Differential PI3K–Akt–mTOR activation by Semliki Forest and 1 chikungunya virus, dependent on nsP3 and connected to replication 2

complex internalisation 3 4 RUNNING TITLE: PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 5

6 Authors: 7 Bastian Thaa (1, 5), Roberta Biasiotto (1), Kai Eng (1), Maarit Neuvonen (2), 8 Benjamin Götte (1), Lara Rheinemann (1), Margit Mutso (3), Age Utt (3), Finny 9 Varghese (4), Giuseppe Balistreri (2), Andres Merits (3), Tero Ahola (4) and 10 Gerald M. McInerney (1) 11 1: Karolinska Institutet, Department of Microbiology, Tumor and Cell Biology (MTC), 12 Nobels väg 16, SE-171 77 Stockholm, Sweden 13 2: University of Helsinki, Institute of Biotechnology, Viikinkaari 9, FI-00014 Helsinki, 14 Finland 15 3: University of Tartu, Institute of Technology, Nooruse 1, EE-50411 Tartu, Estonia 16 4: University of Helsinki, Department of Food and Environmental Sciences, Viikinkaari 9, 17 FI-00014 Helsinki, Finland 18 5: Corresponding author (bastian.thaa@ki.se) 19

JVI Accepted Manuscript Posted Online 2 September 2015J. Virol. doi:10.1128/JVI.01579-15Copyright © 2015, American Society for Microbiology. All Rights Reserved.

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 2 Abstract 20 Many viruses affect or exploit the phosphatidylinositol-3-kinase (PI3K)–Akt–21 mammalian target of Rapamycin (mTOR) pathway, a crucial pro-survival signalling 22 cascade. We report that this pathway was strongly activated in cells upon infection with 23 the Old World alphavirus Semliki Forest virus (SFV), even under conditions of complete 24 nutrient starvation. We mapped this activation to the hyperphosphorylated/acidic 25 domain in the C-terminal tail of SFV non-structural protein nsP3. Viruses with a deletion 26 of this domain (SFV-Δ50), but not other regions in nsP3, displayed a clearly delayed and 27 reduced capacity of Akt stimulation. Ectopic expression of nsP3-wildtype, but not -Δ50, 28 equipped with a membrane anchor was sufficient to activate Akt. We linked PI3K–Akt–29 mTOR stimulation to the intracellular dynamics of viral replication complexes, which are 30 formed at the plasma membrane and subsequently internalised in a process blocked by 31 the PI3K inhibitor Wortmannin. Replication complex internalisation was observed upon 32 infection of cells with SFV-wt and SFV mutants with deletions in nsP3, but not with SFV-33 Δ50, where replication complexes were typically accumulated at the cell periphery. In 34 cells infected with the closely related chikungunya virus (CHIKV), the PI3K–Akt–mTOR 35 pathway was only moderately activated. Replication complexes of CHIKV were 36 predominantly located at the cell periphery. Exchanging the hypervariable C-terminal 37 tail of nsP3 between SFV and CHIKV induced the phenotype of strong PI3K–Akt–mTOR 38 activation and replication complex internalisation in CHIKV. In conclusion, infection 39 with SFV but not CHIKV boosts PI3K–Akt–mTOR through the 40 hyperphosphorylated/acidic domain of nsP3 to drive replication complex 41 internalisation. 42

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 3 Importance 43 SFV and CHIKV are very similar in terms of molecular and cell biology, e.g. regarding 44 replication and molecular interactions, but are strikingly different regarding pathology: 45 CHIKV is a relevant human pathogen, causing high fever and joint pain, while SFV is a 46 low-pathogenic model virus, albeit neuropathogenic in mice. We show that SFV and 47 CHIKV both activate the pro-survival PI3K–Akt–mTOR pathway in cells, but greatly 48 differ in their capacity to do so: Akt is strongly and persistently activated by SFV 49 infection, but only moderately by CHIKV. We mapped this activation capacity to a region 50 in the non-structural protein 3 of SFV and could functionally transfer this region to 51 CHIKV. Akt activation is linked to the subcellular dynamics of replication complexes, 52 which are efficiently internalised from the cell periphery for SFV but not CHIKV. This 53 difference in signal pathway stimulation and replication complex localisation may have 54 implications for pathology. 55 56 57 Keywords 58 Chikungunya virus, Semliki Forest virus, Akt, membrane dynamics, replication complex 59

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 4 Introduction 60 Alphaviruses are positive-sense RNA viruses grouped into the family Togaviridae and 61 differentiated into Old World and New World alphaviruses. Prominent examples of Old 62 World alphaviruses comprise well-studied model viruses such as Semliki Forest virus 63 (SFV) and Sindbis virus (SINV) as well as human pathogens such as chikungunya virus 64 (CHIKV). CHIKV is spread by tropical mosquitoes of the Aedes family and causes 65 chikungunya fever, an illness characterised by high fever and debilitating joint pain. In 66 recent years, several big chikungunya outbreaks have occurred in the Indian Ocean area, 67 in Asia, and recently the Caribbean (www.cdc.gov/chikungunya/geo). 68 SFV is not associated with major disease in humans, but has been employed as a model 69 for viral pathogenesis in mice (1). SFV also serves as a basis for viral vectors for gene 70 therapy and vaccination (2-4). 71 SFV and CHIKV, though different in terms of disease and pathology, are very closely 72 related, evidenced by classification as members of the same serological group, the 73 Semliki Forest antigenic cluster (5). All Old World alphaviruses are very similar 74 regarding their cell biology and replication processes (for review, see (6, 7)). After cell 75 entry and uncoating of the virus, the viral genome serves directly as mRNA for 76 translation of the viral non-structural proteins (nsP) as a polyprotein, cleaved 77 successively by nsP2 into nsP1 (mRNA capping enzyme), nsP2 (RNA helicase, protease), 78 nsP3 and nsP4 (RNA-dependent RNA polymerase). The functions of nsP3 have long been 79 enigmatic, but there is growing evidence that the protein is a relevant player for virus–80 host interaction. Old World alphavirus nsP3 comprises an N-terminal macro domain that 81 binds ADP-ribose moieties (8, 9), an essential zinc-binding region in the middle of the 82 protein (10) as well as a C-terminal hypervariable domain. This intrinsically 83

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 5 unstructured region serves as a hub for protein–protein interactions (11); it contains a 84 hyperphosphorylated/acidic domain, a proline-rich domain, and a C-terminal region 85 with two FGDF motifs. These motifs mediate binding to the cellular protein G3BP (Ras-86 GAP SH3 domain binding protein), an interaction which counteracts the formation of 87 stress granules (12-14). These are dynamic RNA/protein aggregates, known as a cellular 88 response to stress such as virus infection and possibly linked to cellular signalling (15). 89 After processing from the polyprotein, the nsPs stay connected by protein–protein 90 interactions and form the viral replication complex, which is bound to cellular 91 membranes by nsP1. (The nsPs also have other, replication complex-independent 92 subcellular localisations and functions.) The replication complex is initially formed at 93 the plasma membrane and comprises bulb-shaped membrane invaginations termed 94 spherules, which contain double-stranded RNA replication intermediates, shielded from 95 recognition by cytosolic pattern-recognition factors. Later, the spherules are 96 internalised from the plasma membrane to form large intracellular cytopathic vacuoles 97 (CPV-I), modified endosomes/lysosomes containing multiple replication complexes. 98 Replication complex internalisation depends on the phosphatidylinositol-3-kinase 99 (PI3K)–Akt signalling pathway as well as on the actin cytoskeleton and microtubules 100 (16). In the replication complex, the various viral RNA species are generated in a 101 regulated manner: complementary negative strand; full-length positive strand; 102 subgenomic mRNA encoding the structural proteins of the virus, which ultimately give 103 rise to progeny virus particles. 104 Virus infection generally interferes with and exploits numerous cellular functions and 105 pathways. One such pathway is the PI3K–Akt–mammalian target of Rapamycin (mTOR) 106 pathway, which transduces growth factor signals to convey a “pro-survival” state 107

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 6 (reviewed in (17, 18); see Fig. 1A for a simplified schematic representation). The 108 pathway involves PI3K, catalysing the generation of the membrane phospholipid 109 phosphatidylinositol-3,4,5-triphosphate (PtdInsP3), a cue for recruitment of the protein 110 kinase Akt to the plasma membrane. This recruitment is a prerequisite for activation of 111 Akt by phosphorylation at first threonine-308 and then serine-473. Activated Akt 112 phosphorylates a multitude of targets, many of which mediate proliferation, act in an 113 anti-apoptotic fashion or modulate cytoskeleton dynamics. One (indirect) downstream 114 target of Akt is mTOR, the central metabolic regulator in cells. mTOR, forming a large 115 protein complex termed mTORC1, senses nutrient availability (e.g., amino acids, 116 adenosine triphosphate [ATP] and growth factors; the latter signalled through Akt). 117 Active mTOR ensures efficient cellular translation by inducing phosphorylation of 118 downstream targets such as 4EBP1 (eukaryotic initiation factor 4E-binding protein 1) 119 and the ribosomal protein S6. These phosphorylation events commonly serve as a 120 readout for mTOR activity and are impeded by nutrient starvation. 121 The PI3K–Akt–mTOR pathway is affected by many viruses and often required for 122 efficient replication (19-21). While in particular mTOR activation is a common feature 123 especially of DNA virus infection (22), it has also been found that many—but not all—124 RNA viruses activate (23) and others inhibit the PI3K–Akt–mTOR pathway (24). Thus, 125 viruses differ in their mode of influencing this signalling pathway. 126 In spite of its high functional relevance for virus infection, the PI3K–Akt–mTOR pathway 127 has been investigated only sparsely in Old World alphavirus infection, yielding partially 128 conflicting results: SINV infection was found to suppress PI3K–Akt–mTOR late in 129 infection (25), but to activate the pathway in arthropod cells (26), and Akt was recently 130 suggested to be stabilised and activated in CHIKV-infected cells (27). Pharmacological 131

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 7 inhibition of the PI3K–Akt–mTOR pathway generally has no or only little effect on 132 alphavirus titres (16, 25), but it was shown that at least one feature of virus replication 133 is affected by such treatment: the internalisation of replication complexes, which is 134 blocked by the PI3K inhibitor Wortmannin (16). It is at present unclear how the virus 135 regulates the dynamics of the replication complexes, but nsP3 has been suggested to be 136 relevant for the internalisation process (28). 137 In order to clarify these issues, we endeavoured to assess systematically how infection 138 with the Old World alphaviruses SFV and CHIKV interferes with the PI3K–Akt–mTOR 139 pathway and to evaluate the implications on the subcellular localisation of replication 140 complexes. We report here that the PI3K–Akt–mTOR pathway was gradually activated in 141 cells upon infection with SFV or CHIKV. We observed that SFV but not CHIKV infection 142 boosted PI3K–Akt activity to promote replication complex internalisation and mapped 143 this to the hyperphosphorylated/acidic domain of SFV-nsP3. 144 145

