Flavivirus infections induce a Golgi stress response in vertebrate and mosquito cells

The stress of the Golgi apparatus is an autoregulatory mechanism that is induced to compensate for greater demand in the Golgi functions. No examples of Golgi stress responses due to physiological stimuli are known. Furthermore, the impact on this organelle of viral infections that occupy the vesicular transport during replication is unknown. In this work, we evaluated if a Golgi stress response is triggered during dengue and Zika viruses replication, two flaviviruses whose replicative cycle is heavily involved with the Golgi complex, in vertebrate and mosquito cells. Using GM-130 as a Golgi marker, and treatment with monensin as a positive control for the induction of the Golgi stress response, a significant expansion of the Golgi cisternae was observed in BHK-21, Vero E6 and mosquito cells infected with either virus. Activation of the TFE3 pathway was observed in the infected cells as indicated by the translocation from the cytoplasm to the nucleus of TFE3 and increased expression of pathway targeted genes. Of note, no sign of activation of the stress response was observed in CRFK cells infected with Feline Calicivirus (FCV), a virus released by cell lysis, not requiring vesicular transport. Finally, dilatation of the Golgi complex and translocation of TFE3 was observed in vertebrate cells expressing dengue and Zika viruses NS1, but not NS3. These results indicated that infections by dengue and Zika viruses induce a Golgi stress response in vertebrate and mosquito cells due to the increased demand on the Golgi complex imposed by virion and NS1 processing and secretion.


The stress of the Golgi apparatus is an autoregulatory mechanism that is induced to compensate for greater demand in the Golgi functions. No examples of Golgi stress responses due to physiological stimuli are known. Furthermore, the impact on this organelle of viral infections that occupy the vesicular transport during replication is unknown. In this work, we evaluated if a Golgi stress response is triggered during dengue and Zika viruses replication, two flaviviruses whose replicative cycle is heavily involved with the Golgi complex, in vertebrate and mosquito cells. Using GM-130 as a Golgi marker, and treatment with monensin as a positive control for the induction of the Golgi stress response, a significant expansion of the Golgi cisternae was observed in BHK-21, Vero E6 and mosquito cells infected with either virus. Activation of the TFE3 pathway was observed in the infected cells as indicated by the translocation from the cytoplasm to the nucleus of TFE3 and increased expression of pathway targeted genes. Of note, no sign of activation of the stress response was observed in CRFK cells infected with Feline Calicivirus (FCV), a virus released by cell lysis, not requiring vesicular transport. Finally, dilatation of the Golgi complex and translocation of TFE3 was observed in vertebrate cells expressing dengue and Zika viruses NS1, but not NS3. These results indicated that infections by dengue and Zika viruses induce a Golgi stress response in vertebrate and mosquito cells due to the increased demand on the Golgi complex imposed by virion and NS1 processing and secretion.
The endoplasmic reticulum (ER) and the Golgi apparatus are two tightly connected organelles that are key in the synthesis, processing, and secretion of eukaryotic cells proteins. These organelles are bound by membranes that allow for the compartmentalization of specialized cellular functions. Upon an excessive, overwhelming demand of the functions of the ER or the Golgi apparatus, these organelles respond by a mechanism of homeostatic autoregulation termed stress responses, aimed to augment their functional capacity [1][2][3] .
Viral infections are inductors of stress in the infected cell. In particular, flavivirus replication processes leading to the production of progeny virions, interfere with a number of aspects of the cellular metabolism, triggering a homeostatic disequilibrium 4,5 . The flavivirus genome is a single stranded positive sense RNA molecule of approximately 10 kb in length, which encodes one large polyprotein that, after proteolytic processing by host and viral proteases, gives rise to the three structural proteins (capsid, membrane precursor, and envelope) and seven no structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5) that play essential roles in genome replication, polyprotein processing and virion assembly. Translation and replication of the flavivirus genome, as well as virion formation are all processes that occur in the ER and involve extensive ER membrane modifications [6][7][8][9] . Consequently, ER stress and the activation of the Unfolded-Protein-Response (UPR) have been well documented in vertebrate and mosquito cells infected with dengue (DENV) or Zika (ZIKV) viruses 5,10-13 .
