Dual Metal Electrolysis in Theta Capillary for Lipid Analysis

Abstract

Increasing studies associating glycerophospholipids with various pathological conditions highlight the need for their thorough characterization. However, the intricate composition of the lipidome due to the presence of lipid isomers poses significant challenges to structural lipidomics. This study uses the anodic corrosion of two metals in a single theta nESI emitter as a tool to simultaneously characterize lipids at multiple isomer levels. Anodic corrosion of cobalt and copper in the positive ion mode generates the metal-adducted lipid complexes, [M+Co]²⁺ and [M+Cu]⁺, respectively. Optimization of parameters such as the distances of the electrodes from the nESI tip allowed the achievement of the formation of one metal-adducted lipid product at a time. Collision-induced dissociation (CID) of [M+Co]²⁺ results in preferential loss of the fatty acyl (FA) chain at the sn-2 position, thus generating singly charged sn-specific fragment ions. Whereas, multistage fragmentation of [M+Cu]⁺ via CID generated a C=C bond position-specific characteristic ion pattern induced by the π-Cu⁺ interaction. The feasibility of the method was tested on PC lipid extract from egg yolk to identify lipids on multiple isomer levels. Thus, the application of dual metal anodic corrosion allows lipid isomer identification with reduced sample preparation time, no signal suppression by counter anions, low sample consumption, and no need for an extra apparatus.

Graphical Abstract

1. Introduction

Lipids are important biomolecules in the body, exhibiting both structurally complex and functionally diverse.¹ They are an essential component of cellular membranes and play major roles in cell signaling and energy storage.^{2, 3} Variability in lipid chemical structure confers different properties on lipids.³ This variation arises from differences in headgroups, fatty acyl (FA) chain lengths, carbon-carbon double bond (C=C bond) positions in FA chains, the geometry of C=C bonds (cis or trans), and stereo numbering (sn)-positions of the FA chains on the glycerol backbone.^{3, 4} Dysregulation in lipid metabolism and lipid compositional alteration has been linked to various pathological conditions⁵ such as cancer,^6–10 cardiovascular diseases,¹¹ type-II diabetes,^{12, 13} and neurodegenerative diseases (such as Alzheimer’s disease).¹⁴ This has essentialized lipid structure characterization for a better understanding of the onset and progression of these diseases and the development of potential treatment.

Mass spectrometry (MS) coupled with tandem MS such as low-energy collision-induced dissociation (CID) has been used to elucidate lipid structures at the headgroup and FA chain length isomer level.^{15, 16} However, low energy of CID alone does not generate enough fragments to identify lipids at C=C bond positional and sn-positional isomer levels. For this purpose, various activation techniques such as metastable atom activated dissociation (MAD),¹⁷ charge transfer dissociation (CTD),¹⁸ radical-directed dissociation (RDD),¹⁹ ozone-induced dissociation (OzESI-MS),²⁰ electron-impact excitation of ions from organics (EIEIO),²¹ ultraviolet photodissociation (UVPD),²² and charge-remote fragmentation (CRF),²³ has been developed for lipid isomer characterization. Chemical derivitization methods such as the Paterno-Büchi (P-B) reaction,²⁴ epoxidation,^25–30, and ¹ΔO₂-ene reaction,^31–33 have been used to identify lipid C=C bond positional and photocycloaddition was reported to have the additional capability in geometric isomer identification.³⁴ Additionally, the combination of liquid chromatography, ion mobility spectrometry, and MS can further improve isomeric lipid analysis.³⁵ However, methods that can simultaneously characterize lipids at multiple isomer levels are still in need of development.^{7, 36, 37}

Metal-adducted lipid complexes have been used for lipid identification. Alkali metal-adducted lipid complexes such as [M+Na]⁺ and [M+K]⁺ are often formed in nano-electrospray ionization (nESI) with an inert electrode like Pt that generate information on the headgroup and fatty acyl chain length upon fragmentation.³⁸ Other metal-adducted lipid complexes such as silver, barium, manganese, and lithium have been used in isomeric lipid structure characterization.^{37, 39–45} However, these methods often involve premixing lipid solutions with inorganic metal salts to generate the metal-adducted lipid complexes. The use of metal salts has drawbacks including 1) it needs sample preparation, 2) the metal salts such as magnesium and barium are handled as toxic reagents,⁴¹ and 3) it leads to suppression of the signal intensity in the mass spectrum by the counter anion of the inorganic metal salts.

Electrochemistry has been increasingly used in recent days as a means of chemical activation. This method replaced traditional chemical oxidants and reductants and uses electric current as the “traceless” means of redox transformations.^{46, 47} nESI is a method of ionization where an electrode placed inside an emitter acts as an anode in positive ion mode and as a cathode in the negative ion mode.⁴⁸ This inherent electrochemical nature of nESI MS has been used in studies to monitor and study electrochemical reactions.⁴⁹ Further, the use of a theta nESI emitter (two barrels divided by a glass septum) has been used as a modification to generate a divided electrochemical cell.⁵⁰ In this work, a theta nESI is used to achieve the formation of two metal cations through anodic corrosion on the application of voltage to one metal wire at a time (Fig. 1). This metal anodic corrosion in turn produced metal-adducted lipid complexes that are transported to the mass spectrometer via nESI for further lipid structural analysis at multiple isomer levels via tandem MS.⁵¹ Thus this method incorporates ways to 1) analyze lipids at multiple isomer levels, 2) reduce sample preparation time, 3) eliminate use of traditional redox reagents, and 4) reduce signal suppression by counter anions.

