Rapid Detection of Viral Envelope Lipids Using Lithium Adducts and AP-MALDI High-Resolution Mass Spectrometry

Abstract

There is an unmet need to develop analytical strategies that not only characterize the lipid composition of the viral envelope but also do so on a time scale that would allow for high-throughput analysis. With that in mind, we report the use of atmospheric pressure (AP) matrix-assisted laser desorption/ionization (MALDI) high-resolution mass spectrometry (HRMS) combined with lithium adduct consolidation to profile total lipid extracts rapidly and confidently from enveloped viruses. The use of AP-MALDI reduced the dependency of using a dedicated MALDI mass spectrometer and allowed for interfacing the MALDI source to a mass spectrometer with the desired features, which included high mass resolving power (>100000) and tandem mass spectrometry. AP-MALDI combined with an optimized MALDI matrix system, featuring 2′,4′,6′-trihydroxyacetophenone spiked with lithium salt, resulted in a robust and high-throughput lipid detection platform, specifically geared to sphingolipid detection. Application of the developed workflow included the structural characterization of prominent sphingolipids and detection of over 130 lipid structures from Influenza A virions. Overall, we demonstrate a high-throughput workflow for the detection and structural characterization of total lipid extracts from enveloped viruses using AP-MALDI HRMS and lithium adduct consolidation.

Graphical Abstract

INTRODUCTION

As evident from historic and recent pandemics, for example, the influenza pandemic of 1918¹ and the current COVID-19 pandemic,² enveloped viruses are particularly prone to cause human and animal illness, death, and economic distress worldwide due to their route of transmission and replication strategies.³ The viral envelope serves a variety of important functional roles including processes that facilitate viral replication and transmission.⁴ The viral envelope consists of lipids, lipid aggregates, glycoproteins, and lipid–glycoprotein aggregates.^5,6 The coordinated function of envelope glycoproteins and the structural organization of envelope lipids allows for the virus to fuse with the host plasma or endosomal membranes and enter the host cell and facilitates exit of the replicated virus from the host cell via budding and fission typically with the host membrane.⁷ In addition, the structural organization of the envelope lipids serves to protect the viral RNA content from a hostile extracellular environment.⁸ The “building blocks” of the envelope, which establish structure and functionality, are lipids that have been appropriated typically from host cell membranes. The viral envelope is assembled within the host cell to enhance the structural diversity of lipid aggregates to give a wide range of specific protein–lipid interactions.⁹

Lipid analysis of the viral envelope from a variety of different viruses [including human immunodeficiency virus (HIV), herpes simplex virus (HSV), respiratory syncytial virus (RSV), vesicular stomatitis virus (VSV), and influenza virus (IV)] details a common lipid composition.^8,10,11 Viral membranes are enriched in raft membrane lipids (including cholesterol, sphingomyelin (SM), glycosphingolipids (GSL), and saturated fatty acyls) and proportionally deficient in diacyl glycerophospholipids (GP) containing polyunsaturated fatty acids and storage/energy lipids such as cholesterol esters and triacylglycerides. Viral envelopes are also preferentially biased toward a lipid composition that encourages intrinsic membrane curvature.^8,12 The biophysical rational for differential abundance between viral envelope lipids and host cell lipids is attributed to the geometric restraints needed from the viral envelope to allow the virus to enter and exit the host cell and transmit between cells.⁸ For example, raft membrane lipids, such as cholesterol, sphingolipids (SP), and saturated GP, allow tighter packing of laterally associated lipids and a higher gel–liquid phase transition temperature providing membrane rigidity and vertical/lateral heterogeneity.^13,14 Both factors promote a sturdy membrane barrier and facilitate segregation of receptors required for attachment and fusion.¹⁵ For example, the presence of GSL in the viral envelope have associated functions such as coreceptors for viral entry and regulation of receptor protein dynamics by altering the biophysical properties of the membrane.¹⁶ In addition, the levels of cholesterol in the viral membrane can affect both early and late steps in the membrane fusion process during viral entry and the pathognomonic cell–cell fusion process.¹⁷

The analysis of extracted lipids from a variety of biological matrices, including virions, is readily achieved using mass spectrometry.^18–20 Recently, the use of electrospray ionization (ESI) mass spectrometry (MS) combined with and without liquid chromatography has been used to characterize the lipid composition of several different strains of Influenza.^10,11 Another common gas-phase ionization technique for lipid analysis is matrix-assisted laser desorption/ionization (MALDI).²¹ MALDI MS is advantageous for lipid analysis due to its fast analysis time (∼seconds/spectra), low sample consumption (~μL/spot), adaptability with respect to matrix selection, and high tolerance toward salts/buffers. MALDI MS is also compatible with high-resolution mass spectrometers, amenable to automation, and complementary to ESI platforms.²² Our laboratory has recently demonstrated the effectiveness of using lithium adduct consolidation on a MALDI MS platform for increased sensitivity and decreased spectral complexity for lipid detection.²³ The matrix combination of 2′,4′,6′-trihydroxyacetophenone and lithium (THAP + Li) effectively forced lithium adduct consolidation during the MALDI process, resulting in reduced spectral complexity and enhanced lipid detection, especially for sphingolipids.