After leaving the ER complex, immature virions enter the Golgi apparatus and following a classical secretory pathway, are transported to the plasma membrane to be released out of the cell by exocytosis. During their transit thru the Golgi complex, immature virions undergo glycan modifications in the envelope (E) protein and proteolytic cleavage of the membrane precursor protein (M), by action of the Golgi resident protease furin, to become mature infectious virions [14][15][16] . In addition, DENV and ZIKV infected vertebrate and mosquito cells, also secrete the NS1 protein as a hexamer. In vertebrate cells, the NS1 protein is released following a classical www.nature.com/scientificreports/ secretory pathway using the Golgi apparatus, where the N-glycans attached to NS1 suffer further processing 5,6,8,17 . At variance, in mosquito cells, the NS1 protein is secreted using a non-classical secretory pathway in association with the chaperone caveolin complex 17,18 . All those secretory processes may impact and overwhelm the capacity of the Golgi. The Golgi apparatus is an organelle with various functional zones, including the cis-Golgi network, cis-Golgi, medial-Golgi, trans-Golgi, trans-Golgi network, separated by membranes and distinct cisterns, where diverse cellular functions such as glycosylation, sulfation, fucosylation and phosphorylation of secretory and membrane proteins is carried out 2,3 . In the cells that need a greater demand for the functions of Golgi, as could be the case of flaviviruses infected cells, the capacity of this organelle collapses. To compensate this functional demand a mechanism of autoregulation homeostatic is induced, which is termed Golgi stress response which involve several activation or response pathways [1][2][3]19 .
In this work, we explore if the high involvement of the Golgi complex during flavivirus replication results in the induction of the Golgi stress response. Our results indicate that DENV and ZIKV infection activates the Golgi stress response in both vertebrate and mosquito cell lines. Moreover, the sole expression of the NS1 protein can induce the stress response in vertebrate cells. Interestingly, the replication of feline calicivirus, a virus not involved with the Golgi apparatus, do not result in a stress response. The understanding of the activation and regulation of the Golgi stress response is still incomplete 3 ; our results suggest that flavivirus infections can become an additional tool to study this phenomena.

Results
Flavivirus infection results in a Golgi stress response in both vertebrate and mosquito cells. The Golgi stress response is a consequence of an increased demand on the organelle functions. Given that flavivirus replication and NS1 and virion secretion are processes heavily dependent on the Golgi apparatus, we tested if a Golgi stress response is induced in cells infected with DENV or ZIKV. As a positive control, the Na + /H + ionophore monensin was used at non-toxic concentrations. Monensin is an agent that blocks glycosylation and vesicular transport of proteins and induce Golgi-stress response in treated cells 3,20 . Golgi expansion or fragmentation was visualized using GM-130 as a marker. In mock infected cells, GM-130 mark was observed a dotted staining near the nuclei of the cell, that correspond to the reporter Golgi subcellular localization (Fig. 1). Monensin treatment results in a fuzzy pattern of GM-130 staining throughout all the cytoplasm, indicating fragmentation and dispersion of the Golgi apparatus in treated cells (Fig. 1). A pattern like the one obtained with monensin treatment was observed in the Vero, BHK-21 and C6/36 cells infected with DENV and ZIKV at 48 h www.nature.com/scientificreports/ post-infection (Fig. 1). These results suggest that DENV and ZIKV replication results in a Golgi stress response in vertebrate as well as mosquito cells.

DENV and ZIKV infection in vertebrate cells results in the activation of the transcription factor E3 (TFE3) pathway.
One of the pathways activated during the Golgi stress response is the TFE3 pathway.
TFE3 located in the cytoplasm, is dephosphorylated, and translocated to the cell nuclei upon activation of the Golgi stress response. To corroborate the activation of the stress response, vertebrate cells transfected with a recombinant form of TFE3 were infected with DENV and ZIKV. As shown in Fig. 2A, in untreated cells the mark corresponding to TFE3 is located mainly in the cytoplasm. However, monensin treatment or viral infections resulted in a significant translocation of TFE3 from the cytoplasm to the cell nuclei. In Vero-E6 cells, a statistically significant difference was observed in the percentage of TFE3 translocation between the mock condition (27.8%) and cells treated with monensin (95.2%) or cells infected with DENV (88.2%) or ZIKV (83.9%). A similar effect was observed in BHK 21 cells, with a percentage of TFE3 translocation of 21.7% in untreated cells, 96.8% in cells treated with monensin, 78% in DENV infected cells and 80.4% in ZIKV infected cells (Fig. 2B).
