Host sphingomyelin increases West Nile virus infection in vivo

Flaviviruses, such as the dengue virus and the West Nile virus (WNV), are arthropod-borne viruses that represent a global health problem. The flavivirus lifecycle is intimately connected to cellular lipids. Among the lipids co-opted by flaviviruses, we have focused on SM, an important component of cellular membranes particularly enriched in the nervous system. After infection with the neurotropic WNV, mice deficient in acid sphingomyelinase (ASM), which accumulate high levels of SM in their tissues, displayed exacerbated infection. In addition, WNV multiplication was enhanced in cells from human patients with Niemann-Pick type A, a disease caused by a deficiency of ASM activity resulting in SM accumulation. Furthermore, the addition of SM to cultured cells also increased WNV infection, whereas treatment with pharmacological inhibitors of SM synthesis reduced WNV infection. Confocal microscopy analyses confirmed the association of SM with viral replication sites within infected cells. Our results unveil that SM metabolism regulates flavivirus infection in vivo and propose SM as a suitable target for antiviral design against WNV.


Infections and virus titrations
For infections in liquid medium, the viral inoculum was incubated with cell monolayers for 1 h, and then the inoculum was removed and fresh medium containing 1% FBS was added. This time point was considered 1 h postinfection (p.i.). Virus titrations were performed by standard plaque assay on Vero cells ( 28 ). The multiplicity of infection (MOI) used in each experiment was expressed as the number of plaque-forming units (PFUs) per cell and is indicated in the corresponding fi gure legend.

Mice
Age-and sex-matched 5-month-old wt (C57BL/6) or ASMko ( 21 ) mice were used. A breeding colony was established from a couple of ASM heterozygous C57BL/6 mice kindly donated by Prof. E. H. Schuchman (Mount Sinai School of Medicine, New York) and genotyped as described ( 29 ). Mice were intraperitoneally inoculated with 10 4 PFU per mouse of WNV and monitored daily for signs of infection up to 20 days p.i. Animals were kept with ad libitum access to food and water and those exhibiting clear signs of disease were anesthetized and euthanized, as were all surviving mice, at the end of the experiment. All animals were handled in strict accordance with the guidelines of the European Community 86/609/CEE at the BSL-3 animal facilities of the Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA) and the CBMSO . The protocols were approved by the Committee on Ethics of Animal Experimentation of INIA (permit number 2015-004E and PROEX 05/14) and the Animal Welfare Committee of Centro de Biología Molecular "Severo Ochoa" (CBMSO) (permit numbers CEEA-CBMSO-22I149 and PROEX 034/15).

Hippocampal slice cultures
Organotypic slice cultures of the hippocampus were prepared from 5-month-old wt or ASMko mice, as previously described ( 30 ). Six slices (400 m thick each) were placed per insert and cultured for 24 h at 37°C in a 5% CO 2 atmosphere prior to infection. Each slice was individually infected with 10 5 PFU of WNV. Slices were harvested 24 h p.i. and homogenized in PBS for quantitative RT-PCR analysis .

Cell treatments
SM (Santa Cruz Biotechnology, Dallas, TX) was dissolved in ethanol and added to the cells 24 h before infection ( 29 ). D609 (Sigma, St. Louis, MO), SPK-601 (LMV-601; Eurodiagnostico, Madrid, Spain), and MS-209 (dofequidar fumarate; Sigma) were dissolved in DMSO and added to infected cultures 1 h p.i. Control cells were treated in parallel with the same amount of drug vehicle.

SM quantifi cation
Biochemical analysis of SM in brain extracts or cultured cells containing the same amount of protein was performed using an enzymatic fl uorescence assay. Briefl y, lipid extracts were dried in the presence of detergent (Thesit), and SM was subsequently converted into choline by means of sphingomyelinase and alkaline phosphatase, and coupled to the production of fl uorescence with choline oxidase, peroxidase, and homovanillic acid, as modifi ed from Hojjati and Jiang ( 31 ).