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 8 Material and Methods 146 147 Cell culture 148 Human osteosarcoma cells (HOS, ATCC CRL-1543) and immortalised mouse embryonal 149 fibroblasts (MEF, (13)) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) 150 supplemented with 10% foetal bovine serum (FBS), 2 mM L-glutamine and 151 penicillin/streptomycin at 37 °C, 5% CO2, 95% humidity. Baby hamster kidney (BHK) 152 cells (ATCC CCL-10) were kept in Glasgow’s modified Eagle’s medium (GMEM) 153 supplemented with 10% FBS, 10% tryptose phosphate broth, 20 mM HEPES, 1 mM L-154 glutamine and penicillin/streptomycin. For starvation, cells were supplied with Earle’s 155 balanced salt solution (Sigma). Where applicable, the following drugs were used: 156 Rapamycin (Sigma, final concentration 400 nM), Torin-1 (Tocris Bioscience, final 157 concentration 10 nM), Wortmannin (Sigma, final concentration 400 nM). Stock solutions 158 were in dimethyl sulphoxide (DMSO). 159 Viruses and virus infections 160 Wild-type SFV stocks were generated from the SFV4 infectious clone (pSP6-SFV4) as 161 described previously (29). SFV–βgal was derived from the plasmid pSFV-b7lacZ (30) 162 and packaged (31). The SFV mutants with the deletions in nsP3 (Fig. 4A, 7A) were 163 constructed and produced from the infectious plasmid pCMV-SFV4 (32): Δ50 (33), 164 ΔP(1+2) (34)—here referred to as ΔP—, Δ789 (35), Δ26 and Δ26-4S-4A (33, 36), and 165 Δ24 (generated by inverse PCR, sequenced and cloned into the Bsu36I and XhoI sites of 166 the pSFV1 replicon, and further cloned into pCMV-SFV4 using Bsu36I and NotI sites). 167 CHIKV-wt was derived from the wild-type infectious clone CHIKV LR2006-OPY1 (37), 168

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 9 which is also the basis for CHIKV-Δ5 (P3del5, (38)). For generation of SFV/CHIKV 169 chimeras, the hyperphosphorylated/acidic domains or the hypervariable domains of 170 nsP3 were swapped between SFV and CHIKV: nsP3 of SFV/CHIKV5 comprises amino 171 acid residues 1–318 of the SFV protein, followed by residues 323–384 of CHIKV-nsP3 172 and 369–482 of SFV-nsP3; the nsP3 of CHIKV/SFV50 consists of residues 1–322 of 173 CHIKV-nsP3 + 319–368 of SFV-nsP3 + 385–530 of CHIKV-nsP3. The nsP3 sequence of 174 SFV/CHIKV-HVD comprises residues 1–322 of SFV-nsP3 followed by 323–530 of CHIKV-175 nsP3; nsP3 of CHIKV/SFV-HVD encompasses residues 1–322 of CHIKV-nsP3 + 323–482 176 of SFV-nsP3. The corresponding DNA sequences were obtained by gene synthesis 177 (GenScript USA Inc.) and subcloned into pCMV-SFV4 (SFV/CHIKV5, SFV/CHIKV-HVD) or 178 pCMV-CHIKV-ICRES (CHIKV/SFV50, CHIKV/SFV-HVD) using Bsu36I and Not I or SanDI 179 and AgeI sites, respectively. All viruses were rescued and propagated on BHK cells. 180 Titres were determined by plaque assay on BHK cells; infectivity of SFV–βgal was 181 assessed by immunofluorescence. Trans-replication assays were performed as 182 described (39), except that the constructs had a CMV promoter and the replication-183 competent template contained a Gaussia luciferase (Gluc) reporter under control of 184 subgenomic promoter; details of the system will be published elsewhere. Replication 185 efficiency was estimated by comparing activity of Gluc produced in the presence of 186 constructs expressing the replicase under study. The efficiency of infectious virus rescue 187 was determined on BHK cells (40). 188 For experimental infection, virus suspensions were diluted in infection medium (DMEM 189 with 0.2% bovine serum albumin [BSA] and 20 mM HEPES) and added to the cells at the 190 indicated MOI for 1 h. PBS was used as a diluent in the experiments shown in Fig. 2A/B. 191 After infection, cells were washed with PBS and supplemented with growth medium or 192

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 10 EBSS (for starvation experiments) and typically kept at cell culture conditions for 193 another 7 h (if not indicated otherwise) and processed for Western blot or 194 immunofluorescence as described below. 195 Cloning and transfection 196 Expression plasmids were cloned by standard molecular biology techniques. The nsP3-197 coding sequence of SFV and CHIKV was amplified by polymerase chain reaction (PCR) 198 using pCMV-SFV4 (wt and mutations thereof) and pCMV-CHIKV-ICRES (CHIKV LR2006-199 OPY1 and mutations thereof), respectively, as templates and oligonucleotide primers 200 (Eurofins) containing restriction sites for cloning. For generation of Myr-Pal–nsP3 201 constructs, the sequences encoding the myristoylation and palmitoylation signals of 202 murine Lyn kinase (amino acids MGCIKSKRKDNLNDDE) and the C-terminal FLAG tag 203 (amino acids DYKDDDDK) were added by PCR. The PCR products were subcloned into 204 pcDNA3.1(−) using NheI and BsrGI (New England BioLabs). Plasmids were checked by 205 sequencing (Eurofins) prior to transient transfection using Lipofectamine 2000 (Life 206 Technologies) according to the manufacturer’s instructions. 207 Cell lysis, SDS-PAGE, Western blot 208 After treatment and washing with PBS, cells in 6-well plates were lysed with 250 µL lysis 209 buffer (20 mM HEPES pH 7.4, 110 mM potassium acetate, 2 mM magnesium chloride, 210 0.1% (v/v) Tween-20, 1% (v/v) Triton X-100, 0.5% (w/v) sodium deoxycholate and 211 500 mM sodium chloride (41), supplemented with Complete protease inhibitor and 212 PhosSTOP phosphatase inhibitor cocktails (Roche)) on ice. Lysates were cleared by 213 centrifugation at 20,000× g for 10 min at 4 °C, mixed with 4× reducing NuPAGE lithium 214 dodecyl sulphate (LDS) sample buffer (Life Technologies) and heated (95 °C, 5 min). 215 Samples were electrophoresed using precast Novex NuPAGE Bis-Tris (4–12% 216

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 11 polyacrylamide) gels (Life Technologies) and subsequently transferred onto Hybond P 217 polyvinylidene difluoride (PVDF) membranes (GE Healthcare). Membranes were 218 blocked with either 3% BSA in PBS with 0.1% (v/v) Tween-20 (PBST) or 5% skim milk 219 powder in Tris-buffered saline with 0.1% (v/v) Tween-20 (TBST), followed by 220 incubation with primary antibodies as listed below (16 h, 4 °C), washing with either 221 PBST or TBST, incubation with a horseradish peroxidase-coupled secondary antibody 222 (Sigma, 1 h, room temperature), washing and detection by enhanced chemiluminescence 223 (ECL, Pierce) using Hyperfilm chemiluminescence films (GE Healthcare) and a Curix 60 224 film developer (AGFA). The following antibodies were used, following the 225 manufacturers’ instructions: rabbit-anti-phospho-Akt (S473) (Cell Signaling #4060), 226 rabbit-anti-phospho-Akt (T308) (Cell Signaling #2965), rabbit-anti-total-Akt (Cell 227 Signaling #4691), rabbit-anti-phospho-4EBP1 (T37/46) (Cell Signaling #2855), rabbit-228 anti-total-4EBP1 (Cell Signaling #9644), rabbit-anti-phospho-S6 (Cell Signaling #5364), 229 mouse-anti-total-S6 (Cell Signaling #2317), goat-anti-actin (Santa Cruz #1616), mouse-230 anti-SFV-nsP2 ((42), cocktail of 2H3, 2C7 and 3B5, 1:1,000 each). Rabbit antisera 231 directed against SFV-nsP3 (43), CHIKV-nsP3 (44) and CHIKV-nsP2 (45) were generated 232 in the Merits lab and used at 1:5,000. All Western blot data are representative of at least 233 three independent experiments. Densitometry was performed on scanned films using 234 ImageJ; phospho-Akt (S473) signal intensities were normalised to the corresponding 235 total-Akt signals. 236 Immunofluorescence and microscopy 237 For assessment of Akt phosphorylation, cells were seeded on coverslips in 12-well plates 238 and kept in serum-free medium for 32 h before and during treatment to reduce 239 background activation of Akt signalling. After treatment, cells were washed with PBS, 240

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 12 fixed with 3.7% formaldehyde in PBS for 15 min at room temperature and subsequently 241 treated with 5% horse serum (Sigma) in PBS + 0.3% Triton X-100 for 90 min, followed 242 by incubation with primary antibody cocktail for 16 h at 4 °C (rabbit-anti-phospho-Akt 243 (S473, Cell Signaling #4060, 1:100) and either mouse-anti-SFV-nsP1 ((28), 1:300) or 244 mouse-anti-FLAG tag M2 (Sigma #F1804, 1:500), in PBS + 0.3% Triton X-100 + 1% BSA) 245 Then, coverslips were washed with PBS and treated with Alexa488-conjugated donkey-246 anti-rabbit (1:200) and Alexa555-conjugated donkey-anti-mouse-IgG (1:1,000) 247 antibodies (Molecular Probes) plus Hoechst 33258 (Life Technologies, 1 µg/mL) in PBS 248 + 0.3% Triton X-100 + 1% BSA (30 min, room temperature). Coverslips were then 249 washed, mounted on glass slides using Vinol and imaged by epifluorescence microscopy 250 with a Leica DMRB microscope using a 63× objective, a Hamamatsu CCD camera and 251 HiPic software. Images were processed using Adobe Photoshop. 252 For visualisation of replication complexes after infection, cells (grown on coverslips in 253 12-well plates) were washed with PBS, fixed with 3.7% (v/v) formaldehyde in PBS for 254 10 min at room temperature, followed by permeabilisation with methanol at −20 °C for 255 10 min, blocking with 5% horse serum (Sigma) in PBS for 16 h at 4 °C and treatment 256 with primary antibody cocktail in blocking solution for 5 h at room temperature. Mouse-257 anti-dsRNA (English & Scientific Consulting, 1:200) and either rabbit-anti-SFV-nsP3 258 ((43), 1:300) or rabbit-anti-CHIKV-nsP3 ((44), 1:300) were used. Subsequently, 259 coverslips were washed with PBS and incubated for 30 min with Alexa488-conjugated 260 donkey-anti-rabbit (1:200), Alexa555-conjugated donkey-anti-mouse-IgG (1:1,000) 261 (Molecular Probes) and the DNA dye DRAQ5 (Biostatus, 1:1,000). Coverslips were then 262 washed, mounted and imaged by confocal laser-scanning microscopy using a Leica TCS 263

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 13 SP5 X microscope equipped with a supercontinuum pulsed white laser. Microscopy 264 samples were typically processed in a double-blinded manner to avoid bias. 265 266