These results indicate that the TFE3 pathway is activated in vertebrate cells infected with DENV and ZIKV. Since no expression of TFE3 was observed in transfected mosquito cells, C6/36 were not considered in this experiment.
To corroborate the activation of the TFE3 pathway and Golgi stress response in DENV and ZIKV infected cells, the expression of two genes targeted by this pathway was evaluated by qRT-PCR in infected BHK-21 cells, harvested at 24 hpi. Figure 3 shows that upon infection with DENV or ZIKV, the mRNA levels of the GOLGA2 and GCP60 genes, encoding respectively for the structural GM130 and GCP60 proteins, were significantly increased in relation to mock infected cells, and to levels comparable with those observed in monensin treated cells. These results confirm the activation of the TFE3 pathway upon infection.
FCV infection do not activate the Golgi stress response. FCV is a highly lytic virus, whose replication cycle have little or no involvement with the Golgi apparatus 21 . Thus, to address if the observed activation of the Golgi stress response was specific for DENV and ZIKV infections, CRFK cells infected with FCV were analyzed for Golgi expansion or translocation of the TFE3 factor. CRFK cells were chosen because they are highly permissive to FCV and will magnify any putative use of the Golgi complex. GM-130 staining of mock infected CRFK cells was observed as a dotted mark near the nuclei. In addition, TFE3 was observed in the cytoplasm of untreated cells. No changes in the architecture of the Golgi complex or translocation of the TFE3 factor to the cell nuclei were observed in infected CRFK fixed at 5 and 7 hpi (Fig. 4). Yet, CRFK cells can mount a Golgi stress response as indicated by the results obtained in monensin treated cells, where changes in the GM-130 mark and On the other hand, when the ER resident NS3 protease was expressed, no activation of the response was observed (Fig. 5). www.nature.com/scientificreports/ These results suggest that the expression and transit of NS1 alone is sufficient to activate the Golgi stress response in vertebrate cells.

Discussion
The mosquito borne flaviviruses DENV and ZIKV are etiologic agents of important public health diseases affecting tropical and subtropical countries around the globe 22 . The replicative cycle of these and other flavivirus takes place in close contact with membrane bound organelles which are directly or indirectly affected by the viral replicative process. Extensive ER membrane modifications and induction of an UPR, changes in the shape of mitochondria, and even nuclear pore degradation have all been reported in DENV infected cells 13,23,24 . Yet, and despite that the processing and secretion of hexameric NS1 and mature virions in vertebrate cells, as well as the processing and secretion of mature virions in mosquito cells, are all processes that take place in the Golgi apparatus 17 , little is known regarding the response of the Golgi system to flavivirus infections. In this work, evidence obtained in three different cell lines, from vertebrate and mosquito origin, is presented indicating that DENV and ZIKV infections triggers the Golgi stress response and the activation of the TFE3 pathway in infected vertebrate cells. If other mosquito borne flaviviruses not included in this study, such as JEV or WNV, or even the HCV induce a Golgi stress response in infected cells remains to be determined, but given the heavy involvement of all flaviviruses with the Golgi apparatus 7 , this is expected to be the case. Interestingly, the activation of the Golgi stress response seems not to be a generalized phenomenon of viral infections as suggested by the observation that FCV does not trigger these responses in CRFK infected cells, although these cells possess all the molecular mechanisms necessary to go into Golgi stress.