MALDI MS is routinely employed for lipid analysis from microscopic organisms such as bacteria,^24–27 yet to date, the use of MALDI MS for lipid characterization of virus particles has received limited attention.^28,29 The MALDI process is typically carried out under vacuum on a dedicated mass spectrometer.³⁰ This design has been shown to be advantageous for sensitive and reproducible detection of a wide variety of biological analytes,^31,32 including lipids.²¹ More recently, the use of atmospheric pressure (AP)-MALDI has gained traction especially in its application to MS imaging (MSI).^33,34 Several advantages of the AP-MALDI platform are its increased accessibility to a larger swath of mass spectrometers, ease of sample preparation when considering volatile analytes and/or matrices prone to sublimation under vacuum, speed of analysis (no waiting for evacuation of ion source chamber to reach desired vacuum), and convenience of readily switching between AP-MALDI and ESI sources. With these advantages in mind and combined with the use of lithium adduct consolidation, we developed an analytical workflow that utilized AP-MALDI configured to a high-resolution mass spectrometer to rapidly detect and characterize envelope lipids from Influenza Virus A (IVA). We describe a method where the distinct advantages of combining AP-MALDI, lithium adduct consolidation, and high-resolution mass spectrometry (HRMS) offer a unique opportunity to rapidly characterize envelope lipids which subsequently offers new insight into the role these lipids play in viral transmission.

MATERIALS AND METHODS

Materials.

LC–MS-grade acetonitrile, methanol, water, n-propanol, formic acid, and ammonium formate were purchased from Fisher Scientific (Pittsburgh, PA). HPLC-grade tert-butyl methyl ether (MTBE) and chloroform (CHCl₃) were purchased from Sigma-Aldrich (St. Louis, MO). 99% purity 2,5-dihydroxybenzoic acid (DHB) was purchased from Acros Organic (Fair Lawn, NJ). 2′,4′,6′-Trihydroxyacetophenone (THAP, MALDI MS grade) and lithium chloride (LiCl) were purchased from Sigma-Aldrich (St. Louis, MO). Lipid standards were purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). The lipid standards included EquiSPLASH LIPIDOMIX Quantitative Mass Spec Internal Standard (330731). Additional lipid standards included glucosyl ceramide (d18:1/16:0) (GlcCer(d18:1/16:0)) (860539) and galactosyl ceramide (d18:1/16:0) (GalCer(d18:1/16:0)) (860521).

Influenza A Virus Sample Preparation.

Influenza virus A/California/04/2009 was propagated and harvested from allantoic fluid using fertilized research grade specific pathogen free (SPF) eggs (Charles River Laboratories, Norwich, CT) following standard protocols.³⁵ Propagated virus was quantified via plaque assays using standard methods.³⁵ Briefly, MDCK cells were infected using 10-fold dilutions then overlaid with 0.7% agarose and incubated for 7 days followed by staining with 0.5% crystal violet staining in 30% MeOH. Plaques were then counted and quantified to determine viral titers.

Virus Particle Lipid Extraction.

Total lipid extracts of pelleted virus particles were prepared using a modified MTBE lipid extraction protocol.³⁶ Briefly, 250 μL of ice-cold methanol and 10 μL of internal standard mixture (EquiSPLASH lipidomix) were added to each pelleted virus particle sample followed by incubation at 4 °C, 650 rpm shaking for 15 min. Next, 500 μL of ice-cold MTBE was added followed by incubation at 4 °C for 1 h with 650 rpm shaking. Cold water (500 μL) was added slowly, and the resulting extract was maintained 4 °C, 650 rpm shaking for 15 min. Phase separation was completed by centrifugation at 8000g for 8 min at 4 °C. The organic phase was removed and set aside on ice. The aqueous phase was re-extracted with 200 μL of MTBE followed by 15 min of incubation at 4 °C with 650 rpm shaking. Phase separation was completed by centrifugation at 8000g for 8 min at 4 °C. The organic phase was removed and combined with the previous organic extract. The organic extract was dried under a steady stream of nitrogen at 30 °C. The recovered lipids were reconstituted in 200 μL of CHCl₃/MeOH (1:1, v/v) containing 200 μM of butylated hydroxytoluene and stored at −20 °C until analysis.

Atmospheric Pressure Matrix-Assisted Laser Desorption/Ionization High-Resolution Mass Spectrometry.

A 0.5 μL of sample/matrix mixture (1:1, v/v) was spotted on a ABI Opti-TOF 192 target plate (Applied Biosystems, Foster City, CA) using a dried-droplet spotting technique. Matrix solutions included 10 mg/mL DHB and 10 mg/mL THAP with 10 mM LiCl. All matrix solutions were made in n-propanol/methanol (1:1, v/v). A Q Exactive HF mass spectrometer (QE HF, Thermo Fisher Scientific, Bremen, Germany) was coupled to a AP-MALDI source³⁷ (MassTech Inc., Columbia, MD). Target-ng software (MassTech) was used to control the XY stage motion and operation of the laser. The source utilized a diode-pumped solid-state laser (λ = 355 nm) operating at a 0.1–10 kHz repetition rate. Maximum laser pulse energy was 3 μJ at 1 kHz repetition rate. A beam attenuator was used to adjust laser energy. The voltage applied between the MALDI plate and inlet capillary of mass spectrometer was 4 kV. The distance between the plate and the capillary was 3 mm. The inlet capillary was set to 400 °C. Each dried-droplet spot was scanned with a 50 μm wide laser spot. The signals from each spot were integrated over 10 s. Mass spectra were acquired in a positive-ion mode with mass resolution up to 240000 at m/z 200 and a mass range of m/z 450–1600. An automatic gain control (AGC) target was set to 5 × 10⁶ with 200 ms maximum injection time. Tandem MS parameters were the following: mass resolution: 15000; AGC: 5 × 10⁶; max injection time: 200 ms; isolation width: 5 Da; HCD: 35 (normalized collision energy, NCE). Data was acquired and analyzed using Tune and Xcalibur software (Thermo Fisher Scientific, Waltham, MA). Univariate analysis was done using Prism 6 (GraphPad, La Jolla, CA). Each sample was spiked with 10 μL of EquiSPLASH prior to lipid extraction. Based on the spiked amount of the EquiSplash standard, we estimated that the concentration per MALDI spot was approximately 10 pmol/per lipid.

