Tupanvirus-infected amoebas are induced to aggregate with uninfected cells promoting viral dissemination

The discovery of giant viruses in the last years has fascinated the scientific community due to virus particles size and genome complexity. Among such fantastic discoveries, we have recently described tupanviruses, which particles present a long tail, and has a genome that contains the most complete set of translation-related genes ever reported in the known virosphere. Here we describe a new kind of virus-host interaction involving tupanvirus. We observed that tupanvirus-infected amoebas were induced to aggregate with uninfected cells, promoting viral dissemination and forming giant host cell bunches. Even after mechanical breakdown of bunches, amoebas reaggregated within a few minutes. This remarkable interaction between infected and uninfected cells seems to be promoted by the expression of a mannose receptor gene. Our investigations demonstrate that the pre-treatment of amoebas with free mannose inhibits the formation of bunches, in a concentration-dependent manner, suggesting that amoebal-bunch formation correlates with mannose receptor gene expression. Finally, our data suggest that bunch-forming cells are able to interact with uninfected cells promoting the dissemination and increase of tupanvirus progeny.

In this work, we describe that tupanviruses induces the expression of cellular and viral mannose receptor genes, which are well known for promoting the adhesion of amoebas to other cells. Tupanvirus promotes the formation of large bunches of infected and non-infected cells. In this context, we have hypothesized that infected cells act like "zombies, " controlled by tupanviruses, searching and attaching to uninfected cells, and improving the chances of the newly formed viral progeny to find a new host cell in which to propagate.

Materials and Methods
Cell culture, viral production and titration. Acanthamoeba castellanii cells (ATCC 30010) were cultivated in peptone-yeast extract with glucose (PYG) medium supplemented with 25 mg/ml amphotericin B (Fungizone; Cristalia, São Paulo, Brazil), 500 U/ml penicillin (Schering-Plough, Brazil) and 50 mg/ml gentamicin (Schering-Plough, Brazil). Cell culture flasks (175 cm 2 ) containing 7 × 10 6 A. castellanii cells were infected with tupanvirus strain soda lake at an multiplicity of infection (M.O.I) of 0.1 and incubated at 32 °C. After the appearance of a cytopathic effect and evidence of cell lysis, the cells and supernatants were collected and the viruses were purified through ultracentrifugation with a 22% sucrose cushion at 36,000 × g for 30 min. After viral purification, the virus titers were determined using the endpoint method 17,18 . Cytopathic effect and cycle characterization. To  For immunofluorescence, A. castellanii cells were infected with tupanvirus at an M.O.I 10 and at 0 (immediately after contact between tupanvirus and A. castellanii), 1, 8, 12, 16 and 24 h.p.i., approximately 2 × 10 5 cells were collected and centrifuged at 800 × g for 10 min. The pellet was resuspended in 50 µL of Page's amoebae saline (PAS), and the cells were attached to a slide via cytospin and fixed in methanol for 15 min. After fixation, cells were incubated with 3% bovine serum albumin (BSA)-PAS for 30 min, followed by three rinses with 0.1% PAS-Tween. Cells were then stained with polyclonal anti-tupanvirus whole particle antibody produced in mouse (1:400 diluted in 3% BSA-PAS) for 1 h at 37 °C, followed by three rinses with PAS-Tween 0.1%. After a 1 h incubation with an anti-mouse secondary antibody (1:400 diluted in 3% BSA-PAS), one drop of 0.01% Evans Blue (Sigma), which was sufficient to cover the cells, was added and incubated for 15 min at 37 °C, followed by three rinses with PAS-Tween 0.1%. Uninfected cells (control) were also fixed and prepared as described. Fluorescently labeled cells were observed using an Axio Imager Z2-Apotome 2 microscope (Zeiss). The Zen Lite software from Zeiss microscopy was used for image processing.