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 14 Results 267 Characterisation of SFV-induced PI3K–Akt–mTOR activation 268 First, we determined whether and how the PI3K–Akt–mTOR pathway is affected by 269 infection of cells with SFV, and to what extent this is influenced by nutrient availability, 270 generally a prerequisite for this pro-survival pathway. We mock-infected or infected 271 human osteosarcoma (HOS) cells with SFV-wildtype (wt; strain SFV4) at a multiplicity of 272 infection (MOI) of 10 for 1 h. Cells were then either kept in standard growth medium or 273 starved for all nutrients and growth factors. Cell lysates were prepared 8 h post-274 infection (p.i.) and were subjected to Western blot analysis to detect phosphorylation of 275 Akt at either activation residue (threonine-308, serine-473) as well as phosphorylation 276 of 4EBP1 and the ribosomal protein S6, downstream readouts for mTORC1 activity. 277 Infection was assessed by probing for nsP3. These analyses (Fig. 1B) revealed that the 278 phosphorylation status of Akt was strongly increased in infected cells when compared to 279 the Akt phosphorylation levels in non-infected cells (very low under standard 280 conditions, undetectable under starvation conditions). Akt phosphorylation upon 281 infection was seen both in non-starved and starved cells, pointing towards an active, 282 infection-induced stimulation. Concomitantly, the mTOR targets S6 and 4EBP1 were 283 phosphorylated in infected cells, not only under normal growth conditions (where 284 mTOR is active), but also under starvation conditions (where mTOR is normally 285 inactive). 286 In a complementary approach, we assessed S6 phosphorylation by immunofluorescence 287 8 h p.i. (Fig. 1C, co-staining for phospho-S6 [green] and viral nsP1 [red]) and obtained 288 comparable results: S6 was phosphorylated in non-starved cells irrespective of infection 289

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 15 and was not phosphorylated in starved cells except those that had been infected (Fig. 1C, 290 rightmost panel). Thus, mTOR was kept active by infection, not only under normal 291 growth conditions, but even in starved cells. The stimulation of the PI3K–Akt–mTOR 292 pathway by infection was not limited to HOS cells as we obtained similar results also in 293 other mammalian cell types such as immortalised mouse embryonal fibroblasts (MEF, 294 Fig. 1D) and baby hamster kidney (BHK, Fig. 1E) cells. We therefore conclude that PI3K–295 Akt–mTOR activation is a common feature of SFV infection. 296 Next, we performed a time-course experiment to assess the kinetics of PI3K–Akt–mTOR 297 activation upon infection. HOS cells were infected with SFV at MOI 10 or mock-infected 298 for 1 h, employing phosphate-buffered saline (PBS) for dilution to prevent any activation 299 of mTOR by nutrients in the infection medium. Cells were subsequently kept under 300 standard or starvation conditions; cell lysates were prepared at different times after 301 infection and assessed by Western blot (Fig. 2A/B). Akt and S6 phosphorylation were 302 reduced after 1 h of starvation (mock-infection or infection in PBS, compare lanes 2 and 303 6 to lane 1 in Fig. 2A) and reappeared after 1 h of incubation in nutrient-rich medium 304 (Fig. 2A, lane 3). Levels of S6 phosphorylation subsequently stayed high throughout the 305 timeframe of analysis (up to 8 h) in both non-infected and infected cells (lanes 3–5 and 306 7–9, respectively). Akt activation levels in non-infected cells reached starting levels 307 (compare lanes 4 and 5 to lane 1), with a transient boost shortly after resupply of 308 nutrient-rich medium (lane 3). In contrast, Akt activation levels increased gradually in 309 infected cells, exceeding the starting levels (compare lanes 8 and 9 to lane 1). When the 310 same experiment was performed under conditions of starvation (Fig. 2B), significant Akt 311 activation was seen 4 h after infection (lane 8) and increased further up to 8 h p.i. 312 (lane 9), while staying undetectably low in non-infected, starved cells (lanes 2–5). S6 313

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 16 phosphorylation faded out in the course of starvation in non-infected cells (lanes 2–5), 314 but was restored to full levels in infected cells in spite of starvation (compare lane 9 to 315 lane 1). These data indicate that the PI3K–Akt–mTOR pathway was actively stimulated 316 by infection, even in starved cells, starting as early as 4 h p.i. 317 In order to better understand the mode of virus-induced PI3K–Akt–mTOR activation, we 318 employed specific inhibitors of the pathway. Cells were infected for 1 h with SFV or 319 mock-infected and were then kept in standard medium or under conditions that 320 differentially block mTOR: starvation (general deactivation of the pathway by nutrient 321 withdrawal), Rapamycin (an inhibitor of mTORC1) or Torin-1 (an inhibitor of mTORC1 322 as well as another mTOR complex, mTORC2, the kinase for Akt activation residue S473). 323 Cell lysates were prepared 8 h p.i. and analysed by Western blot (Fig. 2C). In non-324 infected cells, S6 phosphorylation, the downstream readout for mTOR activity, was not 325 detected upon each of these treatments (lanes 2, 3 and 4), which confirms their efficacy. 326 In infected cells, S6 phosphorylation was sustained in untreated and in starved cells 327 (lanes 5 and 6) but not upon application of either Rapamycin (lane 7) or Torin (lane 8), 328 indicating that the infection could not overcome mTOR inhibition by these drugs and 329 thus acted upstream of mTOR itself. When we assessed Akt phosphorylation, we 330 observed that this modification was prevented by starvation (absence of growth factors, 331 lane 2) and Torin-1 (inhibition of mTORC2, lane 4) but not Rapamycin (lane 3), while in 332 infected cells (lanes 5–8), very strong Akt activation was seen under all conditions 333 except for Torin-1 treatment (lane 8). This indicates that the virus-induced stimulation 334 of mTOR acts upstream of or at the level of Akt activation. 335 We then applied the PI3K inhibitor Wortmannin for further characterization of the 336 virus-induced Akt activation. After mock-infection or infection at MOI 10 for 1 h, cells 337

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 17 were kept in growth or starvation medium, with or without Wortmannin at a high 338 concentration (400 nM). Cell lysates were prepared 8 h p.i. and analysed by Western 339 blot (Fig. 2D). The high level of Akt phosphorylation seen in infected cells (lane 5) was 340 reduced—but not abolished—upon treatment with Wortmannin (lane 7). S6 341 phosphorylation was not affected by Wortmannin (owing to intact mTOR stimulation 342 through other pathways). Phosphorylation of Akt and S6 was not detected when 343 Wortmannin treatment was combined with starvation (lane 8). 344 In summary, the data in Fig. 1 and 2 show that infection of cells with SFV led to strong 345 and permanent, partially Wortmannin-insensitive activation of Akt. This activation 346 occurred even under starvation conditions and resulted in sustained activity of mTOR. 347 Viral requirements for SFV-induced PI3K–Akt–mTOR activation 348 Next, we investigated the viral requirements for PI3K–Akt–mTOR activation. To evaluate 349 whether the structural proteins play a role for this activation, we employed SFV–βgal, an 350 SFV replicon in which the open reading frame for the structural proteins had been 351 replaced by the gene encoding β-galactosidase. Western blot analysis of cell lysates 352 prepared 8 h p.i. showed that Akt was phosphorylated upon infection with SFV–βgal 353 virus replicon particles to a similar extent as observed with SFV-wt (Fig. 3A), indicating 354 that expression of the structural proteins is dispensable for SFV-induced Akt activation. 355 In addition, we employed immunofluorescence for phosphorylated Akt to assess the 356 activation of Akt at a single-cell level and corroborated that Akt was phosphorylated 357 only in cells that had been infected with SFV-wt or SFV–βgal, but not in the non-infected 358 neighbouring cells (Fig. 3B). In addition, we transfected cells with a plasmid encoding 359 P123 to determine whether Akt activation depends on the presence and action of the 360 RNA-dependent RNA polymerase nsP4, which is lacking in P123. We employed the 361

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 18 immunofluorescence assay and detected that Akt was considerably phosphorylated in 362 the cells expressing P123 (Fig. 3B, rightmost panel). Thus, virus-induced Akt activation 363 in cells could be narrowed down to the presence of the replication complex proteins 364 nsP1, 2, and 3, but did not require replication, the presence of nsP4 or the structural 365 proteins. 366 It has been proposed that nsP3 mediates the internalisation of SFV replication 367 complexes from the plasma membrane (28), a process that is linked to the PI3K–Akt 368 pathway (16). We therefore asked whether mutations of nsP3 in the viral context would 369 affect the virus-induced Akt activation. We hypothesised that the C-terminal 370 hypervariable region of nsP3, a scaffold for interaction with cellular factors, could be 371 relevant for PI3K–Akt activation and focused on mutations in this region. We employed 372 virus mutants characterised earlier, encoding deletion variants of nsP3 as depicted in 373 Fig. 4A: In SFV-Δ50, 50 residues (319–368) comprising the hyperphosphorylated/acidic 374 region of nsP3 are deleted (33); SFV-ΔP does not contain the proline-rich regions 375 (residues 408–440, (34)); SFV-Δ789 lacks the G3BP-interaction domain (residues 449–376 472 (35)). We infected BHK cells with SFV-wt or either of these viruses and assessed Akt 377 activation at 8 h p.i. by Western blot (Fig. 4B) and phospho-Akt immunofluorescence 378 (Fig. 4C). While the extent of Akt phosphorylation was similar for SFV-wt as well as SFV-379 ΔP and SFV-Δ789, infection with SFV-Δ50 led to considerably weaker phospho-Akt 380 signals. All the mutant viruses were similarly attenuated with respect to the wildtype (as 381 judged from the intensities of the nsP3 signals). The result indicates that the stretch of 382 nsP3 deleted in SFV-Δ50, but none of the other nsP3 regions under study, substantially 383 contributed to SFV-induced Akt activation. The weak Akt activation by SFV-Δ50 was 384 completely blocked by application of Wortmannin (Fig. 4D), and exhibited slower 385

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 19 kinetics than in the case of SFV-wt: The Akt phosphorylation in SFV-Δ50-infected cells 386 was at background levels at 4 h p.i., a time where Akt was already strongly activated by 387 infection with SFV-wt (Fig. 4E). 388 After observing that the hyperphosphorylated/acidic region of SFV-nsP3 is relevant for 389 Akt activation, we next asked whether expression of nsP3 alone is enough to induce Akt 390 phosphorylation. To test this, we cloned expression vectors for nsP3 (wt and Δ50) with a 391 C-terminal FLAG tag, transfected HOS cells and performed immunofluorescence analysis 392 with anti-phospho-Akt (S473) and anti-FLAG-tag antibodies. We could not detect Akt 393 phosphorylation above background levels in cells expressing nsP3–FLAG (neither wt nor 394 Δ50), indicating that nsP3 expression did not detectably activate Akt (Fig. 5A–C). 395 There is a significant difference in subcellular localisation between nsP3 expressed on 396 its own (cytosolic) and nsP3 in association with the replication complex (stably 397 membrane-associated through interaction with nsP1, which binds to membranes (28, 398 46)). Since P123 activated Akt (Fig. 3B), we postulated that nsP3 needs to be membrane-399 associated to mediate Akt phosphorylation, a process that requires plasma membrane 400 recruitment of Akt (47). We thus engineered a membrane-attached version of nsP3 by 401 fusing the N-terminal myristoylation and palmitoylation signal of Lyn kinase (Myr-Pal) 402 to the N-terminus of nsP3–FLAG in the context of the expression vector. This signal leads 403 to covalent modification of the protein with fatty acids (myristic acid at glycine-2, 404 palmitic acid at cysteine-3) with the potential to target the protein to the plasma 405 membrane (48). When we expressed Myr-Pal–nsP3–FLAG in HOS cells, the protein was 406 typically present at intracellular vesicles; some cells exhibited a concentration of the 407 signal at the cell periphery (Fig. 5D). When we assessed Akt phosphorylation by 408 immunofluorescence, cells expressing membrane-attached Myr-Pal–nsP3–FLAG 409

