Fig 1.
Sphingolipids and influenza virus infections.
Several studies have demonstrated that inhibition of distinct enzymes in the sphingolipid pathway results in alterations to influenza virus infection levels. Pharmacological inhibition of sphingomyelin synthesis (through serine palmitoyltransferase) and sphingosine kinase, as well as genetic ablation of sphingomyelin synthase and glucosylceramidase (red Xs) led to decreased influenza virus infection [23–25,34]. Conversely, reductions in ceramide synthesis through inhibition of ceramide synthase led to an increase in influenza virus replication [22]. In this study we tested the role of UGCG (green box) in influenza virus infection by using the pharmacological inhibitor PPMP as well as by knocking out the gene encoding for UGCG enzyme expression in two cell lines.
Fig 2.
CRISPR/Cas9 mediated knockout of glucosylceramide synthase.
(A) HEK 293 cells were pretreated with 20 μM PPMP (for 48 hours) or 100 nM bafilomycin (for 1 hour) and then infected with PR8 influenza virus encoding an NS1-GFP chimeric protein, in the presence of the indicated drug, for 18–24 hours (selected time points chosen after optimization). Cells were then lifted, fixed, and analyzed by flow cytometry for GFP expression. PPMP-treated samples exhibited a 50% reduction in GFP signal compared to WT, indicating a role for UGCG in influenza virus infection. Data represent the mean values of 4 biological replicates (each performed in triplicate) ± SE. (B) HEK 293 and A549 cells were transfected with plasmids encoding GFP as well as Cas9-sgRNA targeting UGCG. Cells were selected for GFP expression and single cell colonies were expanded and monitored for UGCG knockout as described in the Methods. (C,D) Selected cell clones (see S1 Fig) were assayed for UGCG activity by incubating cells with 5 μM C6-ceramide nanoliposome for 4 hours. Cells containing functional UGCG are able to convert C6-ceramide to C6-GlcCer, as seen in WT samples. HEK 293 and A549 UGCG KO cells displayed no C6-GlcCer, indicating a complete loss of UGCG activity. (E,F) Lipids from WT and KO cells were analyzed by mass spectrometry. In agreement with the measured enzyme activity (C,D), levels of total basal endogenous GlcCer were significantly reduced in both HEK 293 and A549 KO cells as compared to WT. Data represent the mean ± SE (n = 6 samples). ** p<0.01 using a Mann-Whitney non-parametric test.
Fig 3.
A549 UGCG KO cells exhibit haploinsufficiency.
(A) Relative loss of UGCG expression was confirmed in HEK 293 cells by western blot analysis. In comparison, A549 UGCG KO cells (based on DNA analysis; see Methods) displayed only a reduced level (not an absence) of UGCG protein on Western blots. Since we detected no UGCG activity in these cells (Fig 2D and 2F), they are functionally null for UGCG and therefore haploinsufficient. (B) Next generation sequencing indicated the A549 UGCG KO line was heterozygous, containing one WT allele and one allele with a 14 base pair insertion in the gene encoding UGCG. This analysis supports the proposal that the loss of UGCG activity seen in Fig 2D and 2F is the result of haploinsufficiency. (C) Sequence analysis of the CRISPR-modified UGCG allele in A549 cells revealed a frameshift mutation beginning at the codon for amino acid N68 and terminating with an early stop codon at the position of C86.
Fig 4.
Glucosylceramide synthase regulates influenza virus reinfection.
Cells were infected with influenza virus as in Fig 2A, and analyzed 18–24 hours later by flow cytometry. (A) HEK 293 UGCG KO cells exhibited an ~40% reduction in influenza virus infection as compared to WT. (B) A549 UGCG KO cells exhibited ~70% reduction in influenza virus infection as compared to WT. (mean ± SE; n = 6). ** p<0.01 using a Mann-Whitney non-parametric test.
Fig 5.
UGCG maintains optimal entry of VLPs bearing the glycoproteins of VSV, WSN influenza virus and EBOV.
VLPs were generated on an influenza virus βlaM1 backbone with the indicated viral glycoprotein. VLPs were added to prechilled cells which were then centrifuged at 4° for 1 hour. Next the cells were incubated for 3 hours at 37°, and then for 1 hour at room temperature in the presence of the βlaM substrate CCF2. Cells were washed, stored in the dark at room temperature, and (the following day) harvested, fixed, and analyzed for β-lactamase activity via flow cytometry. (A,B) Entry by VLPs bearing VSV G was reduced in HEK 293 UGCG KO cells, but unaffected in A549 KO cells. WSN influenza virus glycoprotein-mediated entry was reduced in both KO cell lines, consistent with the findings in Fig 4. Entry mediated by the EBOV glycoprotein was reduced in both 293 and A549 UGCG KO cells, and to a greater extent than seen with VLPs bearing the glycoproteins from WSN or VSV (mean ± SE; n = 6). ** p<0.01 using a Mann-Whitney non-parametric test.
Fig 6.
Effects of loss of UGCG on infections by VSV pseudoviruses bearing the glycoproteins of measles virus, VSV, and EBOV.
Pseudoviruses were generated using a VSV helper virus encoding GFP and the indicated viral glycoprotein(s). Pseudoviruses were then adhered to prechilled cells assisted by centrifugation at 4°C for 1 hour. The cells were then washed, incubated at 37°C and, the following day, harvested, fixed, and analyzed for GFP expression by flow cytometry. (A,B) Infection by VSV pseudoviruses bearing the measles virus H and F proteins was unaffected in HEK 293 UGCG KO cells, and increased in the A549 UGCG KO cells. Infection by VSV pseudoviruses bearing the VSV glycoprotein was decreased in HEK 293 UGCG KO cells but unaffected in A549 UGCG KO cells, consistent with the findings in Fig 5. Infection by VSV pseudoviruses bearing the EBOV glycoprotein was decreased in both UGCG KO lines tested (mean ± SE; n = 6), also consistent with the results in Fig 5. ** p<0.01 using a Mann-Whitney non-parametric test.