For transmission electron microscopy (TEM), the cells were pelleted for 10 min at 800 × g. The pellet was washed twice with 0.1 M phosphate buffer (pH 7.4) and fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer for 1 h at room temperature. The pellet was then washed twice with 0.1 M phosphate buffer and suspended in the same buffer. After, amoebas were embedded in Epon resin the using a standard method, as follows: 2 h of fixation in 2% osmium tetroxide, five washes in distilled water, overnight incubation in uranyl acetate 2% at 2-8 °C, two washes in distilled water, 10 min dehydration in increasing ethanol concentrations (35, 50, 70, 85, 95 and 100%), 20 min incubation in acetone and embedding in EPON resin. Ultrathin sections were subsequently analyzed under TEM (Spirit Biotwin FEI-120 kV).
For scanning electron microscopy (SEM) assays, 50 µL of supernatant was added to round glass coverslips coated with poly-L-lysine and fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer for 1 h at room temperature. The samples were washed three times with 0.1 M cacodylate buffer and post-fixed with 1% osmium tetroxide for 1 h at room temperature. After a second fixation, the samples were washed three times with 0.1 M cacodylate buffer and immersed in 0.1% tannic acid for 20 min. The samples were then washed in cacodylate buffer and dehydrated for 10 min by serial passages in ethanol solutions (35, 50, 70, 85, 95 and 100%). Samples were subsequently subjected to critical point drying using CO 2 , placed in stubs and metalized with a 5 nm gold layer. The analyses were completed using SEM (FEG Quanta 200 FEI).

Mannose binding protein transcripts analyses and mannose assays.
To evaluate the effect of mannose during tupanvirus infection, 1 × 10 6 A. castellanii cells were infected with tupanvirus at an M.O.I of 10 in PAS medium containing different concentrations of alpha-D-mannopyranoside (25, 50, 100, 400, 600 and 1000 mM) and incubated at 32 °C. The cells were observed under an optical microscope at 8 h.p.i. In addition, the culture flasks were divided (2.5 cm 2 /field) and the counting of bunches was performed under an optical microscope. Cells infected by tupanvirus in PAS medium without mannose and PAS medium with 600 mM of mannose were titrated at 24 h.p.i. As controls, uninfected A. castellanii cells either cultivated in PAS medium without mannose or PAS medium containing 600 mM of mannose were used. To investigate the role of the mannose in the expression pattern of the A. castellanii and tupanvirus mannose binding protein genes, total RNA was purified from 5 × 10 5 A. castellanii cells that had been infected with tupanvirus (M.O.I of 10) in the presence or absence of mannose (600 mM) at 1, 2, 4, 6, or 8 h.p.i. using the RNAeasy Mini Kit (Qiagen). Uninfected A. castellanii cells were used as controls. An control of infection using APMV was used to evaluate the expression of cellular MBP in amoebae infected with APMV, total RNA was purified from 5 ×   Evaluation of the interaction between uninfected amoebas and bunches. To evaluate whether bunches are able to interact with uninfected cells, 1 × 10 6 A. castellanii cells in 25 cm 2 culture flasks were infected with tupanvirus (M.O.I. of 10) and at 16 h.p.i. the bunches were collected and used to inoculate uninfected A. castellanii cells in suspension in a separate 25 cm 2 culture flasks. After incubation for 30 min at 32 °C the non-adherent cells were removed from the culture flasks and the cells that remained adhered were counted. As a control, 16 h.p.i. APMV-infected amoebae were used in the same procedure. The experiment was performed in triplicate. The results were plotted on a column graph, and the statistical significance was calculated using a two-tailed Student's t test, performed using Graph Pad Prism. In addition, the uninfected cells that interacted with bunches and were carried by them (the cells that did not adhere to the culture flasks again) were assayed using IF and SEM, as described above. To evaluate the importance of bunches with fresh cells for propagation of tupanvirus, 7 × 10 6 A. castellanii cells were infected (M.O.I of 10) and at 16 h.p.i. the medium was removed; the bunches were washed and used to infect new A. castellanii cells and as a control, the bunches were also added to a flask containing fresh medium. At 36 h.p.i. the supernatants were collected and titrated as described above. Graphs were constructed using GraphPad Prism software (v7.0 for Windows).  (Fig. 1A). Interestingly, tupanvirus exhibits a peculiar CPE that is characterized by amoebae aggregates, called "bunches. " The formation of bunches starts at approximately 6 h.p.i. and becomes more evident at 8, 12 and 16 h.p.i., when almost all the cells are seen incorporating into large bunches (Fig. 1A). From 24 h.p.i., despite the presence of bunches, cell lysis is more evident and is concomitant with the reduction of bunches size. At 32 h.p.i., the complete disaggregation of bunches was observed, and at 48-72 h.p.i. the majority of amoebae were lysed (Fig. 1A). The formation of bunches was also observed when a low M.O.I. (0.01) was used, although there was a temporal delay in bunch appearance (Suppl. Fig. 1). The CPE triggered by tupanvirus is unlike that of the other mimivirus, as well as other known giant amoebal viruses. Acanthamoeba polyphaga mimivirus, the prototype of the genus Mimivirus, triggers rounding and lysis in A. castellanii cell cultures, being a CPE commonly described for other giant viruses (M.O.I. of 10) (Suppl. Fig. 2). Different timepoints during infection (at an M.O.I. of 10) were selected for an immunofluorescence assay (IF) and transmission/scanning electron microscopy (TEM and SEM). IF assays using anti-tupan antibody revealed that at 0 and 1 h.p.i. tupanvirus particles appeared attached to amoebas (Fig. 2), even after the cell washing step, suggesting virion fibrils early interactions with host-cells, as demonstrated for other mimiviruses 19 . VFs start to be detected by IF in amoebas' cytoplasm using anti-tupanvirus whole-particle antibodies at 12 h.p.i. (Fig. 2). Curiously, even at high a M.O.I. it was not possible to observe a synchronous infection of all cells, something that was peculiarity related to tupanvirus and that needs further investigation. Remarkably, we observed amoebas with and without VFs or viral particles in the cytoplasm by IF, especially at 12 h.p.i., suggesting that early bunches are formed by the association of cells at different stages of the viral replication cycle (Fig. 2). Even at 16 h.p.i., when the viral titer reached its plateau (Fig. 1B) and the significant majority of cells already presented viral factories and particles (Fig. 2), we were able to find a very few cells with no detectable signs of tupanvirus by IF. By SEM (Fig. 3A) and TEM (Fig. 3B), it was possible at 16 h.p.i. to observe bunches of amoebas with damaged cells filled with or releasing tupanvirus particles, which were associating with cells with an intact plasma membrane and only a few particles inside them (Fig. 3).

Early amoebal bunches reform after mechanical separation of cells.
Another phenomenon that drew our attention was the occurrence of bunch reformation after mechanical breakdown (Fig. 4). Initially, we observed that early and late bunches could be easily disaggregated by homogenization using a pipette or by vortexing. However, 30 min after disaggregation, cells from early bunches (12 h.p.i.) were attracted to each other, forming large clusters of cells and assuming an appearance similar to early bunches before cell separation. In contrast, reformation of late bunches (24 h.p.i.) was not observed. In addition, using light microscopy we observed that some cells belonging to late bunches were larger in size, which was especially obvious after the disaggregation procedure (Fig. 4). The lack of bunch reformation at late timepoints indicates that at 24 h.p.i., cells were already killed by the viral infection, and a simple homogenization was able to permanently disaggregate the cells that formed late bunches. To confirm this hypothesis, we performed SEM on late bunches. Interestingly, we observed cells with the plasma membranes almost completely degraded and filled with intricate networks of tupanvirus particles and cell debris (Fig. 5).