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 20 displayed strong signals for phosphorylated Akt (Fig. 5D); expression of Myr-Pal–nsP3–410 FLAG with the Δ50 mutation however failed to induce phosphorylation of Akt (Fig. 5E). 411 Taken together, the data presented in Fig. 5 show that nsP3, when attached to 412 membranes (thus imitating the situation in the replication complex) activated Akt, 413 dependent on the hyperphosphorylated/acidic region. 414 415 Subcellular localisation of SFV replication complexes 416 In the course of SFV infection, the viral replication complexes (spherules) are formed at 417 the plasma membrane and subsequently internalised in a process that depends on 418 PI3K–Akt signalling (16). We hence hypothesised that the activation of the PI3K–Akt–419 mTOR pathway upon SFV infection is linked to replication complex dynamics, possibly 420 resulting in spherule internalisation. To assess this, we infected BHK cells with either 421 SFV-wt, SFV-Δ50, -ΔP or Δ789 at MOI 10, fixed cells 8 h p.i. and performed 422 immunofluorescence staining for double-stranded (ds) RNA to detect replication 423 complexes; co-staining for nsP3 was performed to confirm non-structural protein 424 expression. Representative confocal micrographs are shown in Fig. 6. In cells infected 425 with SFV-wt, dsRNA-positive foci were present throughout the cytosol without 426 prominent staining of the cell surface (Fig. 6A). Application of Wortmannin (for 7 h, 427 added 1 h p.i.) resulted in accumulation of replication complexes at the cell periphery 428 (Fig. 6B), which is indicative of a block in spherule internalisation and corroborates 429 earlier results (16). Administration of either Torin or Rapamycin did not affect 430 replication complex internalisation (Fig. 6C/D). When cells were infected with SFV-Δ50, 431 which does not induce strong PI3K–Akt–mTOR activation (Fig. 4), the subcellular 432 localisation of replication complexes (Fig. 6E) was noticeably different from the 433

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 21 appearance in SFV-wt-infected cells (Fig. 6A), but strongly resembled the situation with 434 SFV-wt + Wortmannin (Fig. 6B). Thus, replication complexes accumulated at the cell 435 periphery in most cells infected with SFV-Δ50. Conversely, the localisation of replication 436 complexes upon infection with SFV-ΔP (Fig. 6F) or -Δ789 (Fig. 6G) was indistinguishable 437 from SFV-wt, thus indicating unperturbed spherule internalisation. The nsP3 signals 438 were either punctate, sometimes co-localising with dsRNA and thus representing 439 replication complex-associated nsP3, or diffusely spread in the cytosol. This is in 440 accordance with the notion that nsP3 is engaged in various complexes (49). 441 In summary, the results presented in Fig. 6 provide a link between strong PI3K–Akt–442 mTOR activation and replication complex internalisation and show that both functions 443 are disrupted by the Δ50 mutation in nsP3, where the hyperphosphorylated/acidic 444 region of nsP3 is lacking. 445 Next, we endeavoured to decipher whether a specific part of the 446 hyperphosphorylated/acidic region is responsible for the effect on PI3K–Akt–mTOR 447 signalling and replication complex dynamics. To this end, we employed recombinant 448 viruses with partial deletions in this domain of nsP3 (residues 319–368). These 449 deletions comprise either the 24 N-terminal residues (Δ24, i.e., 319–342) or the 26 C-450 terminal residues (Δ26, 343–368) of the region (see Fig. 7A, (33)). Also, to assess the 451 influence of phosphorylation, we made use of the mutant Δ26-4S-4A, which carries the 452 Δ26 deletion and an additional exchange of four serines—the major remaining 453 phosphorylation sites—by alanines, a set of mutations that eliminates detectable 454 phosphorylation of nsP3 (36). We infected BHK cells with either of these recombinant 455 viruses and assessed the subcellular localisation of replication complexes by 456 immunofluorescence staining for dsRNA and SFV-nsP3 8 h after infection (Fig. 7B–D). 457

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 22 For each of these viruses, we observed that dsRNA-positive foci were localised in the 458 cytosol, very similar to the situation for SFV-wt (Fig. 6A); no prominent accumulation of 459 replication complexes at the cell periphery (as in the case of SFV-Δ50, Fig. 6E), was seen. 460 In complementary experiments, BHK cells were infected with SFV-wt, -Δ50 or one of the 461 partial deletion mutants at MOI 10 to assess Akt phosphorylation at 8 h p.i. by Western 462 blot (Fig. 7E). As before, we observed high levels of Akt phosphorylation in SFV-wt-463 infected cells, which were clearly reduced for SFV-Δ50. Conversely, infection with SFV-464 Δ24, -Δ26 and also -Δ26-4S-4A induced Akt phosphorylation to comparable levels as 465 SFV-wt, higher than in the case of SFV-Δ50. Thus, the capacity to strongly activate Akt 466 was not lost with any of the partial deletions within the hyperphosphorylated/acidic 467 region of nsP3. 468 Taken together, the partial deletions within the hyperphosphorylated/acidic region 469 neither prevented strong PI3K–Akt–mTOR activation nor replication complex 470 internalisation. This implies that individual portions of this region are sufficient to 471 support these functions, and that phosphorylation is not essential. 472 473 PI3K–Akt–mTOR activation and replication complex dynamics in CHIKV 474 infection 475 Next, we assessed whether the closely related CHIKV also activates the PI3K–Akt–mTOR 476 pathway in a similar fashion as SFV. To this end, we infected BHK cells with CHIKV-wt at 477 MOI 10 and prepared cells lysates at different times after infection (up to 16 h p.i., before 478 the onset of extensive cell death), and determined Akt phosphorylation by Western blot 479 (Fig. 8A). The extent of Akt phosphorylation increased within 8 h after infection with 480

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 23 CHIKV and was not further augmented by 16 h p.i. Hence, Akt was activated by CHIKV 481 infection as well; strikingly however, the extent of Akt activation was much weaker than 482 the intensity seen 8 h after infection with SFV-wt, performed in parallel under the same 483 conditions (Fig. 8A). 484 To assess whether CHIKV-induced Akt activation requires the same domain in nsP3 as 485 seen for SFV, we made use of a CHIKV mutant termed CHIKV-Δ5, which carries a 486 deletion in nsP3 comparable to the SFV-nsP3-Δ50 deletion (38). The respective 487 stretches of nsP3 are equivalent in position but not sequence. 488 When BHK cells were infected with the CHIKV-Δ5 mutant and assessed by Western blot, 489 we observed time-dependent phosphorylation of Akt, which reached comparable levels 490 to the Akt activation seen with CHIKV-wt (Fig. 8A). Further, we noted that the CHIKV-491 induced Akt stimulation in BHK cells was completely blocked in the presence of 492 Wortmannin, both for CHIKV-wt and CHIKV-Δ5 (Fig. 8B). This is reminiscent of the 493 situation with SFV-Δ50 (Fig. 4D), but unlike that with SFV-wt, where Akt stimulation 494 was only partially Wortmannin-sensitive (Fig. 2D). When we assessed whether CHIKV-495 wt is capable of activating the PI3K–Akt–mTOR pathway also under starvation 496 conditions, as seen above for SFV (Fig. 1), we found that neither Akt nor S6 were 497 strongly phosphorylated in CHIKV-infected, starved cells, indicating that the CHIKV-498 induced PI3K–Akt–mTOR activation is sensitive to starvation (Fig. 8C). Taken together, 499 these results imply that CHIKV infection led to only moderate activation of the PI3K–500 Akt–mTOR pathway, which was prevented by Wortmannin or starvation and not 501 dependent on the region of nsP3 that corresponds to the hyperphosphorylated/acidic 502 domain of SFV-nsP3. 503

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 24 Next, we assessed the subcellular localisation of replication complexes in BHK cells 8 h 504 after infection with CHIKV-wt or -Δ5 at MOI 10 by immunofluorescence staining for 505 dsRNA and CHIKV-nsP3 and confocal microscopy (Fig. 8D–F). We noted a difference in 506 nsP3 localisation between CHIKV-wt and -Δ5, probably reflecting engagement of nsP3 in 507 various complexes in cells (49). The dsRNA-positive replication complexes were 508 markedly accumulated at the cell periphery in most cells, both for CHIKV-wt and -Δ5, at 509 8 h p.i. (Fig. 8E/F) and also at later times (not shown). This is a clear difference to the 510 typical replication complex localisation upon infection with SFV-wt and resembles the 511 situation for SFV-Δ50 (compare with Fig. 6). These results suggest that the 512 hyperphosphorylated/acidic region of SFV but not CHIKV mediates strong PI3K–Akt–513 mTOR activation and replication complex internalisation. 514 515 PI3K–Akt–mTOR activation and replication complex dynamics of chimeric 516 SFV/CHIKV viruses 517 We next asked whether the hyperphosphorylated/acidic domain can be functionally 518 exchanged between SFV and CHIKV, i.e. whether the SFV domain induces the “SFV 519 phenotype” (strong PI3K–Akt–mTOR activation and replication complex internalisation) 520 also in the context of CHIKV. To this end, we generated recombinant SFV and CHIKV 521 genomes in which the residues deleted in SFV-Δ50 were replaced by the sequence 522 deleted in CHIKV-Δ5 and vice versa, giving rise to CHIKV/SFV50 (CHIKV with the SFV 523 domain) and SFV/CHIKV5 (SFV with the CHIKV domain), respectively. The efficiency of 524 infectious virus rescue was drastically reduced compared to the respective wildtype 525 viruses. In particular, rescue efficiency of CHIKV/SFV50 was reduced by at least 10,000-526 fold; correspondingly, virus stocks had very low initial titres and acquired secondary 527

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 25 (adaptive) mutations upon propagation in BHK cells. Interestingly, these mutations did 528 not reside in the swapped region or in the rest of the C-terminal region of nsP3; instead, 529 a compensatory methionine-to-isoleucine mutation was detected in the zinc-binding 530 domain of nsP3 (position 1552 in P1234, where the corresponding SFV-wt residue is 531 leucine). Furthermore, another compensatory mutation occurred even outside of nsP3 532 (N1318I in P1234; C-terminal region of nsP2). The ineffective rescue and low infectivity 533 of the CHIKV/SFV50 virus most likely resulted from severely compromised replicase 534 activity in the recombinant virus: we could not detect any replicase activity above the 535 background for the CHIKV/SFV50 replicase in a trans-replication system. The activity of 536 the SFV/CHIKV5 replicase was also severely diminished (reduced to approximately 1% 537 as compared to SFV-wt); consequently, the rescue of infectious SFV/CHIKV5 virus was 538 reduced by at least 1,000-fold. Hence, while deletion of the hyperphosphorylated/acidic 539 region in SFV- and CHIKV-nsP3 was well tolerated (SFV-Δ50 and CHIKV-Δ5), swap of the 540 domains was not. This precluded biochemical experiments. 541 We employed the recombinant CHIKV/SFV50 and SFV/CHIKV5 viruses to determine the 542 subcellular localisation of replication complexes in infected BHK cells at 8 h p.i. Staining 543 for dsRNA revealed that replication complexes were predominantly localised at the cell 544 periphery, for SFV/CHIKV5 (Fig. 9A) as well as CHIKV/SFV50 (Fig. 9B). Thus, the 545 hyperphosphorylated/acidic domain of CHIKV could not functionally replace the domain 546 in SFV, and the SFV domain did not mediate replication complex internalisation when 547 introduced into CHIKV. 548 We also addressed the activation of the PI3K–Akt–mTOR pathway. To this end, we 549 generated expression constructs for membrane-associated nsP3 (Myr-Pal–nsP3–FLAG) 550 for use in phospho-Akt immunofluorescence experiments (see Fig. 5), with the nsP3 551

