow fever virus NS1, also 312 associates with CAV-1 and usurps the cholesterol traffic pathways for secretion. Thus, the lipid 313 content or molecular structure of the DENV NS1 seems to have taken advantage of mosquito 314 cell cholesterol transport to guarantee the efficient NS1 secretion in the mosquito vector. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 16 325 Few works have reported a luciferase secretion system in insect cell lines [26] . Yet, as an 326 additional tool to study classical secretion in mosquito cell lines, an efficient Gaussia luciferase 327 reporter system that enables the detection of Luciferase in cell supernatants 6 hours after 328 transfection was developed. The secretion of luciferase was significantly affected after treatment 329 with GCA and BFA (data not shown); these data, together with the observed inhibition of viral 330 particle release, indicates the existence of a robust classical secretion pathway in mosquito cells. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint 18 371 (CSD) with a distance of 5Ǻ approximately. Further analysis also showed, high energy binding 372 through hydrophobic dynamics, between the NS1 -roll domain and FKBP52, CyA and Cy40 373 with distances under 6Ǻ. However, the model predicts all these proteins interacting at the same 374 domain in NS1 (data not shown), which seems unlikely, due to steric effects. Thus, we favor an 375 alternative molecular docking model, based on interaction energies, in which NS1 is bound 376 directly to CAV-1, and in turn FKBP52, CyA and Cy40 are bound to CAV-1 (Fig. 10D ). This 377 model still allowed interaction of NS1 with the CCC through membranes. However, if the NS1-378 CCC complex will traffic to the plasma membrane as a free-cytosolic complex or membrane 379 associated is totally unknown. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint . CC-BY 4.0 International license is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. The copyright holder for this preprint (which was not peer-reviewed) is the author/funder. It . https://doi.org/10.1101/370932 doi: bioRxiv preprint