Taken together, these results indicate that tupanvirus promotes the formation of large aggregates of amoebas, which are able to re-aggregate as long as cells are still alive. These findings motivated us to investigate possible factors that may interfere in bunch formation and the biological relevance of tupanvirus-promoted bunches. Amoebal-bunch formation correlates with increased mRNA levels of mannose binding protein genes. Acanthamoeba spp. is widely distributed in the environment and can cause human diseases, mainly associated with immunocompromised patients and contact lenses wearers. Furthermore, Acanthamoeba spp. has a relevant role in ecosystems, due to its ability to act as a host for microbial pathogens [20][21][22][23] . Mannose receptor proteins are well known for promoting interactions between amoebas and other cells. In this way, previous studies revealed that mannose-binding protein (MBP) present on the surface of Acanthamoeba is one of the major virulence proteins and may be essential for pathogenesis and invasion of host cells 11,15,24 demonstrated that the potential of Acanthamoeba pathogenicity is directly dependent on the level of MBP 24 . In addition, previous studies showed that A. castellanii treated with mannose had reduced adhesion and cytotoxicity, strongly suggesting that its pathogenicity is associated with MBP 11,25 . Interestingly, after an in-depth analysis of the tupanvirus soda lake The viral particles are in green (anti-tupan particle antibody) and amoeba cytoskeleton in red (stained by Evans Blue). Scale bar, 100 µm. genome, we found that the virus has an 525 bp MBP gene. The domain involved in mannose recognition contains a three-fold internal repeat, and the consensus sequence motif is QXDXNXVXY. A previous study has suggested that the QDN subdomains may be essential for functionality of the protein. In addition, the importance of Y in subdomain of the MBP-mannose interaction was demonstrated 26 . This finding led us to investigate the MBP viral and cellular expression during the replication of tupanvirus and the possible effects of free mannose on the formation of bunches. An analysis of cellular MBP gene expression in uninfected A. castellanii cells showed a basal level of MBP transcripts, meaning it is likely essential for cellular processes (Fig. 6A). However, during tupanvirus infection, the levels of cellular MBP transcripts significantly increased (p = 0.0001) at earlier times during infection (1, 2 and 4 h.p.i.). The increase of cellular MBP gene   (Fig. 6B). Therefore, our data indicate that the expression of cellular and viral MBP genes is induced during tupanvirus infection, suggesting a possible relevance of these genes in the viral replication cycle (Fig. 6). Interestingly, orthologous MBP genes were found in the genome of other members of the Mimiviridae family, such as in the prototype APMV. Phylogenetic analysis of the amino acid sequence of viral MBP showed that tupanviruses forms an out-grouping among amoebal-mimiviruses (Suppl. Fig. 3A). Considering APMV does not induces the formation of bunches in infected amoebas, an analysis was performed to determine whether APMV was able to induce the expression of cellular MBP, as well as whether this virus express its own MBP gene throughout the time-course of infection. We observed that APMV induced a decrease in cellular MBP gene expression, even when compared to uninfected cells, in contrast to tupanvirus soda lake, which significantly increases cellular MBP gene expression which may be related to the induction of bunch formation (Suppl. Fig. 4B).
The expression of APMV MBP gene transcripts was also analyzed during the time-course of APMV infection in A. castellanii, revealing that this gene was significantly expressed at 6 and 8 h.p.i. (Suppl. Fig. 4A). However, an analysis of the APMV MBP amino acid sequence revealed polymorphisms in all three of the repeats that compose the catalytic site of the protein (the mannose-binding domain), more specifically in the QDN subdomains (Suppl. Fig. 3B), which have been associated with a loss of MBP function 26 . Polymorphisms in the Y amino acids have also been observed in the first repeat of the APMV MBP catalytic site (Suppl. Fig. 3B), which has also been associated with a loss of MBP function 26 . In contrast, tupanvirus MBP has no polymorphisms in any of the repeats in the catalytic site, when compared to the functional protein previously studied (Suppl. Fig. 3B) 26 . In summary, mutations in the mannose-binding domains of the APMV MBP gene could indicate that this protein is not able to bind to mannose, although more studies into structural level are needed to confirm this hypothesis. Previous work have demonstrated that free-mannose negatively affects Acanthamoeba adhesion to cell surfaces and interferes with amoeba pathogenicity 16,25 . Therefore, the impact of free-mannose on cellular and viral MBP gene expression was here evaluated in uninfected and tupanvirus-infected cells. Our results showed that free mannose negatively affects the expression of the cellular MBP gene, maintaining it at close to basal levels (Fig. 6A). The same negative effect was observed in relation to the expression of the tupanvirus MBP gene (Fig. 6B). Considering these results, the effect of free mannose on the cytopathic effect of tupanvirus was evaluated as well. Interestingly, when Acanthamoeba was pre-treated with free-mannose, there was a gradual inhibition of bunch formation in a dose-dependent manner (Fig. 7), suggesting that formation of amoebal-bunches correlates with the suppression of viral and cellular MBP genes (Fig. 7).