Immunohistochemistry
Right brain hemispheres from four randomly selected mice per condition were fi xed in 4% paraformaldehyde, dehydrated, and embedded in paraffi n. Serial sagittal sections (3 m thick) is in turn the central core of SM ( 15 ). The sphingolipid content of biological membranes defi nes important physical and biological properties ( 16,17 ). Viruses can take advantage of these properties to develop specialized membrane sites for RNA replication and particle biogenesis ( 18 ). Lipidomic analyses have shown an increase in the content of both ceramide and SM in fl avivirusinfected cells ( 8,12 ), and ceramide has specifi cally been associated with WNV replication and viral particle biogenesis ( 8,9 ). Regarding SM, it is enriched in membranes from the replication complex of viruses phylogenetically related to WNV, such as dengue virus and hepatitis C virus ( 12,19 ). For WNV, an enrichment of SM in the viral envelope relative to total cellular membranes has been described ( 8 ). Despite this observation, the contribution of SM, the most abundant complex sphingolipid in mammalian cells ( 15 ), has not been addressed in detail in WNV infection.
The levels of SM are tightly controlled by cellular sphingomyelinases ( 15 ). Consistently, defects in the activity of acid sphingomyelinase (ASM) result in the accumulation of SM, causing a sphingolipidosis termed Niemann-Pick disease type A (NPA) ( 20 ). Because ASM is a ubiquitous enzyme, SM accumulates in all kinds of cells in NPA patients leading to peripheral symptoms, especially evident in enlarged liver and spleen. However, the most affected organ is the brain, resulting in severe cognitive defi cits, neurodegeneration, and early death. Mice lacking ASM [ASM knockout mice (ASMko)] reproduce NPA and accumulate SM, but not ceramide, in their tissues (21)(22)(23). Accumulation of SM in the ASMko mouse brain cells is progressive, reaching a 5-fold increase compared with wild-type (wt) mice (20)(21)(22). This increase is similar to that reported in the cerebral cortex of NPA patients ( 24 ). ASMko mice and primary cell cultures from NPA patients have provided useful models to study the role of SM in viral infection ( 25,26 ). Using these models, we here show, for the fi rst time, that SM levels modulate WNV infection in vitro and in vivo, thus identifying this sphingolipid as a key cellular factor for WNV replication.

Cells and viruses
All virus manipulations were performed in our biosafety level 3 (BSL-3) facilities. The origin and passage history of WNV strain NY99 has been described ( 27,28 ). Vero 76 cells, clone E6 (ATTC CRL-1586), were used for amplifi cation and titration of viruses. Primary human skin fi broblasts from unaffected individuals (AG07310 and AG07471) and NPA patients (GM13205 and GM16195) were from Coriell Institute for Medical Research (Camden, NJ). GM13205 carried a deletion of a single C in exon 2 at codon 330 of the SMPD1 gene (which encodes human ASM) resulting in a frameshift leading to the formation of a premature stop at codon 382. GM16195 is homozygous for a T to C transition at nucleotide 905 of the SPMD1 gene, resulting in a Leu to Pro substitution at codon 302. Cells were cultured (37°C, 5% CO 2 ) in DMEM supplemented with 2 mM glutamine, an ER-targeted red fl uorescent protein ( 5 ). Appropriate secondary antibodies labeled with Alexa Fluor 488, 594, or 647 were from Invitrogen. For plasma membrane lysenin staining, cells were observed using an Axioskop (Zeiss, Oberkochen, Germany) epifl uorescence microscope with a Plan-Neofl uar Ph3 40×/1.3 oil immersion objective. Images were acquired with a monochrome Coolsnap FX camera (Roper Scientifi c, Tucson, AZ) and fl uorescence was quantifi ed using ImageJ software (http://imagej.nih. gov/ij/). In the case of confocal microscopy analyses, cells were observed using a Leica TCS SPE confocal laser scanning microscope using an HCX PL APO 63×/1.4 oil immersion objective. Optical slice thickness for all confocal images displayed was of 1 airy unit.