In their transit to the extracellular space, immature DENV and ZIKV virions enter the Golgi apparatus where the E protein suffers trimming of the N-glycan residues 25 and later the prM protein is proteolytically processed by the Golgi resident acidic protease furin to finally form mature virions 15 . In addition, in vertebrate cells, the secretion of NS1 protein from infected cells utilizes the entire Golgi system, where the two N-glycan residues attached to NS1 undergo further processing 25 . Moreover, the NS1 located on the plasma membrane is GPIanchored protein, another function that take place in the Golgi cisternae 26 . More recently, multiple components of the Golgi apparatus were found in the interactome of NS1 in vertebrate cells 27 . In this work, we demonstrate that activation of the Golgi stress response occurs also in cells expressing NS1 by itself. So, along with vesicle transport, flavivirus proteins such as E, pr-M and NS1 suffer several post-transcriptional modifications in their transit over the Golgi cisternae. The TFE3 pathway activation ends in the expression of genes involved in augmenting the functions of the Golgi apparatus in general, including Golgi structure, N-glycosylation functions and vesicular transport, rather than a specific function 3 . Thus, the activation of the TFE3 pathway in DENV and ZIKV infected cells is in line with the import and multiple roles played by the Golgi complex during flavivirus replication. Other pathways also activated during the Golgi stress response, such as the CREB3 and the HSP47 pathways conduct to the activation of pro-apoptotic and anti-apoptotic genes 3 . The infection of mouse endothelial cells by HSV-1 causes GM-130 degradation and Golgi apparatus fragmentation that finally leads to cell apoptosis 28 . Both, activation and suppression of apoptosis have been reported in DENV infected cells 29,30 . If the CREB3 or the HSP47 pathways are activated during DENV and ZIKV replication is unknown. www.nature.com/scientificreports/ The UPR triggered during flavivirus replication have been linked with mechanisms favoring viral replication such as modulation of innate immunity and induction of autophagy 10,11,29 . The significance of the activation of the Golgi stress response during flavivirus replicative cycle is unclear, but it is reasonable to speculate that augmenting the capacity of the Golgi apparatus will favor the secretion of NS1 and mature virions. Consequently, a better understanding of the interplay between DENV and ZIKV proteins and the Golgi apparatus may help to identify antiviral targets common to vertebrate and mosquito cells.
Finally, we like to remark that several molecular aspects of the Golgi stress response, as for example, the sensor molecule for the activation of the TFE3 pathway, are still unknown. Currently, the TFE3 pathway is activated mainly using pharmacological treatments, such as monensin and nigericin 3 . Our results revealed, for the first time, that the Golgi stress response is indeed activated under physiological conditions. In addition, the results here presented suggest that infections with DENV and ZIKV, or expression of recombinant NS1 may become additional experimental tools in the study of the Golgi apparatus and its stress response. Numerous are the examples of how the study of viruses had been rewarding to study the biology of cells. Virus infections. Confluent cell monolayers grown in 24-well plates (2 × 10 5 cells per well) were infected for 2 h, using a MOI = 3 for DENV and ZIKV, and a MOI = 5 for FCV. After infection, the monolayers were washed three times with PBS to remove unabsorbed viruses, and infections allowed to proceed for 48 h for DENV or ZIKV, and 5 and 7 h for FCV. Afterward, cells were washed, fixed and stained for immunofluorescence.

Plasmids transfection.
Plasmids expressing recombinant DENV-NS1 and DENV-NS3 were kindly donated by Dr. Ana Fernández-Sesma (Icahn School of Medicine at Mount Sinai, New York). Recombinant plasmids expressing the transcriptional factor E3 (pTFE3-Myc) were kindly donated by Dr. Hiderou Yoshida (University of Hyogo, Japan). Plasmids were transfected into confluent monolayer of cells grown in 24-well plates using lipofectamine reagent Lipofectamine TM 2000 (11668019, Invitrogen); each well was transfected with 1 μg of plasmid DNA and 2 μL of Lipofectamine in a final volume of 0.25 ml of EMEM w/o FBS. After 5 h of transfection, 0.25 ml of EMEM supplemented with 10% FBS were added per well. Cells transfected for DENV-NS1 and DENV-NS3 were fixed and stained for immunofluorescence at 24 h post transfection. Cells transfected with pTFE3-Myc were infected with DENV or ZIKV 24 h after transfection and the infection allowed to proceed for additional 48 h before fixing and staining for immunofluorescence. In addition, cells transfected with pTFE3-Myc were treated with monensin for 24 h, fixed and stained for immunofluorescence.