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 26 sequences of CHIKV-wt as well as the chimeric variants. Note that adaptive mutations do 552 not occur in this virus replication-independent approach. These constructs were 553 expressed in HOS cells and assessed by immunofluorescence microscopy for phospho-554 Akt. Contrary to the result with Myr-Pal–SFV-nsP3–FLAG (Fig. 9C, see also Fig. 5D), we 555 did not detect significant phosphorylation of Akt above background levels in cells 556 expressing Myr-Pal–nsP3–FLAG in the CHIKV, CHIKV/SFV50 or SFV/CHIKV5 variants 557 (Fig. 9D–F). Taken together, the results in Fig. 9 indicate that the 50 residues of the 558 hyperphosphorylated/acidic domain of SFV-nsP3 (deleted in SFV-Δ50) are necessary 559 but not sufficient for replication complex internalisation and Akt activation. 560 Our approach of domain swap likely disturbed correct folding of nsP3 and accurate 561 assembly of replication complexes, probably due to sequence incompatibility upon 562 insertion. We supposed that the risk of such adverse effects would be reduced if the 563 entire C-terminal hypervariable domain of nsP3 were swapped between CHIKV and SFV. 564 Hence, we generated a recombinant CHIKV genome where the sequence encoding the 565 hypervariable domain of nsP3 (starting at residue 323, where the high sequence 566 conservation ends) was replaced by the corresponding sequence of SFV, yielding 567 CHIKV/SFV-HVD, and a recombinant SFV genome where the C-terminal tail of nsP3 was 568 analogously replaced by that of CHIKV (SFV/CHIKV-HVD). Using this approach, 569 infectious chimeric viruses could be rescued efficiently. Subsequently, BHK cells were 570 infected with these viruses to assess the subcellular localisation of replication complexes 571 by immunofluorescence staining for dsRNA at 8 h p.i. Confocal microscopy showed that 572 the replication complexes of CHIKV/SFV-HVD (Fig. 10B) were efficiently internalised 573 from the cell periphery, contrary to the situation with CHIKV-wt (Fig. 10A) and 574 resembling the phenotype of SFV-wt (Fig. 10D). Conversely, replication complexes of 575

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 27 SFV/CHIKV-HVD (Fig. 10C) were predominantly localised at the cell periphery, similarly 576 to those of CHIKV-wt and unlike the situation with SFV-wt. Hence, the C-terminal 577 domain of nsP3 determined the subcellular localisation of replication complexes: the 578 domain of SFV-nsP3 induced efficient replication complex internalisation in the context 579 of both SFV and CHIKV, while the corresponding domain of CHIKV largely failed to 580 promote replication complex internalisation in CHIKV as well as SFV. 581 We then analysed activation of the PI3K–Akt–mTOR pathway and asked whether the 582 presence of the hypervariable domain of SFV-nsP3 confers the “SFV phenotype” 583 (strongly activated Akt signalling) to the recombinant CHIKV/SFV-HVD and whether this 584 phenotype is lost in SFV/CHIKV-HVD. BHK cells were mock-infected or infected with 585 either of the wildtype or chimeric viruses at MOI 10 and lysed at 8 h p.i., followed by 586 SDS-PAGE and Western blot (Fig. 10E). Efficient infection was verified with antisera 587 against nsP2 of CHIKV and SFV. Akt phosphorylation was significantly stronger in cells 588 infected with CHIKV/SFV-HVD than in the case of CHIKV-wt, reaching similar levels as 589 SFV-wt. Conversely, the Akt phosphorylation levels upon infection with SFV/CHIKV-HVD 590 were clearly lower than for SFV-wt. This is evidence that the hypervariable domain of 591 SFV, but not CHIKV, boosted PI3K–Akt–mTOR signalling in the context of both SFV and 592 CHIKV. 593 In a complementary approach, we generated expression constructs for membrane-594 associated nsP3 (Myr-Pal–nsP3–FLAG) and performed phospho-Akt 595 immunofluorescence in transfected HOS cells as above (Fig. 5 and 9). Contrary to the 596 CHIKV version of this construct (see Fig. 9F), expression of membrane-attached CHIKV-597 nsP3 with the hypervariable domain of SFV induced clearly detectable levels of Akt 598 phosphorylation (Fig. 10F), like the analogous SFV-nsP3 construct (cf. Fig. 5D, 9C). 599

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 28 Conversely, no Akt activation above background levels was visible for membrane-600 attached chimeric SFV/CHIKV-HVD-nsP3 (Fig. 10G). Thus, the hypervariable C-terminal 601 tail of SFV-nsP3 had the capacity to confer strong Akt activation in the context of 602 membrane-attached nsP3 of both SFV and CHIKV. Finally, we assessed whether the 603 hypervariable domain of SFV-nsP3 (residues 319–482) is sufficient to induce Akt 604 phosphorylation when equipped with an N-terminal myristoylation and palmitoylation 605 signal and a C-terminal FLAG tag (Myr-Pal–SFV-HVD–FLAG). When we expressed this 606 construct in HOS cells and performed phospho-Akt immunofluorescence, we observed 607 prominent Akt activation in most transfected cells (Fig. 10H). This indicates that the 608 essential elements of the PI3K–Akt–mTOR activation domain reside in the C-terminal 609 tail of SFV-nsP3. 610 611

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 29 Discussion 612 In this work, we describe that infection of a variety of cells with SFV-wt led to a very 613 strong and persistent phosphorylation and thus activation of the pivotal pro-survival 614 kinase Akt, leading to sustained activation of the downstream metabolic regulator mTOR 615 (Fig. 1 and 2). This was very likely due to activation of the Akt pathway rather than 616 inhibition of a negative regulator or mere stabilisation of activated components since 617 SFV infection could overcome the inactivation of Akt and mTOR by starvation (see time 618 course in Fig. 2B). 619 Expression of membrane-bound nsP3 in the absence of other viral proteins was 620 sufficient to activate Akt, as demonstrated by using nsP3 with an N-terminal membrane 621 anchor (Myr-Pal–nsP3–FLAG, Fig. 5), indicating that (plasma) membrane localisation of 622 the viral protein is necessary for Akt activation. During infection, stable membrane 623 association of nsP3 is achieved by interaction with the peripheral membrane protein 624 nsP1 within the replication complex (46, 50, 51). There are other, non-membraneous 625 nsP3-containing structures in infected cells (49), which are however unlikely to be 626 relevant for Akt activation since membrane anchorage of nsP3 was required. 627 Even though there have been several comprehensive investigations into molecular 628 interactions of Old World alphavirus nsP3 (41, 49, 52), none of these reported an 629 interaction partner of nsP3 known to have a direct function in PI3K–Akt–mTOR 630 signalling. Presumably, the interaction of nsP3 with the Akt signalling module is 631 transient or dynamic. Since proximity to membranes appears to be required (Fig. 5), an 632 intriguing possibility is that the protein mimics a cellular activation feature for the Akt 633 module. Such molecular mimicry might also explain why the SFV-induced Akt activation 634 was not completely blocked by the PI3K inhibitor Wortmannin. 635

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 30 We identified that the hyperphosphorylated/acidic region of nsP3, deleted in the virus 636 mutant SFV-Δ50, is necessary for strong and persistent PI3K–Akt–mTOR activation 637 (Fig. 4). Interestingly, PI3K–Akt–mTOR activation did not require the proline-rich region 638 of nsP3 (deleted in SFV-ΔP), a motif shown to mediate binding to the SH3 domain of 639 amphiphysin (34). This is markedly different to the situation with other viruses such as 640 influenza A virus (IAV), where the protein NS1 activates Akt through the interaction of a 641 proline-rich sequence with the SH3 domain of p85, the regulatory subunit of PI3K (53). 642 Thus, the proline-rich regions of different viral proteins differ in their specificity for 643 cellular proteins containing SH3 domains; specificity is likely conferred by additional 644 adjacent binding motifs. In the case of IAV-NS1, a (presumably phosphorylated) tyrosine 645 interacts with the SH2 domain of p85 (54). The non-structural proteins of SFV however 646 have never been shown to contain phosphorylated tyrosines (55). 647 While our results show that the hyperphosphorylated/acidic region of SFV-nsP3 is the 648 key part for the strong stimulation of the PI3K–Akt pathway (Fig. 4 and 5) and efficient 649 replication complex internalisation (Fig. 6), it remains unclear by what molecular 650 mechanism this is achieved. A distinctive feature of the domain is that it comprises a 651 high density of negative charges: 8 aspartic acid residues as well as 6 threonine and 12 652 serine residues (in SFV4), most of which are phosphorylated (at least in the case of SFV 653 and also SINV) by largely undefined cellular kinases, presumably in a non-regulated 654 manner (33, 55, 56). Tyrosine phosphorylation of nsP3 has never been detected; thus, 655 the region is unlikely to mimic activated growth factor receptors which employ 656 phospho-tyrosine for binding to PI3K.Phosphorylation of nsP3 was not critical for Akt 657 activation and replication complex internalisation since these phenotypes were also 658 observed for the virus mutant SFV-Δ26-4S-4A, where no phosphorylation is detectable 659

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 31 (36). Strong and persistent Akt activation upon SFV infection might require the presence 660 of a certain number or density of negative charges in the structural context of nsP3 (as is 661 the case in nsP3-wt, -Δ24, -Δ26 and also -Δ26-4S-4A, but not -Δ50). 662 In cells infected with SFV-Δ50, the PI3K–Akt–mTOR pathway was activated to a 663 moderate level (clearly above-the-background in Western blot but not above the 664 detection threshold in immunofluorescence). This stimulation was completely blocked 665 by Wortmannin and was thus entirely dependent on PI3K activity. Taken together with 666 the finding that SFV-wt induced much stronger and partially Wortmannin-insensitive 667 Akt activation, we propose that there are two types of alphavirus-induced PI3K–Akt–668 mTOR activation (see Fig. 11): 669 (I) Basal, completely Wortmannin-sensitive (PI3K-dependent) activation upon infection, 670 which is also seen with SFV-Δ50 as well as CHIKV (Fig. 8). 671 (II) Additional strong Akt activation, observed for SFV-wt, but not SFV-Δ50 or CHIKV. 672 This activation is at least partially resistant to starvation or Wortmannin treatment (but 673 not both, see Fig. 2D) and leads to potent and sustained activation of Akt–mTOR 674 signalling. This strong PI3K–Akt–mTOR activation phenotype can be induced in CHIKV 675 by swapping the hypervariable C-terminal domain of nsP3 (Fig. 10). 676 Taken together with the data on the subcellular localisation of replication complexes, it 677 becomes evident that boosted (II) but not basal (I) Akt activation is linked to 678 internalisation of replication complexes from the cell periphery: All viruses that were 679 capable of very strong Akt activation (II) exhibited replication complex internalisation; 680 SFV-Δ50 and CHIKV, which only induced moderate Akt activation (I), did not. Thus, the 681 hyperphosphorylated/acidic region in SFV-nsP3 is required both for the boost in PI3K–682

