Synaptic vesicle-like lipidome of human cytomegalovirus virions reveals a role for SNARE machinery in virion egress

Modest Impact of HCMV on Total Cell Glycerophospholipids.

Using mass spectrometry, we monitored 146 unique glycerophospholipids, with acyl chains containing a total of 30–42 carbons, over the course of the HCMV replication cycle. Confluent fibroblast cultures were fed with serum-free medium at 24 h before infection and maintained in the absence of serum throughout the infection cycle. This treatment synchronized the cells in G0, reducing cell-to-cell variation in response to infection, and it removed a complex source of extraneous metabolites (2). A comparison of mock-infected vs. virus-infected fibroblasts is displayed in Fig. 1A, where each data point represents the average concentration of a single lipid at a specific time. Complete quantitative data for each lipid species is available in Dataset S1.

Glycerophospholipids in mock-infected samples exhibited variable but modest changes over the time course. The glycerophospholipids generally showed less than twofold changes in response to infection. The largest changes were increases in saturated and monounsaturated PA species; for example, PA (32:0) increased with time after infection, exhibiting an approximately fourfold increase relative to mock-infected cells at 96 hpi (P = 0.013; Fig. 1B). Multiple long chain PC species increased by typically 15–40% (e.g., at 96 hpi 38:6 P = 0.012, 40:6 P = 0.015), whereas several short chain PC species decreased ≈15–30% (e.g., at 96 hpi 32:0 P = 2.2E-4, 34:2 P = 0.0028). PI levels generally dropped over the course of infection, with PI (38:4) showing the largest reductions from basal levels (approximately twofold, P = 0.0054 at 96 hpi) during later stages of infection. Saturated and monounsaturated PG species generally increased with infection (e.g., 32:0, 32:1, 34:0, 34:1 all by more than 60% by 96 hpi, P < 0.05 by two-way ANOVA), whereas many polyunsaturated fatty acid species of PG and PS decreased. In summary, HCMV infection induced reproducible but generally modest changes in the levels of cellular glycerophospholipids.

Virion Glycerophospholipid Composition Is Different from That of Whole-Cell Lysates.

Virions were purified from the medium of infected fibroblast cultures by velocity sedimentation in glycerol-tartrate gradients (26), and judged to be substantially free of cellular membrane contaminants by electron microscopy (Fig. S1). Analyses of three independent virion preparations (Dataset S2) showed modest differences between virions and mock-infected or HCMV-infected cells in terms of acyl chain length (for measured lipids, which contained a total of 30–42 carbons in the fatty acid chains) or degree of unsaturation (Fig. 2A). The head group composition of the glycerophospholipids from the virion was, however, substantially different from that of the infected host (Fig. 2B). Whereas PE constituted ≈21% of glycerophospholipids assayed in whole-cell lysates, it was the dominant head group (≈48%) in purified virions. PS, PA, and PI were lower by factors of approximately 3, 2, and 2, respectively, whereas PC levels did not change the ratio. Interestingly, only one species each of PA and PG was detected in virions: PA (38:4) and PG (36:2). PA (38:4) was higher in infected compared to mock-infected cells by a factor of approximately 3.5, and PG (36:2) did not change significantly in infected cells through 96 hpi (Dataset S1).

Finally, the virion PE population contained multiple plasmalogen species that were found in lesser amounts or not detected in cell lysates. Plasmalogens are lipids in which the fatty acid linked to the first carbon of the glycerol backbone is attached through an ether or vinyl-ether bond rather than an ester bond. Although the fraction of PEs that were plasmalogens was lower in virions than in cells (19% and 28% of their total PE populations, respectively) (Fig. S2A), given the large increase in PEs in the virions, the representation of plasmalogen PEs as a percentage of the total glycerophospholipid pool was higher in virions (Fig. S2B). The distinct composition of the virion envelope relative to that of whole cells argues that the envelope is derived from a specialized cellular membrane.

SNAP-23 Is Required for the Efficient Production of HCMV Progeny.