Quantifi cation of viral RNA
For quantifi cation of cell-associated viral RNA, infected cell monolayers were washed, supernatants were replaced by fresh medium, and the cells were subjected to three freeze-thaw cycles. In the case of animal samples or organotypic slice cultures, tissues were homogenized in PBS using TissueLyser II equipment (Quiagen, Venlo, The Netherlands). Viral RNA was extracted from samples with Speedtools RNA virus extraction kit (Biotools, Madrid, Spain). Positive-strand viral RNA was quantifi ed by realtime RT-PCR ( 34 ) as genomic equivalents to PFUs by comparison with RNA extracted from previously titrated samples ( 35 ).

Data analysis
ANOVA, applying Bonferroni's correction, was performed with the statistical package, SPSS 15. Nonparametric data were analyzed using the Mann-Whitney U test (GraphPad Prism 2.01).
were deparaffi nized in xylene and rehydrated in graded alcohol. Endogenous peroxidase was inactivated by incubation in 1.5% H 2 O 2 in methanol for 20 min. Sections were placed in blocking buffer (1% BSA, 5% FBS, 2% Triton X-100 in PBS) for 1 h and incubated overnight at 4°C with an anti-WNV E glycoprotein (3.67G) mouse monoclonal antibody (Millipore, Temecula, CA) in blocking buffer. Subsequently, sections were incubated with biotinylated anti-mouse IgG (Vector Laboratories, Burlingame, CA) and stained using the Elite Vectastain ABC kit (Vector). Immunoreactivity was developed with diaminobenzidine (Dako, Glostrup, Denmark), which yields a brownish precipitate, and sections were counterstained with hematoxylin. Contiguous brain sections were processed in parallel without primary antibody in order to determine nonspecifi c staining. Images were acquired using a Leica DM LB microscope equipped with a HCX PL 40×/0.75 objective and a digital camera, Leica DC 100.

Immunofl uorescence and confocal microscopy
Immunofl uorescence was performed as described ( 5 ). Plasma membrane staining with lysenin, which specifi cally binds to SM, was accomplished as described ( 23 ). For intracellular SM detection, lysenin staining was performed in a similar manner, except that cells were permeabilized with 50 g/ml digitonin after fi xation ( 32 ). Lysenin (Peptanova, Sandhausen, Germany) was detected using a specifi c rabbit antiserum (Peptanova). Doublestrand RNA (dsRNA) and WNV E protein were detected using mouse monoclonal antibodies J2 (English and Scientifi c Consulting, Hungary) and 3.67G, respectively. Actin microfi laments were visualized with TRITC-labeled phalloidin (Sigma). The ER was visualized by transfection with plasmid IgLdR1kdel ( 33 ) encoding ASMko mice that exhibited a 2.7-fold increase in their brain SM content relative to wt mice were used (218 ± 19 nmol/mg and 81 ± 5 nmol/mg of protein, respectively) ( Fig. 1A ). To determine whether the alteration of SM levels could modulate WNV infection in vivo, we fi rst infected ASMko mice ( 21 ) with a highly neurovirulent WNV NY99 strain ( 28 ) derived from the virus causing the outbreak of encephalitis in the Northeastern United States in 1999 ( 37 ). ASMko mice were signifi cantly more susceptible to WNVinduced lethality than wt mice when challenged with 10 4 PFU by intraperitoneal inoculation ( Fig. 1B ). Quantitative RT-PCR revealed a signifi cantly higher viral load in the brains of dead ASMko mice than in those of dead wt mice ( Fig. 1C ), indicating that WNV replicated to higher levels in ASMko mice. To test this hypothesis, the viral load in several neural and peripheral tissues was analyzed in randomly chosen animals at 6 days p.i. The values found were Kaplan-Meier survival curves were analyzed by a log-rank test using GraphPad Prism 2.01. Data are presented as mean ± SD. Differences with P values of <0.05 were considered statistically signifi cant. Asterisks in the fi gures indicate P values: * P < 0.05; ** P < 0.005.