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 32 Akt–mTOR activation and the internalisation of replication complexes. The 683 corresponding domain of CHIKV-nsP3 did not mediate these phenotypes, neither in 684 CHIKV (Fig. 8) nor when introduced into SFV (Fig. 9 and 10). Conversely, the C-terminal 685 domain of SFV-nsP3 led to strong PI3K–Akt–mTOR activation and efficient replication 686 complex internalisation in the context of both SFV and CHIKV (Fig. 10). 687 The PI3K–Akt–mTOR activation domain is not congruent with the 50 residues deleted in 688 SFV-Δ50. This is evidenced by the finding that partial deletions within this region did not 689 prevent strong PI3K–Akt activation and replication complex internalisation (Fig. 7). 690 Also, the 50 residues did not induce the “SFV phenotype” when transferred into CHIKV-691 nsP3 (Fig. 9), while the complete C-terminal hypervariable region did (Fig. 10). This 692 implies that PI3K–Akt–mTOR activation and replication complex internalisation do not 693 critically depend on any specific amino acids within the 50 residues; several regions 694 inside and outside of this region may act cooperatively and form a (potentially 695 discontinuous) activation feature. Elements downstream of the 50 residues are likely 696 part of the PI3K–Akt activation domain, which is disrupted by the Δ50 deletion. The 697 conserved N-terminal domains of nsP3 (macro domain and zinc-binding region) 698 however do not significantly contribute to the PI3K–Akt activation structure since the 699 hypervariable C-terminal region of SFV-nsP3 alone induced detectable Akt 700 phosphorylation when equipped with a membrane anchor (Fig. 10H). 701 702 The non-structural proteins of alphaviruses are initially targeted to the plasma 703 membrane, where the replication complexes (spherules) are formed (16, 57). It is hence 704 likely that SFV-nsP3 mediates activation of the PI3K–Akt signalling module at this stage, 705

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 33 at the plasma membrane, the general cellular site of PI3K action and Akt activation. 706 Replication complex internalisation would then follow, likely involving additional 707 cellular features such as the cytoskeleton (16). The interdependence of infection-708 induced PI3K–Akt activation and replication complex dynamics is evident from the 709 finding that SFV replication complexes are stalled at the cell periphery by PI3K 710 inhibition (Fig. 6B; (16)). The signal for replication complex internalisation is, however, 711 not downstream of Akt since Torin blocked SFV-induced Akt phosphorylation, but not 712 replication complex internalisation (Fig. 6C). It remains unclear for what purpose the 713 replication complexes of SFV are efficiently internalised, but it is conceivable that this 714 subcellular localisation ensures spatial proximity of the viral replication products to the 715 cellular translation machinery. 716 The robust stimulation of Akt and mTOR may have other functional consequences for 717 the cell biology of SFV infection. Thus, the strong perturbation of cellular signalling may 718 pose a potential caveat for the use of SFV-based vectors for studies of molecular 719 pathogenesis, vaccination or gene therapy. The potential to activate this survival 720 pathway might have evolved to allow for establishment of virus persistence in the 721 mosquito host, as proposed for SINV (26); such a function is not manifested in 722 mammalian cells, where infection is lytic. There, activation of the PI3K–Akt–mTOR 723 pathway may help the virus to ensure efficient virus replication under suboptimal 724 growth conditions such as starvation. Intriguingly, the presence of the 725 hyperphosphorylated/acidic domain in nsP3 is one of the features that have been linked 726 to pathogenicity (especially neuropathology) of SFV in mice (33, 58), possibly connected 727 to the strong and persistent PI3K–Akt–mTOR activation. Yet, the domain and its 728 functional roles are unlikely to be the decisive reason for neurovirulence; virulence 729

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 34 determinants have been mapped to several features of the non-structural and also the 730 structural proteins of SFV (59-61). 731 The question remains why SFV but not CHIKV robustly activates PI3K–Akt–mTOR 732 signalling and efficiently internalises replication complexes. It is surprising that SFV and 733 CHIKV show these clear differences because these two Old World alphaviruses are 734 otherwise very closely related and thus expected to be very similar in many aspects of 735 the infected cell’s biology. Indeed, they do not differ regarding their replication 736 processes and other cell biological features such as the nsP2-dependent degradation of 737 RNA-polymerase II (62) or the nsP3-mediated sequestration of the stress granule 738 component G3BP (12). The differential activation of the PI3K–Akt–mTOR pathway by 739 SFV and CHIKV may have a cellular connection to disease and pathology, which are 740 distinctly different for these two related viruses. 741

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 35 Acknowledgements 742 743 We thank Karl Ljungberg, Peter Liljeström and Dan Grandér (Karolinska Institutet) for 744 provision of reagents and Mohammedyaseen Syedbasha (University of Helsinki) for 745 technical assistance. 746 Funding was provided by the Swedish Cancer Foundation (CAN 2012/789, to GMM), the 747 Swedish Research Council (621-2014-4718, to GMM), the German Research Foundation 748 (TH1896/1-1, to BT), the Estonian Ministry of Education and Research (IUT20-27, to 749 AM) and the Academy of Finland (grant 265997, to TA). The authors declare that no 750 competing interests exist. 751 752

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 36 Figure Legends 753 754 Fig. 1: Activation of the PI3K–Akt–mTOR pathway in SFV-infected cells, even 755 upon starvation. (A) The PI3K–Akt–mTOR pathway. Only proteins and inhibitors used 756 in this study are shown. Arrows indicate (indirect or direct) activation. Starvation 757 denotes complete absence of growth factors, amino acids and nutrients, to shut off 758 PI3K–Akt and mTOR activation. PDK, phospholipid-dependent kinase; S6K, S6 kinase; 759 other abbreviations see text. (B) HOS cells were infected with SFV-wt (MOI 10 for 1 h) or 760 mock-infected, then supplemented with growth medium (no starvation) or EBSS 761 (starvation) prior to lysis and Western blot analysis for the indicated proteins (P: 762 phosphorylated). Position of molecular mass marker (in kDa) indicated on the left. (C) 763 Immunofluorescence analysis of HOS cells, infected for 1 h with SFV-wt (MOI 0.2) or 764 mock-infected, then incubated in growth medium (no starvation) or EBSS (starvation) 765 for another 7 h, fixed and stained for phosphorylated S6 (pseudocoloured green) and 766 SFV-nsP1 (red). Nuclei stained with Hoechst 33258 (blue). Scale bar, 25 µm. (D/E) MEF 767 (D) and BHK cells (E) were treated as in B and assessed for activation of PI3K–Akt–768 mTOR by Western blot with the indicated antibodies. 769 770 Fig. 2: Characterisation of SFV-induced PI3K–Akt–mTOR activation. (A/B) Time 771 course experiment in HOS cells, infected for 1 h with SFV-wt (MOI 10) in PBS or mock-772 infected, then supplemented with growth medium (normal conditions, A) or EBSS 773 (starvation, B). Cell lysates were prepared at the indicated times p.i. and probed by 774 Western blot. Position of molecular mass marker (in kDa) indicated on the left. (C) Effect 775

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 37 of mTOR inhibition. HOS cells were infected with SFV-wt (MOI 10 for 1 h) or mock-776 infected, followed by incubation in growth medium (no mTOR inhibition), EBSS 777 (starvation [S]), growth medium with Rapamycin (R, 400 nM) or growth medium with 778 Torin-1 (T, 10 nM) for another 7 h prior to lysis and Western blot. (D) Effect of PI3K 779 inhibition and starvation. HOS cells were infected as in (C), followed by incubation in 780 growth medium or EBSS (starvation), without or with 400 nM Wortmannin as indicated, 781 for another 7 h, then lysed and analysed by Western blot. 782 783 Fig. 3: Viral requirements for SFV-induced PI3K–Akt–mTOR activation. (A) 784 BHK cells were infected with SFV-wt or SFV–βgal (MOI 10, 1 h) or mock-infected, 785 followed by incubation in growth medium for another 7 h, lysis and Western blot for the 786 indicated proteins. (B) HOS cells were infected at MOI 0.2 with the indicated viruses for 787 1 h or mock-infected and fixed at 8 h p.i. or were transfected with the nsP123-encoding 788 plasmid and fixed 24 h later, followed by immunofluorescence for phospho-Akt (S473, 789 green) and SFV-nsP1 (red), nuclei stained with Hoechst 33258 (blue). 790 791 Fig. 4: Requirements in nsP3 for SFV-induced PI3K–Akt–mTOR activation. (A) 792 Schematic depiction of SFV-nsP3 (482 amino acids) and sketch of the nsP3 deletions in 793 the virus mutants employed below. (B) BHK cells were infected with the indicated 794 variants of SFV (MOI 10, 1 h) or mock-infected, incubated in growth medium, lysed at 795 8 h p.i. and assessed by Western blot for the indicated proteins. Position of molecular 796 mass marker (in kDa) indicated on the left. Relative phospho-Akt signal intensities (% of 797 wt): SFV-Δ50: 59 ± 14, -ΔP: 107 ± 10, -Δ789: 99 ± 10 (n = 3) (C) HOS cells were infected 798

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 38 with the indicated SFV variants (MOI 0.2, 1 h) or mock-infected, followed by incubation 799 in DMEM for 7 h, fixation and staining for phospho-Akt (S473, green) and SFV-nsP1 800 (red), nuclei stained with Hoechst 33258 (blue). Scale bar, 25 µm. (D) BHK cells were 801 infected with SFV-Δ50 at MOI 10 for 1 h, incubation was continued in growth medium 802 (no starvation) or EBSS (starvation), without or with 400 nM Wortmannin, for another 803 7 h prior to lysis and Western blot for the indicated proteins. Compare with Fig. 2D. (E) 804 BHK cells were infected with SFV-wt or SFV-Δ50 at MOI 10 for 1 h and incubated for 805 another 3 h (left panel) or another 7 h (right panel), followed by lysis and Western blot 806 for the indicated proteins. 807 808 Fig. 5: Activation of Akt by ectopically expressed SFV-nsP3 variants. HOS cells 809 were mock-transfected (A) or transfected with plasmids encoding FLAG-tagged nsP3-wt 810 (B), nsP3-Δ50 (C), Myr-Pal–nsP3-wt (D) or Myr-Pal–nsP3-Δ50 (E), fixed 24 h post-811 transfection and stained for phospho-Akt (S473, green) and the FLAG tag (red), nuclei 812 stained with Hoechst 33258 (blue). Scale bar, 25 µm. 813 814 Fig. 6: Subcellular localisation of SFV replication complexes. BHK cells were 815 infected with the indicated SFV variants (schematic depiction in Fig. 4A) at MOI 10 for 816 1 h or mock-infected, incubated in growth medium for 7 h and fixed. Drug treatment 817 (added 1 h p.i. and present until fixation at 8 h p.i.): Wortmannin (B), 400 nM, Torin (C), 818 10 nM, Rapamycin (D): 400 nM. Immunofluorescence staining for dsRNA (red) and SFV-819 nsP3 (green); nuclei stained with DRAQ5 (blue). Representative confocal micrographs. 820 Scale bar, 25 µm. 821