Previous studies have shown that Acanthamoeba MBP is itself a glycoprotein containing mannose 15 , which may be one of the factors that explains the formation of the bunches in cells infected by tupanvirus, since this virus induces the expression of the MBP gene. Thus, the induction of mannose receptor genes in the tupanvirus replication cycle may be important for the optimization of the formation of bunches, since the interaction between amoebas may occur through interactions among their receptors.

Bunch-forming cells are able to interact with uninfected cells.
As previously characterized in this study, the formation of the bunches starts at approximately 6 h.p.i. and persists until at least 24 h.p.i. At 16 h.p.i., almost all the cells are infected and clustered forming large bunches (Fig. 1A). An experiment was designed in order to evaluate whether bunch-forming cells were able to interact with uninfected cells (Fig. 8A). For this, A. After incubation, the supernatant (containing bunches and eventual non-infected amoebas interacting with bunches) were removed. The cells that remained adhered to the flask (the ones that did not interact with bunches) were counted to evaluate if the bunches were able to hijack the uninfected cells (Fig. 8). In addition, the supernatant was submitted to SEM and IF, using tupanvirus-specific antibodies. APMV-infected amoebae, which do not induce bunch formation, were used as a control in order to compare whether APMV-infected cells were also able to interact with and hijack uninfected cells (Fig. 8B).
Our results revealed that tupanvirus-induced bunches are able to bind to and hijack many uninfected cells. The number of cells remaining in the flasks after interaction with tupanvirus-induced bunches was significantly lower than those in flasks inoculated with APMV-infected cells (Fig. 8B). The interaction among tupanvirus-induced bunches and uninfected cells was also observed by SEM; it was possible to observe infected cells, which were rounded, interacting with fresh trophozoites presenting their typical pseudopods (Fig. 9). IF assays also showed bunches containing a mix of infected and uninfected cells (data not shown). Lastly, considering the ability of tupanvirus-induced bunches to bind and hijack uninfected cells, we evaluated whether such interactions could promote an increase in viral titer. For this, bunches were collected at 16 h.p.i. and inoculated in flasks containing fresh, uninfected amoebas, as previously described above. After incubation, the supernatant containing bunches attached to uninfected cells was collected and transferred to a new flask. Infectious particles in these population of cells were tittered at 0 and 36 h.p.i, revealing an of almost 2 log increase in viral titer (Suppl. Fig. 5).
Tupanviruses has the broadest host spectrum known among amoebal giant viruses 1 . While most of the giant viruses, such as cedratvirus, marseilleviruses, mollivirus, pandoraviruses, mimivirus, faustovirus and kaumoebavirus, are known to be able to replicate only in a single amoeba genus, tupanvirus is able to replicate in Acanthamoeba spp., Vermoameba vermiformis, Dysctiostelium discoideum and Willartia magna 1,5-9,27,28 . The generalist profile displayed by tupanviruses may be related to the extreme aquatic environments where they were isolated (high salinity, pH and depths); within inhospitable environments there is a lower species richness and abundance, meaning there are more hosts available and a better chance of a future infection 29 . In this way, bunch formation induced by tupanvirus may be important to the recruitment of cells, improving the chances of viral progeny to find a new host cell that may be scarce in those environments. In addition, bunch formation can reduce the dilution effect in aquatic environments, thereby allowing the release of more particles near the host, which could also facilitate the encounter between viral particles and their hosts.

Concluding Remarks
Here we described a new virus-host interaction mechanism in which tupanvirus-infected amoebas are induced to aggregate with uninfected cells, forming bunches that can increases tupanvirus progeny. In addition, our results demonstrated that mannose reduced the transcript levels of cellular and viral mannose receptor genes, as well as inhibited bunch formation in a concentration-dependent manner, suggesting that amoebal-bunch formation correlates with MBP gene expression. Taken together, our results provide insights into the ecology of these intriguing viruses.