RESULTS
ASMko mice are more susceptible to WNV-induced lethality and display enhanced virus replication compared to wt mice SM accumulation is an age-related process in ASMko mice that, close to the end of their life (7 months of age), exhibits a 5-fold higher level of SM in brain, while cholesterol and ceramide content remains unchanged with respect to wt mice ( 22,23,36 ). In our experiments, 5-month-old analyzed by means of immunofl uorescence and confocal microscopy 24 h p.i. Cells showing cytoplasmic cumuli of dsRNA intermediates, which are markers of WNV replication ( 3,5 ), as well as cells positive for WNV E glycoprotein, were observed in infected cultures ( Fig. 4A ), supporting the ability of WNV to replicate in these fi broblast lines. No positive signal for either dsRNA or E protein was observed in uninfected cultures included as negative controls, thus confi rming the specifi city of the staining ( Fig. 4A ). The amount of infectious virus released into the culture medium by NPA and control fi broblasts was measured at 24 and 48 h p.i. Maximum viral production was observed at 24 h p.i., which is consistent with previous reports showing that multiplication of WNV in normal human dermal fi broblast cultures peaks about 24 h p.i. ( 40 ). Remarkably, titration of supernatants from infected cultures indicated that NPA fi broblasts produced signifi cantly more infectious virus than control fi broblasts at both 24 and 48 h p.i. ( Fig. 4B ). This observation was confi rmed when the amount of genome-containing units in the culture medium was analyzed by quantitative RT-PCR 24 h p.i. ( Fig. 4C ). Furthermore, a statistically signifi cant increase in the amount of cell-associated viral RNA in NPA fi broblasts was noticed when compared with control fi broblasts 24 h p.i. ( Fig. 4D ). These results support that WNV replication is increased in NPA patient fi broblasts in comparison to those from unaffected individuals.

SM promotes WNV infection
The results obtained so far in ASMko mouse brains and in fi broblasts from NPA patients, which contain high levels signifi cantly higher (by several orders of magnitude) in the brain of ASMko than in wt mice ( Fig. 1D ). In addition, the amount of viral RNA in peripheral tissues (liver, spleen, kidney, and heart) of ASMko mice was also higher than in wt mice ( Fig. 1D ). These results confi rm that WNV infection is exacerbated in ASMko mice.

Brain tissues from ASMko mice display enhanced WNV infection both in vivo and ex vivo
WNV-infected cells were analyzed by immunohistochemistry in brain sections from wt and ASMko mice at 6 days p.i. using a monoclonal antibody that recognizes the WNV E glycoprotein. Our analysis was focused on the cortex and hippocampus, because both areas are important targets for WNV infection ( 38,39 ). Cells showing cytoplasmic positive staining for viral antigen were found in the prefrontal cortex, as well as in the dentate gyrus and in areas CA1 and CA3 of the hippocampus from both wt and ASMko infected mice ( Fig. 2A ). Quantifi cation of WNV antigen-positive cells revealed that ASMko mice showed signifi cantly more infected cells than wt mice in all the regions analyzed ( Fig. 2B ). To investigate whether this increase of WNV infection in neural tissues of ASMko mice was also observed ex vivo, organotypic hippocampal slice cultures were produced from noninfected wt and ASMko mice. Slice cultures were infected with WNV and the amount of viral RNA in the tissue was analyzed 24 h p.i. by quantitative RT-PCR. A signifi cant increase (about 3-fold) in the content of viral RNA was observed in hippocampal slices derived from ASMko mice as compared with those from wt mice ( Fig. 2C ). Overall, these results confi rm that the replication of WNV is enhanced in brain tissues of ASMko mice.