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 39 822 Fig. 7: Akt activation and replication complex internalisation of recombinant 823 SFV with mutations in the hyperphosphorylated/acidic domain of nsP3. (A) 824 Mutations of nsP3 in the indicated SFV variants. nsP3-Δ26-4S-4A is not phosphorylated. 825 (B–D) BHK cells were infected with SFV-Δ24 (B), -Δ26 (C) or -Δ26-4S-4A (D) at MOI 10 826 for 1 h, incubated in growth medium for another 7 h, fixed 8 h p.i. and stained for dsRNA 827 (red) and SFV-nsP3 (green); nuclei stained with DRAQ5 (blue). Representative confocal 828 micrographs. Scale bar, 25 µm. (E) BHK cells were infected with the indicated viruses at 829 MOI 10 for 1 h or mock-infected, followed by incubation in growth medium for 7 h, lysis 830 and Western blot as indicated. Position of molecular mass marker (in kDa) indicated on 831 the left. Relative phospho-Akt signal intensities (% of wt): SFV-Δ50: 65 ± 3, -Δ24: 94 ± 832 16, -Δ26: 102 ± 17, -Δ26-4S-4A: 93 ± 18 (n = 3). 833 834 Fig. 8: PI3K–Akt–mTOR activation and replication complex dynamics upon 835 infection with CHIKV. (A) BHK cells were mock-infected or infected with the indicated 836 viruses at MOI 10 for 1 h, followed by incubation in growth medium and lysis at 8 h 837 (SFV) or at 4, 8 or 16 h (CHIKV), analysis by Western blot for the indicated proteins. 838 CHIKV-Δ5: deletion of the hyperphosphorylated/acidic domain in nsP3. Position of 839 molecular mass marker (in kDa) indicated on the left. (B) Sensitivity to PI3K inhibition: 840 BHK cells were infected as in (A), followed by incubation in growth medium, without or 841 with 400 nM Wortmannin, for another 15 h, then lysed and analysed by Western blot. 842 (C) Effect of starvation. BHK cells were infected as in (A), followed by incubation in 843 growth medium or EBSS (starvation) for another 7 h prior to lysis and Western blot. (D–844

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 40 F) Immunofluorescence analysis of replication complexes. BHK cells were mock-infected 845 (D) or infected with CHIKV-wt (E) or CHIKV-Δ5 (F) at MOI 10 for 1 h, followed by 846 incubation in growth medium for 7 h, fixation and immunofluorescence staining for 847 dsRNA (red) and CHIKV-nsP3 (green); nuclei stained with DRAQ5 (blue). Representative 848 confocal micrographs are displayed. Scale bar, 25 µm. 849 850 Fig. 9: The hyperphosphorylated/acidic domain of SFV-nsP3 is necessary but 851 not sufficient for replication complex internalisation and Akt activation. (A/B) 852 BHK cells were infected with recombinant SFV where the hyperphosphorylated/acidic 853 domain of nsP3 was replaced with its counterpart from CHIKV-nsP3 (SFV/CHIKV5, A) or 854 CHIKV containing the reciprocal swap (CHIKV/SFV50, B) at MOI 0.1 for 8 h, fixed and 855 stained for dsRNA (red); nuclei stained with DRAQ5 (blue). Representative confocal 856 micrographs; scale bar: 25 µm. (C–F) HOS cells were transfected with the indicated 857 membrane-anchored, FLAG-tagged nsP3 constructs for 24 h, fixed and stained for 858 phospho-Akt (S473, green) and FLAG (red), nuclei stained with Hoechst 33258 (blue). 859 Representative micrographs, scale bar, 25 µm. 860 861 Fig. 10: The hypervariable domain of SFV-nsP3 induces replication complex 862 internalisation and Akt activation in CHIKV. (A–D) BHK cells were infected with 863 CHIKV-wt (A), CHIKV containing the hypervariable domain of SFV-nsP3 (CHIKV/SFV-864 HVD, B), SFV with the hypervariable domain of CHIKV-nsP3 (SFV/CHIKV-HVD, C) or 865 SFV-wt (D) at MOI 10 for 8 h, fixed and stained for dsRNA (red); nuclei stained with 866 DRAQ5 (blue). Representative confocal micrographs; scale bar: 25 µm. (E) BHK cells 867

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 41 were mock-infected or infected with the indicated virus at MOI 10, lysed at 8 h p.i. and 868 subjected to Western blot with the indicated antibodies. (F–H) HOS cells were 869 transfected with the indicated membrane-anchored, FLAG-tagged nsP3 constructs for 870 24 h, fixed and stained for phospho-Akt (S473, green) and FLAG (red), nuclei stained 871 with Hoechst 33258 (blue). Representative micrographs, scale bar, 25 µm. 872 873 Fig. 11: Model of SFV- and CHIKV-induced PI3K–Akt–mTOR activation and 874 replication complex dynamics. (I) CHIKV and SFV-Δ50 induce basal, Wortmannin-875 sensitive Akt activation; replication complexes remain at the cell periphery. (II) SFV 876 carrying the hyperphosphorylated/acidic domain in nsP3 boosts PI3K–Akt–mTOR 877 activation, and replication complex internalisation is promoted. Interactions of the 878 hypervariable domain of SFV-nsP3 depicted in the middle panel (purple box). 879 Representative confocal micrographs of replication complexes (staining for dsRNA) as in 880 Fig. 8E (CHIKV-wt, top) and 6A (SFV-wt, bottom). 881

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 42 References 882 1. Atkins GJ, Sheahan BJ, Liljeström P. 1999. The molecular pathogenesis of 883 Semliki Forest virus: a model virus made useful? The Journal of general virology 884

80:2287–2297. 885 2. Karlsson GB, Liljeström P. 2004. Delivery and expression of heterologous genes 886 in mammalian cells using self-replicating alphavirus vectors. Methods in 887 molecular biology 246:543–557. 888 3. Ljungberg K, Liljeström P. 2015. Self-replicating alphavirus RNA vaccines. 889 Expert review of vaccines 14:177–194. 890 4. Quetglas JI, Ruiz-Guillen M, Aranda A, Casales E, Bezunartea J, Smerdou C. 891 2010. Alphavirus vectors for cancer therapy. Virus research 153:179–196. 892 5. Karabatsos N. 1975. Antigenic relationships of group A arboviruses by plaque 893 reduction neutralization testing. The American journal of tropical medicine and 894 hygiene 24:527–532. 895 6. Jose J, Snyder JE, Kuhn RJ. 2009. A structural and functional perspective of 896 alphavirus replication and assembly. Future microbiology 4:837–856. 897 7. Rupp JC, Sokoloski KJ, Gebhart NN, Hardy RW. 2015. Alphavirus RNA 898 synthesis and nonstructural protein functions. The Journal of general virology: 899 doi: 10.1099/jgv.1090.000249. 900 8. Malet H, Coutard B, Jamal S, Dutartre H, Papageorgiou N, Neuvonen M, Ahola 901 T, Forrester N, Gould EA, Lafitte D, Ferron F, Lescar J, Gorbalenya AE, de 902 Lamballerie X, Canard B. 2009. The crystal structures of Chikungunya and 903 Venezuelan equine encephalitis virus nsP3 macro domains define a conserved 904 adenosine binding pocket. Journal of virology 83:6534–6545. 905 9. Neuvonen M, Ahola T. 2009. Differential activities of cellular and viral macro 906 domain proteins in binding of ADP-ribose metabolites. Journal of molecular 907 biology 385:212–225. 908 10. Shin G, Yost SA, Miller MT, Elrod EJ, Grakoui A, Marcotrigiano J. 2012. 909 Structural and functional insights into alphavirus polyprotein processing and 910 pathogenesis. Proceedings of the National Academy of Sciences of the United 911 States of America 109:16534–16539. 912 11. McInerney GM. 2015. FGDF motif regulation of stress granule formation. DNA 913 and Cell Biology:DNA-2015-2957. 914 12. Panas MD, Ahola T, McInerney GM. 2014. The C-terminal repeat domains of 915 nsP3 from the Old World alphaviruses bind directly to G3BP. Journal of virology 916 88:5888–5893. 917 13. Panas MD, Varjak M, Lulla A, Eng KE, Merits A, Karlsson Hedestam GB, 918 McInerney GM. 2012. Sequestration of G3BP coupled with efficient translation 919 inhibits stress granules in Semliki Forest virus infection. Molecular biology of the 920 cell 23:4701–4712. 921 14. Panas MD, Schulte T, Thaa B, Sandalova T, Kedersha N, Achour A, McInerney 922 GM. 2015. Viral and Cellular Proteins Containing FGDF Motifs Bind G3BP to Block 923 Stress Granule Formation. PLoS pathogens 11:e1004659. 924 15. Kedersha N, Ivanov P, Anderson P. 2013. Stress granules and cell signaling: 925 more than just a passing phase? Trends in biochemical sciences 38:494–506. 926 16. Spuul P, Balistreri G, Kääriäinen L, Ahola T. 2010. Phosphatidylinositol 3-927 kinase-, actin-, and microtubule-dependent transport of Semliki Forest Virus 928

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 43 replication complexes from the plasma membrane to modified lysosomes. Journal 929 of virology 84:7543–7557. 930 17. Manning BD, Cantley LC. 2007. AKT/PKB signaling: navigating downstream. Cell 931 129:1261–1274. 932 18. Laplante M, Sabatini DM. 2012. mTOR signaling in growth control and disease. 933 Cell 149:274–293. 934 19. Dunn EF, Connor JH. 2012. HijAkt: The PI3K/Akt pathway in virus replication 935 and pathogenesis. Progress in molecular biology and translational science 936 106:223–250. 937 20. Cooray S. 2004. The pivotal role of phosphatidylinositol 3-kinase-Akt signal 938 transduction in virus survival. The Journal of general virology 85:1065–1076. 939 21. Diehl N, Schaal H. 2013. Make yourself at home: viral hijacking of the PI3K/Akt 940 signaling pathway. Viruses 5:3192–3212. 941 22. Buchkovich NJ, Yu Y, Zampieri CA, Alwine JC. 2008. The TORrid affairs of 942 viruses: effects of mammalian DNA viruses on the PI3K-Akt-mTOR signalling 943 pathway. Nature reviews. Microbiology 6:266–275. 944 23. Ehrhardt C, Marjuki H, Wolff T, Nürnberg B, Planz O, Pleschka S, Ludwig S. 945 2006. Bivalent role of the phosphatidylinositol-3-kinase (PI3K) during influenza 946 virus infection and host cell defence. Cellular microbiology 8:1336–1348. 947 24. Dunn EF, Connor JH. 2011. Dominant inhibition of Akt/protein kinase B 948 signaling by the matrix protein of a negative-strand RNA virus. Journal of virology 949 85:422–431. 950 25. Mohankumar V, Dhanushkodi NR, Raju R. 2011. Sindbis virus replication, is 951 insensitive to rapamycin and torin1, and suppresses Akt/mTOR pathway late 952 during infection in HEK cells. Biochemical and biophysical research 953 communications 406:262–267. 954 26. Patel RK, Hardy RW. 2012. Role for the phosphatidylinositol 3-kinase-Akt-TOR 955 pathway during sindbis virus replication in arthropods. Journal of virology 956 86:3595–3604. 957 27. Das I, Basantray I, Mamidi P, Nayak TK, Pratheek BM, Chattopadhyay S, 958 Chattopadhyay S. 2014. Heat shock protein 90 positively regulates Chikungunya 959 virus replication by stabilizing viral non-structural protein nsP2 during infection. 960 PloS one 9:e100531. 961 28. Salonen A, Vasiljeva L, Merits A, Magden J, Jokitalo E, Kääriäinen L. 2003. 962 Properly folded nonstructural polyprotein directs the Semliki Forest virus 963 replication complex to the endosomal compartment. Journal of virology 964 77:1691–1702. 965 29. Liljeström P, Lusa S, Huylebroeck D, Garoff H. 1991. In vitro mutagenesis of a 966 full-length cDNA clone of Semliki Forest virus: the small 6,000-molecular-weight 967 membrane protein modulates virus release. Journal of virology 65:4107–4113. 968 30. Sjöberg EM, Suomalainen M, Garoff H. 1994. A significantly improved Semliki 969 Forest virus expression system based on translation enhancer segments from the 970 viral capsid gene. Bio/technology 12:1127–1131. 971 31. Smerdou C, Liljeström P. 1999. Two-helper RNA system for production of 972 recombinant Semliki Forest virus particles. Journal of virology 73:1092–1098. 973 32. Ulper L, Sarand I, Rausalu K, Merits A. 2008. Construction, properties, and 974 potential application of infectious plasmids containing Semliki Forest virus full-975