Initial lipid identification was based on database searching using LIPIDMAPS with a 10 ppm mass accuracy threshold, which is lenient for Orbitrap data. Although the QE HF was mass calibrated prior to installing the AP-MALDI source, we observed lower mass accuracy for all AP-MALDI experiments. We are currently investigating the reason behind this. The rationale for keeping with the 10 ppm criteria was the confidence in lipid identification when using lithium adducts. For tandem MS experiments, it should be noted that the quadruple on the QE HF can achieve near 1 Da isolation. This is advantageous when performing tandem MS on complex samples. The lower isolation window results in less interference from overlapping m/z values. However, we observed a considerable drop in signal intensity when lowering the isolation window. The nature of the MALDI acquisition (discrete sampling across the MALDI spot) combined with the presence of low abundant species for some lipids of interest necessitated a 5 Da isolation window. The precursor isolation window for several high abundant m/z values was lowered to 1–2 Da, and we observed highly comparable tandem MS spectra when compared to the 5 Da isolation window (i.e., minimal interference from overlapping m/z values). In addition, the use of lithium adduct consolidation reduced spectral complexity which in turn reduced overlapping m/z interferences. Consequently, we used a 5 Da isolation window for all experiments.

We also investigated the possibility of in-source fragmentation resulting from the lithium adduction. This was investigated via the use of authentic standards with (THAP + Li). We did not observe in-source fragmentation for our experimental conditions (data not shown). In addition, the use of THAP, unlike other common matrices (e.g., DHB and CHCA), is considered a “cold” MALDI matrix and is expected to preserve intact lipids during ionization (including GSL’s and some GP, which are susceptible to in-source fragmentation).³⁸ This was consistent with our data (Figure 1).

Figure 1.

AP-MALDI HRMS mass spectra of IVA envelope lipids. (A) Mass spectrum using DHB. (B) Mass spectrum using (THAP + Li). PC: glycerophosphocholine, PE: glycerophosphoethanolamine, PE-P: vinyl ether PE, SM: sphingomyelin, HexCer: hexosylceramide. Precursor ions were colored coded: [M + H]⁺, blue; [M + Na]⁺, green; [M + H − H₂O]⁺, purple; [M + Li]⁺, red. Lipid identification at the sum composition level was done using accurate mass database searching via LIPIDMAPS.

Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry.

Prior to MALDI TOF MS analysis, samples were spotted on a MTP 384 ground steel Bruker MALDI plate using a dried-droplet spotting technique as described above. The same matrices used in the AP-MALDI experiments were used for the MALDI-TOF experiments. MALDI TOF MS analysis was done using an ultra-fleXtreme mass spectrometer (Bruker Daltonics, Bremen, Germany) in positive reflectron mode. Laser properties included medium smartbeam at a frequency of 1 kHz for 5000 shots per sample. Random walk, partial sample mode was enabled. The m/z range was set to 450–1600. Data was analyzed via Bruker’s flexAnalysis Version 3.4.

RESULTS

To develop a rapid detection platform for viral envelope lipids, we combined lithium adduct consolidation and AP-MALDI HRMS. The result was a mass spectrometry-based platform that allowed for rapid, in-depth, and confident identification of lipids extracted from IVA virions.

AP-MALDI MS and Lithium Adduct Consolidation.

Lipid analysis was accomplished with the use of AP-MALDI HRMS which included coupling Mass Tech’s AP-MALDI source to a Thermo QE HF mass spectrometer. As is common for MALDI MS analysis of lipids, the AP-MALDI HRMS platform yielded a mass spectrum detailing the relative abundance of the most abundant lipids present in the total lipid extract of the IVA virions (Figure 1). Our initial analysis used DHB as the MALDI matrix as it is considered the most common and versatile matrix for lipid analysis.³⁹ Using DHB, the AP-MALDI mass spectrum contained high abundant GP, including diacyl glycerophosphocholine (PC), diacyl glycerophosphoethanolamine (PE), and ether PE (PE-P) (Figure 1A). Also present at relatively high abundance were several SP, including hexosylceramide (HexCer) with and without hydroxylation, and SM (Figure 1A).

A common occurrence within MALDI spectra of lipids is the presence of multiple adducts per lipid structure (e.g., observation of both the protonated ([M + H]⁺) and sodiated ([M + Na]⁺) adducts for the same lipid structure).³⁹ The multiadduct phenomenon complicates lipid identification and potentially reduces sensitivity by splitting analyte signal across the ion series. The addition of an exogenous alkali metal source to the MALDI matrix can alleviate the issue of multiadduction, and in the case of lithium chloride (LiCl), lithium adduct consolidation is promoted.²³ Also, the use of THAP as the MALDI matrix in combination with LiCl was shown to be superior for lithium adduction compared to other common matrices including DHB.^23,38 With this in mind, we analyzed the IVA lipid extract using THAP plus 10 mM of LiCl (THAP + Li) as the MALDI matrix. The resulting AP-MALDI HRMS spectrum clearly showed the benefits of lithium adduct consolidation for decreasing spectral complexity and increasing detection sensitivity (Figure 1B). See Figure S1 for comparative mass spectra of Figure 1 detailing absolute abundance, noise, and baseline calculations. The (THAP + Li) mass spectrum had a lower overall absolute abundance with concomitant lower noise and baseline. The reduced absolute intensity of the (THAP + Li) mass spectrum did not hinder sensitivity due to reduced noise and the lower baseline. Lithium adduct consolidation drastically reduced the presence of multiple adducts which in turn substantially simplified lipid identification. At this stage, lipid identification was achieved using accurate mass database searching (<10 ppm, LIP-IDMAPS) and reported as sum composition. The use of lithium adducts allowed for a less stringent mass accuracy threshold.