WNV infection is enhanced in fi broblasts from NPA patients
Together with brain cells, which are particularly affected by WNV, dermal fi broblasts are also a relevant type of cell for WNV infection, because they are one of the fi rst exposed to the virus-infected mosquitoes ( 40 ). For this reason, and to extend our analyses to the human scenario, we tested the vulnerability to infection of skin fi broblasts derived from NPA patients, which also showed SM accumulation ( 26,29 ). Quantifi cation of SM confi rmed that NPA fi broblasts displayed signifi cantly higher levels of SM than control fi broblasts (52.05 ± 1.26 nmol/mg and 13.09 ± 0.56 nmol/mg of protein, respectively) ( Fig. 3A ). In addition, because SM is enriched in the plasma membrane, the levels of SM at this cellular site were also analyzed in both control and NPA fi broblasts by lysenin staining of nonpermeabilized cells ( 29 ). A signifi cant 3-fold increase in lysenin staining was noticed in NPA fibroblasts when compared with control fi broblasts ( Fig. 3B, C ), which further confi rmed the accumulation of SM in cells derived from NPA patients. To test the vulnerability to infection by WNV of skin fi broblasts derived from NPA patients, two NPA fi broblast lines (GM13205 and GM16195) derived from different NPA patients and two control fi broblast lines derived from unaffected individuals (AG07310 and AG07471) were infected with WNV. Virus replication was vehicle alone ( Fig. 5C ). Moreover, the production of WNV in control line AG07471 was increased by the addition of SM, reaching the level observed in NPA fi broblasts. Consistent with the unchanged SM levels in NPA fi broblasts upon addition of the lipid, the amount of virus in these cells, which was already high compared with control fi broblasts, did not increase after SM addition ( Fig. 5C ). Again, this lack of effect of addition of SM could be due to the high levels of SM already existing in NPA fi broblasts. In fact, no signifi cant increase on SM levels was noticed upon addition of exogenous SM to NPA fi broblasts ( Fig. 5A, B ). To further analyze this point and to test whether high SM levels had the same effect in other cell lines, we added SM to Vero cells, which are a widely used model for WNV infection ( 3,5 ). The addition of SM signifi cantly increased SM content of Vero cells (5.95 ± 1.23 nmol/mg and 12.44 ± 2.34 nmol/mg of protein in control and treated cells, respectively) ( Fig. 5D ), which correlated with enhanced viral production ( Fig. 5E ). These results further support the positive role of high SM levels on the infection of WNV. of SM ( 22,26,29,36 ), suggest that the accumulation of SM could result in increased WNV multiplication. To directly test this, we fi rst added 40 M SM to NPA or control fi broblasts. Enzymatic quantifi cation of SM content confi rmed that addition of the lipid signifi cantly increased SM levels in control fi broblasts by 1.3-fold (16.93 ± 0.39 nmol/mg and 12.69 ± 0.32 nmol/mg of protein in treated and nontreated control fi broblasts, respectively) ( Fig. 5A ). Treatment with SM did not signifi cantly increase SM levels in NPA fi broblasts ( Fig. 5A ), likely due to the already very high SM levels in these cells. Similar results were obtained when SM was detected by lysenin staining ( Fig. 5B ), confi rming that addition of exogenous SM successfully increased this lipid content in control fi broblasts. To test the effects of SM increase in WNV infectivity, control and NPA fi broblasts were treated with the lipid and then infected with WNV. The titration of culture supernatants from infected fi broblasts revealed that the addition of SM signifi cantly increased WNV production in control fi broblasts when compared with those cultures treated with specialized membranes derived from the ER (see Introduction), Vero cells were transfected with plasmid IgLdR1kdel that encodes an ER-targeted red fl uorescent protein ( 33 ) to allow visualization of this organelle. Transfected cells were infected with WNV and both SM and dsRNA were labeled by immunofl uorescence ( Fig. 6B ). Confocal microscopy analyses revealed colocalization among dsRNA, SM, and the ER markers in discrete cytoplasmic foci, which confi rms the association of SM to the membranes derived from the ER where WNV replication takes place.