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 44 length cDNA with an inserted intron. Journal of virological methods 148:265–976 270. 977 33. Vihinen H, Ahola T, Tuittila M, Merits A, Kääriäinen L. 2001. Elimination of 978 phosphorylation sites of Semliki Forest virus replicase protein nsP3. The Journal 979 of biological chemistry 276:5745–5752. 980 34. Neuvonen M, Kazlauskas A, Martikainen M, Hinkkanen A, Ahola T, Saksela 981 K. 2011. SH3 domain-mediated recruitment of host cell amphiphysins by 982 alphavirus nsP3 promotes viral RNA replication. PLoS pathogens 7:e1002383. 983 35. Varjak M, Žusinaite E, Merits A. 2010. Novel functions of the alphavirus 984 nonstructural protein nsP3 C-terminal region. Journal of virology 84:2352–2364. 985 36. Vihinen H, Saarinen J. 2000. Phosphorylation site analysis of Semliki forest 986 virus nonstructural protein 3. The Journal of biological chemistry 275:27775–987 27783. 988 37. Pohjala L, Utt A, Varjak M, Lulla A, Merits A, Ahola T, Tammela P. 2011. 989 Inhibitors of alphavirus entry and replication identified with a stable 990 Chikungunya replicon cell line and virus-based assays. PloS one 6:e28923. 991 38. Hallengärd D, Kakoulidou M, Lulla A, Kümmerer BM, Johansson DX, Mutso 992 M, Lulla V, Fazakerley JK, Roques P, Le Grand R, Merits A, Liljeström P. 2014. 993 Novel attenuated Chikungunya vaccine candidates elicit protective immunity in 994 C57BL/6 mice. Journal of virology 88:2858–2866. 995 39. Spuul P, Balistreri G, Hellström K, Golubtsov AV, Jokitalo E, Ahola T. 2011. 996 Assembly of alphavirus replication complexes from RNA and protein components 997 in a novel trans-replication system in mammalian cells. Journal of virology 998 85:4739–4751. 999 40. Gorchakov R, Frolova E, Williams BR, Rice CM, Frolov I. 2004. PKR-dependent 1000 and -independent mechanisms are involved in translational shutoff during 1001 Sindbis virus infection. Journal of virology 78:8455–8467. 1002 41. Cristea IM, Carroll JW, Rout MP, Rice CM, Chait BT, MacDonald MR. 2006. 1003 Tracking and elucidating alphavirus-host protein interactions. The Journal of 1004 biological chemistry 281:30269–30278. 1005 42. Kujala P, Rikkonen M, Ahola T, Kelve M, Saarma M, Kääriäinen L. 1997. 1006 Monoclonal antibodies specific for Semliki Forest virus replicase protein nsP2. 1007 The Journal of general virology 78:343–351. 1008 43. Tamberg N, Lulla V, Fragkoudis R, Lulla A, Fazakerley JK, Merits A. 2007. 1009 Insertion of EGFP into the replicase gene of Semliki Forest virus results in a 1010 novel, genetically stable marker virus. The Journal of general virology 88:1225–1011 1230. 1012 44. Scholte FE, Tas A, Albulescu IC, Žusinaite E, Merits A, Snijder EJ, van Hemert 1013 MJ. 2015. Stress Granule Components G3BP1 and G3BP2 Play a Proviral Role 1014 Early in Chikungunya Virus Replication. Journal of virology 89:4457–4469. 1015 45. Utt A, Das PK, Varjak M, Lulla V, Lulla A, Merits A. 2015. Mutations conferring 1016 a noncytotoxic phenotype on chikungunya virus replicons compromise 1017 enzymatic properties of nonstructural protein 2. Journal of virology 89:3145–1018 3162. 1019 46. Peränen J, Laakkonen P, Hyvonen M, Kääriäinen L. 1995. The alphavirus 1020 replicase protein nsP1 is membrane-associated and has affinity to endocytic 1021 organelles. Virology 208:610–620. 1022

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 45 47. Downward J. 1998. Mechanisms and consequences of activation of protein 1023 kinase B/Akt. Current opinion in cell biology 10:262–267. 1024 48. Resh MD. 1999. Fatty acylation of proteins: new insights into membrane 1025 targeting of myristoylated and palmitoylated proteins. Biochimica et biophysica 1026 acta 1451:1–16. 1027 49. Gorchakov R, Garmashova N, Frolova E, Frolov I. 2008. Different types of 1028 nsP3-containing protein complexes in Sindbis virus-infected cells. Journal of 1029 virology 82:10088–10101. 1030 50. Ahola T, Kujala P, Tuittila M, Blom T, Laakkonen P, Hinkkanen A, Auvinen P. 1031 2000. Effects of palmitoylation of replicase protein nsP1 on alphavirus infection. 1032 Journal of virology 74:6725–6733. 1033 51. Spuul P, Salonen A, Merits A, Jokitalo E, Kääriäinen L, Ahola T. 2007. Role of 1034 the amphipathic peptide of Semliki forest virus replicase protein nsP1 in 1035 membrane association and virus replication. Journal of virology 81:872–883. 1036 52. Frolova E, Gorchakov R, Garmashova N, Atasheva S, Vergara LA, Frolov I. 1037 2006. Formation of nsP3-specific protein complexes during Sindbis virus 1038 replication. Journal of virology 80:4122–4134. 1039 53. Shin YK, Li Y, Liu Q, Anderson DH, Babiuk LA, Zhou Y. 2007. SH3 binding motif 1040 1 in influenza A virus NS1 protein is essential for PI3K/Akt signaling pathway 1041 activation. Journal of virology 81:12730–12739. 1042 54. Hale BG, Jackson D, Chen YH, Lamb RA, Randall RE. 2006. Influenza A virus 1043 NS1 protein binds p85beta and activates phosphatidylinositol-3-kinase signaling. 1044 Proceedings of the National Academy of Sciences of the United States of America 1045

103:14194–14199. 1046 55. Peränen J, Takkinen K, Kalkkinen N, Kääriäinen L. 1988. Semliki Forest virus-1047 specific non-structural protein nsP3 is a phosphoprotein. The Journal of general 1048 virology 69 ( Pt 9):2165–2178. 1049 56. Li GP, LaStarza MW, Hardy WR, Strauss JH, Rice CM. 1990. Phosphorylation of 1050 Sindbis virus nsP3 in vivo and in vitro. Virology 179:416–427. 1051 57. Frolova EI, Gorchakov R, Pereboeva L, Atasheva S, Frolov I. 2010. Functional 1052 Sindbis virus replicative complexes are formed at the plasma membrane. Journal 1053 of virology 84:11679–11695. 1054 58. Galbraith SE, Sheahan BJ, Atkins GJ. 2006. Deletions in the hypervariable 1055 domain of the nsP3 gene attenuate Semliki Forest virus virulence. The Journal of 1056 general virology 87:937–947. 1057 59. Ferguson MC, Saul S, Fragkoudis R, Weisheit S, Cox J, Patabendige A, 1058 Sherwood K, Watson M, Merits A, Fazakerley JK. 2015. The ability of the 1059 encephalitic arbovirus Semliki Forest virus to cross the blood brain barrier is 1060 determined by the charge of the E2 glycoprotein. Journal of virology 89:7536–1061 7549. 1062 60. Tuittila M, Hinkkanen AE. 2003. Amino acid mutations in the replicase protein 1063 nsP3 of Semliki Forest virus cumulatively affect neurovirulence. The Journal of 1064 general virology 84:1525–1533. 1065 61. Saul S, Ferguson MC, Cordonin C, Fragkoudis R, Ool M, Tamberg N, 1066 Sherwood K, Fazakerley JK, Merits A. 2015. Differences in processing 1067 determinants of non-structural polyprotein and in the sequence of non-structural 1068 protein 3 affect neurovirulence of Semliki Forest virus. Journal of virology: 1069 JVI01186 (accepted for publication). 1070

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

THAA ET AL., PI3K–AKT–MTOR ACTIVATION BY SFV AND CHIKV 46 62. Akhrymuk I, Kulemzin SV, Frolova EI. 2012. Evasion of the innate immune 1071 response: the Old World alphavirus nsP2 protein induces rapid degradation of 1072 Rpb1, a catalytic subunit of RNA polymerase II. Journal of virology 86:7180–1073 7191. 1074 1075

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

on March 27, 2018 by guest

http://jvi.asm.org/

Dow

nloaded from

Correction for Thaa et al., Differential Phosphatidylinositol-3-Kinase-Akt-mTOR Activation by Semliki Forest and Chikungunya Viruses IsDependent on nsP3 and Connected to Replication ComplexInternalization

Bastian Thaa,a Roberta Biasiotto,a Kai Eng,a Maarit Neuvonen,b Benjamin Götte,a Lara Rheinemann,a Margit Mutso,c Age Utt,c

Finny Varghese,d Giuseppe Balistreri,b Andres Merits,c Tero Ahola,d Gerald M. McInerneya

Karolinska Institutet, Department of Microbiology, Tumor and Cell Biology, Stockholm, Swedena; University of Helsinki, Institute of Biotechnology, Helsinki, Finlandb;University of Tartu, Institute of Technology, Tartu, Estoniac; University of Helsinki, Department of Food and Environmental Sciences, Helsinki, Finlandd

Volume 89, no. 22, p. 11420 –11437, 2015. Page 11431, Fig. 8: Due to an error in figure file processing, the immunofluorescence data inFig. 8F do not display replication complexes (dsRNA, red) and nsP3 (green) of CHIKV-nsP3�5 but instead show those of SFV-�50(as in Fig. 6E). The figure legend is correct as originally published.

Fig. 8F should appear as shown below.

Page 11430, column 1, lines 13 to 16: The sentence “We noted a difference in nsP3 localizations between CHIKV-wt and -�5, probablyreflecting engagement of nsP3 in various complexes in cells (49)” should be deleted, because no such difference is obvious from theimages.

The authors regret the errors. These corrections do not change the interpretation of and the conclusions from the data.

Citation Thaa B, Biasiotto R, Eng K, Neuvonen M, Götte B, Rheinemann L, Mutso M,Utt A, Varghese F, Balistreri G, Merits A, Ahola T, McInerney GM. 2015. Differentialphosphatidylinositol-3-kinase-Akt-mTOR activation by Semliki Forest andchikungunya viruses is dependent on nsP3 and connected to replication complexinternalization. J Virol 90:4256. doi:10.1128/JVI.00202-16.

Copyright © 2016, American Society for Microbiology. All Rights Reserved.

AUTHOR CORRECTION

crossmark

4256 jvi.asm.org April 2016 Volume 90 Number 8Journal of Virology