By comparing the glycerophospholipid composition of HCMV virions with published datasets, we noted a resemblance to synaptic vesicles (20). HCMV virions and synaptic vesicles contain similar ratios of phospholipid head group classes (i.e., both have higher PE content and lower PS). In addition, both are enriched for PE plasmalogens. These similarities suggest that HCMV might acquire its envelope by budding into vesicles within fibroblasts that are similar in lipid composition to neuronal synaptic vesicles.

Synaptic vesicles are specialized trafficking vesicles that store neurotransmitters and undergo Ca²⁺-dependent exocytosis in response to an action potential. SNAP-25 is a key component of the SNARE complex that mediates exocytosis of synaptic vesicles in neurons and hormonal exocytosis in endocrine cells (reviewed in ref. 27). SNAP-23 is a SNAP-25 homolog that is expressed in many cell types (28), including primary human fibroblasts (29). Given the role of SNAP-25 in fusion of synaptic vesicles, the close relationship of SNAP-23 to SNAP-25, and the presence of SNAP-23 in fibroblasts, we tested the possibility that SNAP-23 was used by HCMV during its assembly/egress process. SNAP-23 was quantified by Western blot, and its levels did not change substantially with time after infection (Fig. 3A). Virus-coded IE-1 and pUL99 were assayed to confirm the progression of infection, and β-actin was used as a loading control. Immunofluorescent analysis (Fig. 3B) showed that SNAP-23 was spread through the cytoplasm in mock-infected cells, presumably owing to its association with vimentin filaments (29). In contrast, at 96 hpi, a substantial portion of SNAP-23 colocalized with pUL99 and pUL55 (gB), markers for the HCMV assembly compartment, the site at which the virion acquires an envelope (9–12).

The functional significance of the relocalization was probed by performing a knockdown experiment with siRNA against SNAP-23 (30). A Western blot showed a substantial decrease in SNAP-23 levels relative to the nontargeting control using β-actin for normalization, and the cells remained competent to express the IE1 protein after infection with HCMV (Fig. 4A). However, three independent experiments showed an ≈50% decrease in the production of extracellular virus in knockdown compared with normal cells (Fig. 4B). In these experiments the siRNA was introduced into fibroblasts by transfection, and, consequently, some cells did not receive the siRNA and produced a normal yield of virus. Accordingly, we repeated the knockdown experiment using three different shRNAs expressed from a lentivirus vector. The most efficient knockdown occurred with shRNA1 (≈85%; Fig. 4C), and it was used to generate cultures in which SNAP-23 was reduced in all cells. Upon infection, these cells produced markedly less extracellular virus; the yield of infectious virus was reduced by a factor of approximately 1,000 at 120 hpi (Fig. 4D). Immunofluorescent analysis demonstrated that the HCMV pUL99 late protein accumulated and properly localized to the assembly compartment (Fig. 4E), arguing that the infection progressed to a very late stage but failed to generate normal levels of infectious progeny in the knockdown cells.

Virus Growth and Purification.

Primary human foreskin fibroblasts and MRC5 fibroblasts (ATCC) were cultured in medium containing 10% FBS. HCMV infections used the AD169 strain BADwt (56). Virus stocks were produced in fibroblasts by pooling cell-free and cell-associated virus, concentrated by centrifugation through sorbitol cushions, resuspended in serum-free medium, and titered by tissue culture infective dose (TCID₅₀) or fluorescent focus assay.

Sample Preparation and Mass Spectrometry.

For analysis of cellular lipids, primary fibroblasts were serum starved for 24 h, infected at a multiplicity of 3.0 TCID₅₀ units per cell, and then maintained in serum-free medium. To extract phospholipids, 2 × 10⁶ cells were washed twice with ice-cold PBS, scraped into 800 μL of ice-cold 0.1 N HCl:CH₃OH (1:1), and transferred into a cold microfuge tube. After addition of 400 μL of ice-cold CHCl₃, samples were mixed by vortexing for 1 min at 4 °C, and phases were separated by centrifugation in a microfuge (18,000 × g for 5 min, 4 °C). The lower organic layer was isolated, synthetic odd-carbon phospholipids (four per each phospholipid class) were added as internal standards, and solvent was evaporated. The resulting lipid film was dissolved in 100 μL of isopropanol (IPA):hexane:100 mM NH₄COOH_(aq) 58:40:2 (mobile phase A) (14).