In addition to the visual inspection of the mass spectra (Figure 1), the differentiation of multiadduction between DHB and (THAP + Li) was visualized by considering percent adduction for a particular lipid subclass (Figures 2 and 3). The two most abundant GP, PC and PE, ionized readily with DHB to both the protonated (~60%) and sodiated (~40%) adducts (Figure 2A,B). This revealed the analyte signal was split almost evenly between the two adducts. Percent adduction of individual PC and PE structures is represented in Figure S2. PC structures displayed relatively high multiadduction variability. This was attributed to the presence of isobaric species of which a portion were a result of overlapping protonated and sodiated adducts. For example, protonated PC(36:4) (m/z 782.5694) is isobaric with sodiated PC(34:1) (m/z 782.5670) (seen in Figure 1A, m/z 782.5600). Note that the mass resolution needed to resolve these isobaric species is >326000. The effective mass resolution for this m/z range in these experiments was approximately 120000. Thus, with insufficient mass resolution the protonated PC(36:4) and sodiated PC(34:1) were not distinguished, leading to ambiguous PC structure assignment.

Figure 2.

Percent adduct abundance for PC and PE lipids using DHB (A, B) and (THAP + Li) (C, D) as the MALDI matrix. [M + H]⁺ = protonated, [M + Na]⁺ = sodiated, and [M + Li]⁺ = lithiated. Mean was calculated using cumulative percent abundance for all identified PC and PE, respectively. CV = coefficient of variation.

Figure 3.

Percent adduct abundance for HexCer and SM using DHB (A, B) and (THAP + Li) (C, D) as the MALDI matrix. [M + H − H₂O]⁺ = protonated minus H₂O, [M + H]⁺ = protonated, [M + Na]⁺ = sodiated, and [M + Li]⁺ = lithiated. Mean was calculated using cumulative percent abundance for all identified PC and PE, respectively. CV = coefficient of variation.

The effect of lithium adduct consolidation when using (THAP + Li) was seen when comparing the percent adduction for the same PC and PE lipids evaluated with DHB (Figure 2). The use of (THAP + Li) effectively pushed the primary adduct to be overwhelmingly lithiated: 66% for PC and 78% for PE (Figure 2C,D). Of significance, lithium adduct consolidation for both PC and PE was associated with low variability (coefficient of variation (CV) < 7%). The introduction of an exogenous adduct source, LiCl, shifted the dependency of adduct formation away from a variable source of sodium (which likely accumulated from sample matrix, solvents, and laboratory ware) to a quantitatively controlled parameter (LiCl concentration). Further confirmation of the advantages of (THAP + Li) for lithium adduct consolidation was seen with individual PC and PE structures (Figure S2C,D); lithium paired with THAP effectively yielded a predominant, consistent lithium adduct.

A similar, yet more pronounced, lithium adduct consolidation effect was observed for SP. Both HexCer and SM were readily detected in the MADLI mass spectra (Figure 1). In MALDI, HexCer readily ionizes to yield three adducts (sodiated ([M + Na]⁺), protonated minus H₂O ([M + H − H₂O]⁺), and protonated ([M + H]⁺)) and SM preferentially ionizes to produce two adducts (sodiated ([M + Na]⁺) and protonated ([M + H]⁺)).^23,39 As seen with DHB as the MALDI matrix, HexCer were identified as an ion series of roughly 2:1:1 of sodiated, protonated minus H₂O, and protonated (Figure 3A). SM were observed as sodiated and protonated in a ratio of nearly 2:3, respectively (Figure 3B). The individual HexCer and SM structures detailed a similar pattern as the collective lipid subclass sum (Figure S3 A-C). The addition of lithium to THAP for MALDI MS analysis predictably forced adduct formation to the lithium adduct with 91% (HexCer) and 80% (SM) effectiveness. This was observed for the lipid subclass as a whole and individual lipid structures (Figure 3C,D and Figure S3D–F). The use of (THAP + Li) effectively and consistently forced lithium adduct consolidation, which simplified the mass spectra and enabled more sensitive detection of sphingolipid structures.

AP-MALDI HRMS.

The complexity of the total lipid extracts from the IVA virions warranted the use of a mass spectrometer capable of delivering high mass resolution. Although the use of (THAP + Li) reduced the spectral complexity and isobaric overlap resulting from multiadduction, there was still the need for high mass resolution to tease apart isobaric m/z values from overlapping isotopic envelopes. For example, the zoomed in m/z region of (800–850) of the AP-MALDI HRMS mass spectrum of IVA envelope lipids using (THAP + Li) displayed the isotopic envelope for several identified HexCer and SM structures (Figure 4). Close inspection of the isotopic envelopes of HexCer(d42:1), HexCer(d41:1-OH), and SM(d42:1) was highly revealing. The monoisotopic peak, denoted as m, of SM(d42:1) (m/z 821.7009) is isobaric with both the m+1 peak of HexCer- (d41:1-OH) (m/z 821.6814, where m = monoisotopic peak and m+1 = second isotopic peak)) and the m+3 peak of HexCer(d42:1) (m/z 821.7172) (Figure 4). The calculated mass resolution where mass resolution is defined as [m/z value at apex/full width half maximum (fwhm)] was greater than 120000 for each of these peaks. At this mass resolution, all three isobaric m/z values were baseline resolved and readily distinguished. To achieve this mass resolution, the spectra were acquired with a data acquisition setting of 240 K resolution. In comparison, when the data acquisition parameters were adjusted to lower resolving power (e.g., 60K and 120 K acquisition settings), the acquired mass spectra showed requisite decreases in mass resolution to the point that isobaric m/z values were not distinguished (Figure S4). As expected, we observed a near linear decrease in mass resolution. For the example above, mass resolution at m/z 821.7 decreased from 120 K to 70 K to 30 K for data acquisition settings of 240 K, 120 K, and 60 K, respectively (Figure S4).

