Amphotericin B increases influenza A virus infection by preventing IFITM3-mediated restriction

Abstract

The IFITMs inhibit influenza A virus (IAV) replication in vitro and in vivo. Here, we establish that the antimycotic heptaen, amphotericin B (AmphoB), prevents IFITM3-mediated restriction of IAV, thereby increasing viral replication. Consistent with its neutralization of IFITM3, a clinical preparation of AmphoB, AmBisome, reduces the majority of interferon's protective effect against IAV in vitro. Mechanistic studies reveal that IFITM1 decreases host-membrane fluidity, suggesting both a possible mechanism for IFITM-mediated restriction and its negation by AmphoB. Notably, we reveal that mice treated with AmBisome succumbed to a normally mild IAV infection, similar to animals deficient in Ifitm3. Therefore, patients receiving antifungal therapy with clinical preparations of AmphoB may be functionally immunocompromised and thus more vulnerable to influenza, as well as other IFITM3-restricted viral infections.

Graphical abstract

Figure 1

AmphoB Increases IAV Infection by Preventing IFITM3-Mediated Restriction In Vitro (A) HeLa cells stably expressing either of two negative control short hairpin RNAs (shRNAs) against the firefly luciferase gene (shLuc-1 or shLuc-2) or shRNAs against IFITM3 (shIFITM3-1 and shIFITM3-2) were incubated for 1 hr in the presence (red) or absence (blue) of 1 μM AmphoB, then challenged with IAV WSN/33 (multiplicity of infection [moi] 0.2) or PR8 (moi 0.02) followed by immunostaining for HA. Numbers represent the mean percentage of infected cells of three separate experiments ± SD, as determined by imaging analysis software. (B) Whole cell lysates from the indicated cells in (A) were subjected to immunoblotting using the noted antibodies. kDa, kilodaltons. RAN serves as a loading control. (C) IFITM3’s restriction of HA-mediated entry is overcome by AmphoB. A549 cells stably transduced with IFITM3 or with vector alone (Vector) were treated with 1 μM AmphoB (red) or buffer (blue) then challenged with the indicated pseudotyped MLV-GFP particles: WSN/33 (H1), PR8 (H1), Thai (H5), FPV (H7), or the g protein of VSV. Relative infection represents the percentage of GFP-positive cells normalized to that of A549-vector cells as determined by fluorescence-based imaging. Results are the mean of three independent experiments ± SD. (D and E) A549-vector (blue) or A549-IFITM3 (red) cells were infected with increasing amounts of IAV (WSN/33, relative moi) in the presence or absence of AmphoB (D) or AmBisome (E). Infectivity was determined by immunostaining for surface HA protein. (F) IAV infection (WSN/33, moi 2) of A549-vector (blue) or A549-IFITM3 (red) cells in the presence (closed symbols) or absence (open symbols) of AmBisome or IFN-α. (G) Structure of the AmphoB derivative DS-AmpB-020. (H) A549-vector (Vec, open symbols) or -IFITM3 (M3, closed symbols) cells were infected with IAV (WSN/33, moi 2) in the absence (buffer, green lines) or presence of increasing concentrations of either AmphoB (blue) or DS-AmpB-020 (red). Results are representative of three independent experiments.

Figure 2

AmphoB Prevents IFITM2- and IFITM3-Mediated Restriction, but Not Restriction by IFITM1 (A) HeLa-vector or HeLa-IFITM3 cells were incubated with AmphoB or buffer for 1 hr prior to being incubated on ice with IAV PR8. Warm media was added at time zero. Cells were fixed at the indicated time points, immunostained for the IAV NP protein (green) and stained for DNA, then analyzed by confocal microscopy. Image analysis software was used to define each cell’s nuclear peripheries (blue lines based on the staining of nuclear DNA). Numbers represent the number of NP particles (green) present per nucleus at the indicated time points and are the mean ± SD of three independent experiments. Scale bar, 10 μM. (B) A549 cells stably transduced with vector alone (Vector) or with IFITM1 (M1), IFITM2 (M2), or IFITM3 (M3) were incubated for 1 hr with increasing concentrations of AmphoB followed by infection with IAV WSN/33 for 12 hr and immunostaining for HA. (C) Schematic diagram of the IFITM chimeras. The N- and C-terminal domains (NTD, CTD) and intramembrane domains 1 and 2 (IM1 and IM2) are shown, separated by the conserved intracellular loop (CIL). M1, IFITM1 (gray); M2, IFITM2 (green); M3, IFITM3 (red). The chimeric proteins (M3M1 and M1M3) are shown below, with the first M# representing the NTD’s origins and the second M# denoting the consignation of the remaining portion of the protein. (D) A549 cells stably transduced with vector alone (Vector) or the indicated chimeras were incubated for 1 hr with increasing concentrations of AmphoB prior to infection with IAV WSN/33 as above. (E) Immunoblot of the indicated cell lines using the CIL antisera, which recognizes an epitope that is identical across all of the IFITM WT proteins and chimeras. Results are representative of three independent experiments.