For analysis of virion lipids, virus was purified as previously described (26). Serum-starved, primary fibroblasts were infected at a multiplicity of 0.1 TCID₅₀ units per cell, maintained in serum-free medium, and supernatant was harvested 6 d later. Cell debris was removed by centrifugation, bacitracin was added to prevent clumping, and virus particles were pelletted by centrifugation through the 20% sorbitol cushion at room temperature. Pellets were resuspended in buffered saline, and virions were resolved by centrifugation in sodium tartrate gradients. The virion band was removed with a syringe, diluted, and purified through a second round of centrifugation. Purified virions were resuspended in buffer containing 100 mM NaCl and 50 mM Tris·HCl, pH7.4. Infectivity of the product was quantified by TCID₅₀ assay, and purity was assessed by electron microscopy (Fig. S1). Phospholipids were extracted from an average of 3.4 × 10⁶ TCID₅₀ of purified virions by using acidified methanol followed by chloroform extraction, as described for cells above.

Mass spectrometric analysis and quantitation were performed essentially as previously described (14). An MDS SCIEX 4000QTRAP hybrid triple quadrupole/linear ion trap mass spectrometer (Applied Biosystems) was used for the analyses. Coupled to it was a Shimadzu HPLC system (Shimadzu Scientific Instruments) consisting of an SCL 10 APV controller, two LC 10 ADVP pumps, and a CTC HTC PAL autosampler (Leap Technologies). Phospholipids were separated on a Phenomenex Luna Silica column (2 × 250 mm, 5-μm particle size) using a 20-μL sample injection. A binary gradient consisting of IPA:hexane:100 mM NH₄COOH_(aq) 58:40:2 (mobile phase A) and IPA:hexane:100 mM NH₄COOH_(aq) 50:40:10 (mobile phase B) was used for the separation. Instrumentation parameters and solvent gradient were as previously described (14). Data are presented as means plus SEs based on three independent experiments. Statistical analysis was performed by ANOVA. For time courses, a two-way ANOVA using factors of time and virus state (infected, uninfected) was used, with post hoc t test at individual time points.

Protein Analysis.

For Western blots, MRC5 cell pellets were resuspended in lysis buffer (150 mM NaCl, 1% nonidet-p40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris pH 8.0, and Roche protease inhibitor mixture), sonicated, rocked for 30 min at 4 °C, and subjected to centrifugation in a microfuge to remove insoluble material. Equal amounts of protein were separated by electrophoresis in SDS-containing 12% polyacrylamide gels and transferred to nitrocellulose membranes, which were then probed with antibodies. For immunofluorescence, cells were plated on glass coverslips and mock infected or infected with HCMV. After various time intervals, cells were fixed and permeabilized with methanol. Cells were blocked and incubated with mouse or rabbit antibodies. The following antibodies were used: HCMV IE-1 [1B12 (57)], pUL99 [10B4-29 (9)], pUL55 gB (Goodwin Institute), SNAP-23 (111-202; Synaptic Systems), β-actin-HRP (ab49900; Abcam), and GAPDH (G8795; Sigma-Aldrich).

Knockdown of SNAP-23.

MRC5 fibroblasts were transfected with siRNA (10 nM) directed against SNAP-23 (5′CCAACAGAGAUCGUAUUGATT3′) (30), an siRNA specific for IE-2 (5′AAACGCAUCUCCGAGUUGGACTT3′) (58), or the Mission siRNA Universal Negative Control #1 (Sigma-Aldrich) by using Oligofectamine (Invitrogen) in Optimem (Gibco). After 6 h, transfection reagent was removed and replaced with medium containing 10% FBS. Twenty-four hours later, cells were infected and analyzed at 72 or 96 hpi. shRNAs in pLKO.1-puro (Sigma-Aldrich) also were used for knockdown of SNAP-23. MRC5 cells were infected with the lentivirus vector, selected with 500 μg/mL G418, and then used for analysis. shRNA1 (TRCN0000144789) was chosen for further study (5′CCGGGCAATGAGATTGATGCTCAAACTCGAGTTTGAGCATCAATCTCATTGCTTTTTTG-3′).