Figure 4.

AP-MALDI HRMS mass spectrum of IVA envelope lipids using (THAP + Li) for zoomed m/z region of 800–850 (bottom). Inset spectra for zoomed in m/z 818–822 (middle) and expanded region centered on m/z 821.70 (top). Isotopic envelope for HexCer(d42:1) m/z values are in red, HexCer(d41:1-OH) in green, and SM(d42:1) in blue. m = monoisotopic peak, m+1 = second isotopic peak, m+2 = third isotopic peak, and m+3 = fourth isotopic peak.

For further evaluation, we compared the mass spectra generated from our newly developed AP-MALDI HRMS workflow to that of a more conventional vacuum (v)MALDI TOF MS workflow. The same MALDI matrix (THAP + Li) and IVA lipid extracts were analyzed on a Bruker ultra-fleXtreme MALDI TOF mass spectrometer (Figure 5). The vMALDI TOF mass spectrum displayed a high-quality spectrum that was complementary to the AP-MALDI HRMS spectrum. Several PC, PE, SM, and HexCer were identified at their sum composition. The mass resolution for identified peaks was typically greater than 20000. The inset spectrum in Figure 5 showed the isotopic envelope for HexCer(d42:1), similar to Figure 4. The difference being the increased mass resolution for the AP-MALDI HRMS spectra allowed for greater distinction between isobaric m/z values. This in turn not only allowed for greater confidence in identification but also increased the number of lipid identifications.

Figure 5.

vMALDI TOF mass spectrum of IVA envelope lipids using (THAP + Li). Inset spectra for zoomed in m/z 818–823 (middle) and expanded region centered on m/z 821.7 (top). Isotopic envelope for HexCer(d42:1) m/z values are in red, HexCer(d41:1-OH) in green, and SM(d42:1) in blue. m = monoisotopic peak, m+1 = second isotopic peak, m+2 = third isotopic peak, and m+3 = fourth isotopic peak.

AP-MALDI Tandem MS for Structure Characterization of Hexosylceramide.

Lipid identification based on accurate mass database searching allows for confident assignment of the sum composition of a lipid. The sum composition of a lipid yields structure information about the lipid headgroup and acyl chain composition (i.e., number of acyl chain carbons and degree of unsaturation). Further structure information can be achieved with the use of tandem MS. Tandem MS of lipids has a rich history and the choice of which adduct to fragment heavily influences gas-phase ion fragmentation.⁴⁰ A primary motivation for our initial interest in doping lithium in the MALDI matrix²³ was to take advantage of the information-rich fragmentation spectra resulting from the dissociation of lithiated precursor ions.^41,42 This proved to be especially apropos for fragmentation of HexCer.

A series of HexCer with and without hydroxylation were the most abundant precursor ions in the (THAP + Li) MS1 spectrum (Figure 1B). At the sum composition level for HexCer, there was structural ambiguity as to the sphingoid base, n-acyl chain (length and degree of saturation), location of additional hydroxyl group (if present), and identity of the sugar residue (glucose or galactose). For the most abundant HexCer we were able to generate tandem mass spectra via collision-induced dissociation (CID) of the lithiated precursor ion. Example tandem mass spectra of HexCer with and without hydroxylation are displayed in Figure 6. Fragmentation nomenclature was adopted from the literature.^43–45 Extensive fragmentation of the intact precursor ion corresponding to cleavages across the sphingoid base, n-acyl chain, and glycosidic bond were observed. These cleavages allowed us to confidently assign the identity of the sphingoid base as sphingosine (d18:1), the n-acyl chain as fully saturated C16–C26, and location of the hydroxyl group (when present) to the C2 position on the n-acyl chain. The identity of the sugar residue remained elusive. Our tandem MS experiments which included the use of authentic glucosylceramide (GlcCer) and galactosylceramide (GalCer) standards did not allow us to distinguish between glucose and galactose isomers, even with the use of lithium adducts (SI Figure S5). We are currently investigating the identity of the sugar residue using hydrophilic interaction chromatography (HILIC). HILIC separations have been shown to chromatographically separate sugar isomers including GlcCer and GalCer.⁴⁶ Although this is an ongoing project, our initial results indicated the HexCer structures are a mixture of both GlcCer and GalCer. Our lipid identifications which included the use of “Hex” indicated the sugar portion was a heterogeneous mixture of both glucose and galactose. A total of 11 HexCer structures were assigned via tandem mass spectrometry (Figure 6).

Figure 6.

AP-MALDI HRMS tandem mass spectra of HexCer lithiated precursor ions. (A) Tandem mass spectrum of precursor ion m/z 834.7 corresponding to HexCer(d18:1/24:0(2-OH)). (B) Tandem mass spectrum of precursor ion m/z 818.7 corresponding to HexCer(d18:1/24:0). Inset structures detail annotated fragmentation. Fragmentation nomenclature adopted from the literature.^48–50 Tables next to spectra show identified HexCer structures. *Low abundance for HexCer(d34:1-OH) precursor ion prevented tandem MS; identification left at sum composition.

Identification of Ceramide.