Figure 3

IFITM3’s Inhibition of IAV Is Overcome by AmphoB and Nystatin, but Not by Na⁺ Ionophores (A) HeLa-vector or HeLa-IFITM3 cells were incubated with the indicated compounds and stained for endosomal sodium with the fluorescent indicator dye Asante Na⁺ green 2 (ANG-2, green) and then imaged using confocal microscopy. The concentrations used were AmphoB 1 μM, nystatin 50 μM, and SI-X, SQI-Et, SQI-Pr 5 μM. Scale bar, 10 μm. (B) IAV WSN/33 infection (moi = 2) of A549-vector (blue) or A549-IFITM3 (red) cells in the presence or absence of increasing concentrations of the indicated compounds. Cells were treated with the indicated compounds for 1 hr prior to infection. (C) Schematic diagram of AmphoB or nystatin transmembrane pores (left) permitting the passive diffusion of multiple monovalent cations (purple and red ovals) across the endosomal membrane (arrow indicates high to low concentration gradient). In comparison, the ionophores selectively bind Na⁺ (red ovals) in a central region, effectively shielding the positive charge from the hydrophobic interior of the membrane and thus allowing the cation to be transported down the gradient. Adding TEA or ACh blocks the opening of the AmphoB pore, preventing ion transport (lower panel, gold oval). (D) Confocal images of the relative levels of endosomal sodium based on the fluorescent signal of ANG-2 (green) in A549-vector or A549-IFITM3 cells with or without treatment with AmphoB, TEA, or ACh. Quantitation provided in Figure S2C. Scale bar, 10 μm. (E) IAV WSN/33 infection of A549-vector (blue) or A549-IFITM3 (red) cells in the presence or absence of AmphoB, TEA, or ACh. Serial dilutions of TEA or ACh were added to cells 1 hr prior to infection. Results are representative of three independent experiments throughout.

Figure 4

Role of Cholesterol or VAPA in IFITM-Mediated Restriction of IAV (A) IAV WSN/33 infection (moi 1) of A549-vector (Vec), A549-IFITM1 (M1), or A549-IFITM3 (M3) cells, treated with the indicated small molecules, in the absence (Buffer, blue) or presence of AmphoB (1 μM, red). (B) IAV WSN/33 infection of A549-vector (blue), A549-IFITM1 (red), or A549-IFITM3 (green) cells treated with the indicated compounds prior to infection. (C) HeLa cells were stably transduced with retroviruses expressing either a negative control shRNA against the firefly luciferase gene (shLuc-1), a shRNA against IFITM3 (shIFITM3-1), or the VAPA cDNA (for the latter, two independently derived polyclonal cell populations, VAPA-1 and VAPA-2, are shown). Cells were then treated with (red) or without IFN-α (blue) for 16 hr and then infected with IAV WSN/33. Results represent the mean of three independent experiments ± SD. (D) Whole cell lysates from the indicated cells in (C) were subjected to immunoblotting using the indicated antibodies. RAN serves as a loading control. (E) Infection of A549-vector (blue), A549-IFITM3 (red), or A549-IFITM3+VAPA (green) cells with increasing amounts of IAV PR8 (left) or IAV WSN/33 (right). Results are representative of three independent experiments. (F) Whole cell lysates from the indicated cells in (E) were subjected to immunoblotting using the indicated antibodies. Actin serves as a loading control. (G) Niemann-Pick type c1 (NPC) or wild-type (WT) primary human fibroblasts infected with the indicated IAVs with or without AmphoB (Ampho) treatment. (H) Confocal images of primary human fibroblasts in (G) stained with filipin (white) to detect cholesterol and immunostained for LAMP1 (red) to identify the late endosomal and lysosomal compartments. Scale bar = 15 μM.