Inhibition of SM biosynthesis reduces WNV infection
Considering that our observations were consistent with the involvement of SM on WNV infection, we explored whether the inhibition of the synthesis of this lipid could provide not only further evidence for SM contribution to WNV infection, but also a feasible druggable target for antiviral strategies. SM is produced through the transfer of a phosphocholine headgroup from phosphatidylcholine to ceramide in a reaction catalyzed by SM synthases (SMSs) ( 15 ). Thus, we investigated the effect in WNV infection of SMS inhibitors D609, SPK-601 (an enantiomeric pure isomer of D609), and MS-209 (structurally unrelated to D609 and SPK-601) ( 43,44 ). Unfortunately, the low amount of endogenous SM displayed by Vero cells impaired the experimental confi rmation of the reduction of SM levels (using the enzymatic assay) as a result of the treatment with the inhibitors. Nevertheless, SM levels would be expected to decrease in cells treated with these inhibitors, as shown previously (45)(46)(47). Indeed, the three inhibitors signifi cantly reduced, by several orders of magnitude, the production of infectious particles in WNV-infected Vero cells ( Fig. 7A ). These data are consistent with the idea that whereas an increase in SM content increases virus multiplication, a decrease in SM could reduce virus infection. Interestingly, even when the three drugs exerted a reduction in the amount of genome-containing units released to the culture medium of infected cells ( Fig. 7B ), the extent of these inhibitions was lower than that of infectivity (compare Fig. 7A, B ). A possible explanation for this discrepancy is that treatment with SMS inhibitors increased the production of noninfectious genome-containing WNV particles. Considering that SM could play a structural role in WNV particles ( 8 ), the expected reduction on SM content due to SMS inhibition could result in a reduced availability of SM for viral envelopment, thus causing a decrease in the infectious viral particles produced. Similar effects have been observed in other viral models using other drugs that also block lipid synthesis (48)(49)(50). The amount of cellassociated viral RNA was also measured in cells treated with the inhibitors ( Fig. 7C ). Within the concentration range tested, D609 and MS-209 induced a signifi cant dosedependent reduction of WNV RNA accumulation inside cells that indicated a reduction in viral replication. On the contrary, no signifi cant differences were observed in cells treated with SPK-601, an isomer of D609. In contrast to SPK-601, D609 consists on a mixture of diastereomers ( 44 ), so the differences between both inhibitors could rely on the inhibitory properties of diastereomers present in D609

SM and WNV colocalize in ER membranes within infected cells
When the distribution of intracellular SM within infected Vero cells was evaluated by lysenin staining of permeabilized cells ( 41 ), colocalization between dsRNA and SM was observed in cytoplasmic foci ( Fig. 6A ). This sustained the localization of this lipid at WNV replication complex. Considering that lysenin recognizes SM clusters rather than monomeric SM ( 42 ), these results indicate SM clustering within the membranes where WNV replicates, thus reinforcing the idea of membrane remodeling during WNV infection ( 7 ). Because WNV replication takes place on  ( 51,52 ). Among the lipids co-opted by viruses for this process, sphingolipids are of growing interest ( 17,18 ). In fact, an upregulation of sphingolipid synthesis occurs in cells infected with WNV ( 8 ) and other related viruses ( 12,19 ). However, to our knowledge, the relevance of sphingolipids in fl avivirus pathogenesis in vivo remains unexplored. Our results showed that WNV infection was exacerbated in ASMko mice displaying both increased viral replication and enhanced lethality. Histological analyses confi rmed that the viral content is increased in the brain of ASMko mice, which is but absent in SPK-601. The analysis of the cellular ATP content in drug-treated cells confi rmed that these drugs inhibited WNV at concentrations that exerted negligible cytotoxicity ( Fig. 7D ), pointing to SMS as a feasible druggable antiviral target against WNV.