The identification of individual ceramides from the IVA virions proved challenging. The expanded m/z range for both the DHB and (THAP + Li) spectra showed a series of abundant m/z values between 500 and 700 that corresponded to ceramides as protonated minus water ([M + H − H₂O]⁺) adducts (Figure 7A,B, m/z values indicated with an asterisk). Preliminary identification was done using accurate mass database searching (Figure S6). The identification of these m/z values as ceramides was highly unusual because ceramides are predominately presented in MALDI MS as alkali metal adducts (e.g., Na, K, Li depending on the abundance of each metal in the sample/matrix mixture). The [M + H − H₂O]⁺ adduct if present is typically of low abundance in MALDI spectra. This is in contrast to ESI spectra where the [M + H − H₂O]⁺ adduct is commonly abundant.^47,48 Also of note was the high abundance of these same m/z values in both the DHB and (THAP + Li) spectra. As shown here and in previous reports, the addition of lithium (or any other alkali metal) to the MALDI matrix strongly forces adduct consolidation to the doped alkali metal.^23,49 The use of (THAP + Li) did in fact shift ceramide adduct formation to the lithiated form yet did so for only about 16% of the total adduct formation (Figure 7D,E). The over-whelming adduct for the identified ceramides was the [M + H – H₂O]⁺ at >80% (Figure 7D). These percentages were highly unusual for lithium doping experiments, suggesting the preliminarily identified [M + H − H₂O]⁺ ceramides may be incorrectly assigned. Corroborating evidence that [M + H − H₂O]⁺ adducts of ceramides were likely misassigned was obtained via tandem MS. Tandem MS of the m/z peaks of the preliminarily identified [M + H − H₂O]⁺ ceramides (those with asterisks in Figure 7A,B) did not fragment to give characteristic ceramide product ions (e.g., fragmentation across the sphingoid base and associated water losses) (data not shown). Congruently, tandem MS confirmed the ceramide identify of several lithiated m/z values (identified in Figure 7F). Lastly, comparison to the vMALDI-TOF expanded m/z range spectra for both DHB (data not shown) and (THAP + Li) in the IVA virions lipid extracts (Figure 7C) clearly indicated that ceramides are not present as [M + H − H₂O]⁺ adducts in the IVA virions lipid extracts. In fact, ceramides, consistent with all other lipids identified, uniformly ionized in MALDI to predominately form alkali metal adducts. The identification of contaminant and interfering m/z values in the AP-MALDI spectra highlights the complementary nature of using accurate mass, adduct formation and consolidation, and tandem MS to increase confidence in lipid identification. Note, the mass accuracy threshold used was tolerant for Orbitrap data. The rationale for keeping with the 10 ppm criteria was the confidence in lipid identification when using lithium adducts. For example, a LIPIDMAPS search with a 10 ppm threshold of the m/z values in Figure 7F using the [M + Li]⁺ adduct resulted in ceramides as the only identification except for m/z 544.5227 and 572.5540 where an NEA (N-acylethanolamine) is isomeric. In these cases, the ceramides were confirmed by tandem MS.

Figure 7.

AP-MALDI HRMS mass spectra (expanded m/z range) of IVA envelope lipids for identification of ceramides. (A) Mass spectrum using DHB. (B) Mass spectrum using (THAP + Li). (C) Mass spectrum using (THAP + Li) on vMALDI-TOF. (D) Percent adduction for ceramides with DHB and (THAP + Li). (E) Zoomed in m/z range of 540–660 of (B) (AP-MALDI HRMS spectrum with (THAP + Li)). (F) List of identified ceramides from AP-MALDI HRMS spectrum with (THAP + Li). Blue = [M + H]⁺, green = [M + Na]⁺, and red = [M+Li]⁺; PC: glycerophosphocholine, PE: glycerophosphoethanolamine, SM: sphingomyelin, HexCer: hexosylceramide, Cer: ceramide; * indicates interfering/contaminant m/z values (preliminarily identified as [M + H − H₂O]⁺ ceramides; ppm = mass accuracy in parts per million (ppm); identifications beyond sum compositions were done using tandem MS.

Lipid Identification from IVA Virions via AP-MALDI HRMS and Lithium Adduct Consolidation.

Refer to Table S1 for list of lipids that were identified from the IVA virions. Identification was based on accurate mass database searching of the lithiated adduct and listed as sum composition. Tandem MS was achieved for a number of sphingolipids, and consequently, these lipids had more in-depth structure characterization. There was a total of 132 lipids identified representing eight lipid subclasses. The majority of the identified lipids and the majority of lipid abundance were attributed to SP (Figure 8). The bias toward SP detection was a consequence of SP enrichment in viral envelopes^8,16 and lithium adduct consolidation during the MALDI process is more energetically favorable for SP.^23,50 Accordingly, AP-MALDI HRMS with (THAP + Li) preferentially biased SP detection.

Figure 8.

Number of lipids identified and percent abundance per lipid class and lipid subclass. Data were acquired using AP-MALDI HRMS with (THAP + Li) as MALDI matrix. Individual lipid abundance and identifications are located in SI Table S1. GL = gylcerolipid, GP = glycerophospholipid, SP = sphingolipid, Cer = ceramide, Hex = hexosyl ceramide (including HexCer–OH), SM = sphingomyelin, LPC = lysophosphocholine, PC = glycerophosphocholine (diacyl and ether), PE= glycerophosphoethanolamine (diacyl and ether), and TG = triacylglycerol.

Although the use of (THAP + Li) enhanced ionization of SP, it also enhanced GP ionization for lipids containing an amine group (e.g., PE and glycerophosphoserine (PS)) although to a lesser extent. PC are the exception with the quaternary amine. The enhanced ionization is concomitant with the decreased ionization efficiency of the highly abundant PC. The main advantage of (THAP + Li) was reduced ambiguity for detected lipids due to lithium adduct consolidation. The enhancement of PE and PS when using (THAP + Li) was not overly noticeable in the viral envelope lipids due to the high abundance of SM and HexCer. Given PS propensity toward negative ion mode and low relative abundance compared to other GP, PS was not observed in the AP-MALDI mass spectra. The overall number of GP that were detected was similar for both DHB and (THAP + Li) albeit at different relative abundances. The use of (THAP + Li) simplified identification by reducing multiadduction.