Figure 5

IFITM1 Decreases Membrane Fluidity and Inhibits Cell-to-Cell Fusion (A) COS-7 cells were stably transduced with retroviruses expressing the empty vector (Vector), IFITM1, CAV1-teal fluorescence protein (TFP), or IFITM3. Cells were then confocally imaged to detect the expressed proteins (all in green; additional specificity control images in Figure S4A). Nuclear DNA was stained with DAPI (blue). Scale bar, 10 μM. (B) Fluorescence recovery after photobleaching of the plasma membrane. COS7-vector or COS7-IFITM1 cells were stained with DiO and the unbound dye removed. Cells were photobleached in the indicated area (white square) with a 488 laser, and a series of pictures out to 95 s postbleach were taken. Images are shown for prebleach, postbleach, and 17 s postbleach (t^1/2 of IFITM1 recovery). (C) COS7-vector, COS7-CAV1, or COS7-IFITM1 cell fluorescence recovery was quantified using Leica Lite software, with each individual time point representing the average of ten normalized readouts. PRISM software was then used to plot the best-fit exponential decay line. Results are representative of three independent experiments. (D) The COS7-vector, COS7-CAV1, or COS7-IFITM1 percent immobile fraction was calculated by first subtracting the normalized average of the 95 s postbleach images, then subtracting those values from 1 to determine the amount of membrane that did not recover. Results are representative of three independent experiments. (E) Schematic of IAV-induced cell-to-cell fusion assay. Individual cells (lower portion) exhibit distinct boundaries of actin (red lines) and clearly spaced nuclei (blue ovals). By comparison, after incubation with IAV and the addition of pH 5.0 buffer, the individual cells fuse and form syncytia, which possess multiple clustered nuclei with no intervening actin boundaries. (F) IAV-induced cell-to-cell fusion assay. Chilled IAV (PR8, moi 500–1,000) was incubated with the indicated stably transduced COS-7 cell lines from (A). Warm buffer of either pH 5.0 or 7.5 was added, followed by warm media. Cells were incubated for 5 hr and then fixed and stained to monitor fusion events for both actin (phalloidin, red), to highlight cellular boundaries (white dashed lines), and DNA (blue), to detect closely clustered nuclei present in syncytia. Scale bar, 10 μm. (G) Percentage of fused cells in experiments using pH 5.0 buffer in (F). Fusion events per 100 nuclei were calculated and values represent the average of three to five images from each of three independent experiments ± SD. (H) Immunoblot of the indicated cell lines from (A). The CIL antisera recognize an epitope that is identical in both IFITM1 and IFITM3.

Figure 6

AmBisome Increases Influenza Morbidity and Mortality (A and B) Change in body mass (A) and survival (B) of WT and Ifitm3^−/− mice with or without intravenous administration of AmBisome (3 mg/kg) on days 0, 2, and 4 relative to intranasal inoculation with IAV X-31 (10,000 pfu); n > 3. (C) WT and Ifitm3^−/− mice were treated or not treated with AmBisome and then challenged with X-31 influenza as above. At day 6 postinfection, lung sections were prepared and stained with hematoxylin and eosin. (D) The indicated MEFs were incubated for 1 hr in the presence (red, +) or absence (blue, −) of 1 μM AmphoB and then challenged with X-31 or pandemic A/California/7/2009 (pH1N1) followed by fixation and immunostaining for NP. Numbers represent the mean percentage of infected cells of three separate experiments ± SD. (E) Immunoblot of the indicated cell lines from (D). (F) Model of IFITMs inhibiting viral fusion. Top panels depict HA-directed membrane fusion. Insertion of the viral fusion peptide (red) by the HA protein into the host’s endosomal membrane is followed by an acid-induced conformational change in HA producing a hemifusion transition state (middle, arrows represent force vectors), followed by a fusion pore (right). Arrows represent force vectors throughout. We postulate that IFITM proteins prevent fusion pore formation (bottom panels); situated within the cytosolic-facing leaflet, multiple interacting IFITM proteins act as reinforcing bars to decrease membrane flexibility and prevent the viral fusion machinery from distending the host membrane (red arrows). The intramembrane insertions of each IFITM into the cytosolic leaflet also inhibit fusion by producing a membrane curvature away from the viral fusion machinery. AmphoB (green, rightmost lower panel) is postulated to rescue IAV replication by increasing membrane fluidity and planarity, thereby permitting fusion pore formation.