DISCUSSION
Plus-strand RNA viruses rearrange intracellular membranes to generate specialized organelle-like structures for Fig. 6. Localization of SM within infected cells. A: Vero cells were infected with WNV (MOI, 10 PFU per cell), fi xed, and processed for immunofl uorescence and confocal microscopy at 24 h p.i. SM (green) was detected by lysenin staining. Viral replication complexes were labeled using an antibody against dsRNA (red). Uninfected cells were included as a negative control. Insets show higher magnifi cation images from the indicated areas. B: Confocal image of a Vero cell transfected with plasmid IgLdR1kdel to label the ER and infected with WNV (MOI, 10 PFU per cell) at 24 h post transfection. Cells were fi xed and processed for immunofl uorescence as described for (A). Left panel displays ER fl uorescence to indicate cell shape. Insets show higher magnifi cation images of the ER (red), SM (green), and dsRNA (blue). White spots indicated by circles denote colocalization among the ER marker, SM, and dsRNA. Bars, 10 m. * P < 0.05; ** P < 0.005. lack of ASM activity leading to the accumulation of SM promotes WNV infection.
To further investigate the role of SM in WNV infection, we reproduced the SM accumulation induced by the lack of ASM activity by adding exogenous SM to cell cultures ( 23,29 ). The quantifi cation of SM content confi rmed that treatment with SM was enough to increase cellular SM levels. In these experiments, we noticed that the addition of SM increased WNV multiplication. Thus, we propose that SM positively regulates WNV infection. In fact, the results shown in the present report indicate that a higher content of cellular SM (ASMko mice, NPA fi broblasts, or addition of exogenous SM) promoted WNV infection. This effect of SM on WNV infection is also consistent with the recently reported positive effect of SM on hepatitis C virus replication ( 19,54 ).
On the other hand, treatment with pharmacological inhibitors of SM synthesis, which would be expected to decrease cellular SM levels, resulted in an inhibition of infectious virus production. Interestingly, the extent of the inhibition of infectious particle production was higher than that of the inhibition of RNA replication, suggesting that production of infectious particles could be more sensitive to SM depletion than the replication of the viral genome. These results also support a functional role of SM on the infectivity of WNV particles, which is consistent with our previous observation of an enrichment of this lipid in the viral envelope of WNV ( 8 ).
Moreover, intracellular SM detected by lysenin staining was found associated to dsRNA and also to an ER marker by confocal analyses of WNV-infected cells. These results support previous data pointing to the clustering and localization of this lipid within the replication complexes of particularly enriched in SM ( 22,23,36 ). Furthermore, the increased ability of WNV to replicate in neural tissues from ASMko mice was also confi rmed ex vivo using organotypic hippocampal slice cultures. It has to be considered that neurons, which constitute a major target for WNV replication in the nervous system ( 53 ), are specially enriched in SM ( 22 ), which may also reinforce the relevance of this lipid's levels in WNV infection. Although increased levels of other lipids, such as cholesterol or ceramide, can also promote WNV multiplication in cultured cells ( 7,9 ), this possibility was excluded in ASMko mice because neither cholesterol nor ceramide are enriched in their brains ( 36 ). Considering all these factors, the results observed for ASMko mice provide solid evidence of the involvement of the SM pathway in WNV infection and in vivo pathogenesis.
Dermal fi broblasts are one of the fi rst cell types supposed to be exposed to WNV during a blood meal by an infected mosquito and, hence, represent a relevant model for WNV infection ( 40 ). Therefore, to investigate the effect of the alteration of lipid metabolism on WNV lifecycle, we analyzed the infection in fi broblasts from NPA patients. The quantifi cation of SM confi rmed that, as expected, NPA fi broblasts displayed increased levels of SM compared to control fi broblasts. Furthermore, lysenin staining showed that this accumulation of SM was also patent in nonlysosomal membranes such as the plasma membrane, which is consistent with previous data showing that the defi ciency of ASM causes an accumulation of SM in both lysosomal and nonlysosomal membranes ( 23 ). Remarkably, our results confi rmed that WNV multiplication, including RNA replication and infectious virus production, was increased in NPA fi broblasts, which again support that the related fl aviviruses ( 12 ). Flavivirus replication complexes, including those of WNV, are developed in highly modifi ed membranes derived from the ER (3)(4)(5)(6), and both RNA replication and particle biogenesis are coupled in these membranous structures ( 4 ). Along this line, because both the fl aviviral envelope and the replication complex are enriched in SM ( 8,12 ), we propose that SM could positively contribute to both RNA replication and particle biogenesis during WNV infection.
In summary, we have provided evidence showing that sphingolipid metabolism, and especially SM, can modulate WNV infection both in vitro and in vivo, thus making SM a key factor in the WNV lifecycle. Because sphingolipid metabolism is currently considered as a suitable target for antiviral development (55)(56)(57), our data support the potential of modulating SM levels as an antiviral strategy to combat WNV.