DISCUSSION

The viral envelope provides the virus with structural protection and functional bioactivity during viral entry into host cells and viral assembly.⁴ These are key features for the virus to achieve successful infection of host cells and transmission between cells. The viral envelope is composed of a coordinated network of glycoproteins and lipids.⁶ The lipid composition consists of lipids that have been preferentially incorporated from the host cell;^8,9 i.e., the lipid composition of the viral envelope is specifically enriched and deficient in certain lipids when compared to the overall composition of the host cell. The enrichment of certain lipids and consequent deficiency of others is considered to confer advantageous biophysical properties to the viral envelope, such as lateral packing, vertical asymmetry, and high gel–liquid transitions temperatures.^13,14 These biophysical properties promote membrane rigidity, membrane curvature, and protein–lipid interactions, all of which are necessary for virus propagation.^12,15,16 Subsequently, the ability to characterize the lipid composition of the viral envelope is of great importance for understanding how enveloped viruses navigate the extracellular environment and successfully infect host cells. To this end, we have developed an analytical strategy based on AP-MALDI HRMS that rapidly and confidently identifies lipids extracted from enveloped viruses.

The choice of AP-MALDI HRMS over other mass spectrometry-based platforms has several notable advantages not withstanding its complementary nature to other developed methods (e.g., shotgun ESI-MS/MS¹⁰ and LC-ESI-MS/MS¹¹). The choice of MALDI for lipid analysis is expedient for several noteworthy reasons: the MALDI process is fast, requires low sample amounts, tunable by matrix selection and addition of alkali metals, and when interfaced to HRMS provides in-depth analysis of complex samples. In comparison to a more traditional vMALDI source, the use of an AP-MALDI source that readily interfaces with a diverse set of mass spectrometers afforded us with the opportunity to select a mass spectrometer with our desired mass spectrometry attributes (high mass resolution (>100000) and tandem mass spectrometry). The high mass resolution was essential owing to the complex nature of the lipid extracts and lack of chromatographic separation prior to mass spectrometry analysis. Tandem MS was crucial for increased structure characterization especially in combination of lithiated precursor ions. With these two mass spectrometry attributes in mind, we chose to interface the AP-MALDI source to a Thermo QE HF. Of note, other mass spectrometers that have a dedicated MALDI source, specifically MALDI FT-ICR, are used for lipid analysis and achieve both high mass resolution and tandem mass spectrometry.⁵¹ Our choice of the Thermo QE HF relied on the following rationale: (i) instrument performance: high mass resolution and tandem MS, (ii) lab accessibility, (iii) lower operating costs and smaller footprint compared to FT-ICR (QE HF is considered a benchtop MS), (iv) ease of interchanging the AP-MALDI source with ESI, and (v) ease of interchanging the source to other mass spectrometers using same optimized AP-MALDI parameters. Another advantage of using an AP-MALDI source is the lack of dedicated vacuum hardware which reduces the cost and maintenance burden, and the luxury of not having to wait for the desired vacuum settings to be reached expedites sample analysis time.

The combination of lithium adduct consolidation, high mass resolution, and tandem MS all achieved on an AP-MALDI HRMS platform allowed us to efficiently identify 132 lipid structures extracted from IVA virions. Inherent in the MALDI process is the need to use a matrix to facilitate gas-phase ionization of the analytes of interest.²² The matrix can be selected to promote the ionization of particular analyte classes and thus the matrix/analyte pairing is a critical component in the MALDI process. The matrix can be further augmented with the addition of alkali metals to bias adduct formation during ionization.⁵² This can be highly beneficial for reducing multiadduct formation by forcing consolidation to a single, primary adduct. The use of lithium as the alkali metal of choice combined with the matrix, THAP, has been shown to achieve superior lithium adduct consolidation.^23,39 This resulted in reduced spectral complexity via near elimination of multiadduction. Consequently, the predominate lithium adducts yielded high specificity for lipid identification by effectively reducing isobaric m/z values resulting from multiadduction. This was exactly what was observed for identification of viral envelope lipids using (THAP + Li) as our MALDI matrix.

Although lithium adduct consolidation proved highly effective for reducing spectral complexity and providing an easily identified primary adduct, there was still the need for utilizing high mass resolution for teasing apart the isobaric overlap of isotopic envelopes from different lipid species. This type of isobaric overlap is a result of the complexity of the lipid extracts corresponding to hundreds (if not thousands) of possible lipid structures of which some will have similar elemental composition and overlapping isotopic patterns. The ability to acquire mass spectrometry data with high mass resolution can potentially resolve isobaric overlap. High mass resolution is commonly defined as a mass resolution of 10000 or greater.^53,54 This is routinely achieved by TOF, Orbitraps, and FT-ICR mass spectrometers with the latter two capable of realizing >100000 mass resolution under optimized data acquisition parameters. In the case of lipid analysis from total lipid extracts, lipid identification at the MS1 level is greatly increased when data is acquired at increasing levels of mass resolution.⁵⁵ Our data strongly supported the use of increasing mass resolution for increased lipid identifications. Data acquisition at an effective mass resolution of >100000 provided the requisite resolving power to resolve many overlapping isotopic envelopes and allowed us to confidently assign lipid identifications. The high mass resolution of the QE HF and lithium adduct consolidation allowed us to overcome the hindrances from isobaric complexity and confidently assign lipid identification at the sum composition level.

Further structure information was gained for several sphingolipid species with the use of tandem mass spectrometry of the lithiated precursor ion. Gas-phase fragmentation of lithiated precursor ions yields information rich product ion spectra corresponding to detailed structure information.^41,42 The use of lithiated precursor ions for tandem MS of sphingolipids provides structural insight into the sphingoid base, n-acyl chain, and headgroup.⁵⁶ An added benefit of lithium adduct consolidation was the presence of abundant lithiated precursor ions where tandem MS was possible. The gas-phase fragmentation of the lithiated precursor ions detailed extensive fragmentation across the sphingoid base, n-acyl chain, and headgroup. This provided sufficient evidence to assign extensive structure information for HexCer structures, including the localization of the hydroxyl modification. We were also able to achieve structural insight for a number of lithiated ceramide structures. Our tandem MS experiments were limited to abundant and/or well-resolved m/z values. Other lipids, including SM, which fragment favorably as lithium adducts, were not able to be appreciably isolated and fragmented due to the combined constraints of insufficient quadruple isolation, precursor ion abundance, and abundance of other closely associated m/z values. We are currently working to resolve these issues via the incorporation of ion mobility and multistage tandem mass spectrometry (MSⁿ). Ultimately, the added benefit of the tandem mass spectrometry experiments enabled us to gain insightful structural information for a number of sphingolipid structures that would not have been otherwise achieved.

Our developed AP-MALDI HRMS platform was applied to the analysis of lipids extracted from IVA virions. IVA is a zoonotic, negative sense RNA enveloped virus.⁵⁷ Although the primary host for IVA are aquatic birds, IVA can crossover and infect humans.⁵⁸ A number of deadly pandemics have been associated with IVA.⁵⁹ Given its history with human health and years of vaccine development, IVA is a model virus for not only investigating the lipid composition of the viral envelope but also ideal for developing new analytical methods. Indeed, the use of IVA virions allowed us to develop a new analytical platform for rapid and robust detection of viral envelope lipids. Our detection of 132 lipids should not be considered all-encompassing but rather an in-depth characterization of the most abundant envelope lipids with explicit enrichment for raft membrane-associated sphingolipids.

Our work has established AP-MALDI HRMS as a viable method to rapidly profile viral envelope lipids. This is highly relevant as enveloped viruses are a present and ongoing threat to human health. The ability to rapidly and confidently profile the lipid composition of the viral envelope has the potential to accelerate our understanding of key biological processes during virus infection, namely viral entry into the host cell, assembly, and extracellular survival. Also of noteworthy importance is the role the lipid composition potentially plays in the development of antivirals and vaccines. A number of current and emerging pandemic pathogens are zoonotic-enveloped viruses (including Ebola, Nipah, Lassa fever, MERS, and SARS).^60,61 Of equal concern is the lack of therapeutic options, vaccines, and/or antivirals available to reduce the significant health burden caused by enveloped viral disease outbreaks.^62,63 One potential strategy is the development of small molecules that target the viral envelope, rendering it inactive.^64–67 A crucial component of the viral envelope are lipids, and thus, the ability to rapidly characterize the lipid composition of the viral envelope when exposed to a membrane-targeting antiviral is an exciting avenue for advancing antiviral and vaccine development.

CONCLUSION

Our newly developed analytical strategy involving lithium adduct consolidation and AP-MALDI HRMS is distinctively suited for rapid and confident detection of lipids extracted from enveloped viruses. The combined use of lithium adduction, high mass resolution, and tandem mass spectrometry on an AP-MALDI HRMS platform allowed us to efficiently characterize the lipid composition of IVA virions. The lipid composition of the viral envelope is of structural and functional importance. Consequently, envelope lipids are a key component for virus transmission and a prospective target for virus inactivation. Taken together, our innovative approach to characterize the lipid composition of the viral envelope has high potential to substantially progress our knowledge of virus transmission. Moreover, the knowledge gained from the characterization of viral envelope lipids promises to provide a foundation for advancing antiviral and vaccine development.

ABBREVIATIONS:

AGC

automatic gain control

AP

atmospheric pressure

BHT

butylated hydroxyltoluene

Cer

ceramide

CHCl₃

chloroform

CV

Coefficient of variation

DHB

2,5-dihydroxybenzoic acid

ESI

electrospray ionization

FT-ICR

Fourier transform ion cyclotron resonance

fwhm

full width half height

GalCer

galactosyl ceramide

GL

gylcerolipids

GlcCer

glucosyl ceramide

GP

glycerophospholipids

GSL

glycosphingolipids

HexCer

hexosylceramide

HexCer–OH

hydroxylated hexosylceramide

HIV

human immunodeficiency virus

HPLC

high pressure liquid chromatography

HRMS

high-resolution mass spectrometry

HSV

herpes simplex virus

IV

influenza virus

IVA

influenza virus A

LC–MS

liquid chromatography–mass spectrometry

LiCl

lithium chloride

LPC

lysophosphocholine

m

monoisotopic peak

m+1

second isotopic peak

m+2

third isotopic peak

m+3

fourth isotopic peak

[M + H]⁺

protonated precursor ion

[M+H−H₂O]⁺

protonated minus water precursor ion

[M + Na]⁺

sodiated precursor ion

MALDI

matrix-assisted laser desorption/ionization

MeOH

methanol

MS

mass spectrometry

MSI

MS imaging

MTBE

tert-Butyl methyl ether

PC

glycerophosphocholine

PE

glycerophosphoethanolamine

PE-P

vinyl ether PE

ppm

parts per million

QE-HF

Q Exactive HF mass spectrometer

RSV

respiratory syncytial virus

SM

sphingomyelin

SP

sphingolipids

THAP

2′,4′,6′-trihydroxyacetophenone

TOF

time-of-flight

vMALDI

vacuum MALDI

VSV

vesicular stomatitis Indiana virus