Fig. 1.

A modified nESI emitter (orifice size: ~10 μm) for simultaneous lipid structure characterization at multiple isomer levels. Lipid standard loaded into one barrel of a theta capillary, while the other barrel is filled with ACN. A cobalt and copper electrode are inserted into the barrel with lipid standard and ACN, respectively. Voltage was applied to one electrode at a time which allows time separation for the formation of the metal-adducted lipid products. MS² of the cobalt-adducted lipid gives characteristic ions for the sn-positional isomer. MS⁴ of the copper-adducted lipid gives characteristic pattern for the C=C bond positional isomer

2. Experimental Section

2.1. Nomenclature

Lipid nomenclature in this study was used in reference to the guidelines mentioned in Liebisch et al.⁵² C=C bond positions are indicated by the Δ-nomenclature system to indicate the first carbon in the fatty acyl chain beginning at the carbonyl group that contains the C=C bond. Determined sn-positions of the fatty acyl (FA) chains are denoted with a “/”, listing the fatty acyl (FA) chain in the sn-1 position first followed by the one in the sn-2 position. When the sn-position is undetermined, the “/” symbol is replaced by “_”. For example, PC 18:1(Δ11)/16:0 refers to a PC lipid with monounsaturated FA 18:1 at the sn-1 position and saturated FA 16:0 at the sn-2 position. The C=C bond position in FA 18:1 is present between the eleventh and the twelfth carbon from the carbonyl group.

2.2. Chemicals and materials

All lipid standards including phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), phosphatidylglycerol (PG), phosphatic acid (PA) and L-α-phosphatidylcholine (95%) (Egg, Chicken) were purchased from Avanti Polar Lipids (Birmingham, AL, USA). Acetonitrile (ACN), chloroform (CHCl₃), and methanol (CH₃OH), cobalt (II) chloride (CoCl₂), and tetrakis(acetonitrile)copper(I) hexafluorophosphate (C₈H₁₂CuF₆N₄P) were purchased from Sigma-Aldrich (St Louis, MO, USA). Formic acid (HCOOH), 99.0+%, Optima ^™ LC/MS Grade and lecithin from soybean (TCI, America^™) were purchased from Fisher Scientific (Waltham, MA). All solvents have a purity greater than 99.9% and were used without further purification. Hydrochloric acid (HCl) was purchased from Sigma-Aldrich (St Louis, MO, USA). Stock solutions of lipid standards were prepared in CHCl₃. Stock solutions for lipid standards were then diluted to a concentration of 50μM with the solvent of ACN (spiked with 0.1% formic acid). PC lipid extract from egg yolk was dissolved in ACN (spiked with 0.1% formic acid) to achieve a concentration of 250μM.

Cobalt wire (0.25mm diameter, Puratronic^®, 99.995% (metals basis)), copper wire (0.25mm diameter, Puratronic^®, 99.9999% (metals basis)), nickel wire (0.25mm diameter, annealed, 99.98% (metals basis)), iron wire (0.25mm diameter, Puratronic^™, 99.995% (metals basis), and silver wire (0.1mm diameter, Premion^®, 99.997% (metals basis)) were purchased from Thermo Fischer Scientific (Ward Hill, MA).

2.3. Instrumentation and MS analysis

NanoESI emitters were made from 0.2mm septum theta borosilicate glass capillary (1.5/1.02mm) from WPI Inc. (Sarasota, FL) with a P-1000 micropipette puller from Sutter Instrument Company (Novato, CA). The following parameters were used to obtain an orifice size of ~ 10 μm: heat 535, pull 35, velocity 30, time 250, and pressure 500.

MS data were acquired on the LTQ XL^™ linear ion trap mass spectrometer from Thermo Fisher Scientific (San Jose, CA). MS data analysis was done using the Xcalibur Qual Browser program (Waltham, MA). The following parameters were used during data acquisition: samples were ionized in the positive ion mode with a positive DC spray voltage of 1.0 – 1.5 kV for the copper electrode and 3.0 – 3.5 kV for the cobalt electrode. In full scan MS, a m/z range of 200–1000 was used. A capillary temperature of 275°C, the S-lens RF level was set to 67.9%, the capillary voltage was 45V, and the tube lens voltage was 85V. A maximum injection time of 100 ms for 2 micro scans was used while acquiring tandem MS (MSⁿ; n =1, 2, 3,..) data. Collision-induced dissociation (CID) energy of 20 a.u. with an isolation window between 1.5 – 3.0 Th was used in tandem MS² of the cobalt-adducted lipid product to obtain the sn-positional isomer characteristic ion pattern. CID energy of 20 a.u. with an isolation window of 1.5 Th was used for MS² and MS³ of the copper-adducted lipid product, whereas a CID energy of 30 a.u. and an isolation window of 1.5 Th was used for MS⁴ of the copper-adducted lipid product to obtain the C=C bond positional characteristic ions.

2.4. In situ electrochemical corrosion of Co and Cu

In this method, theta nESI emitters (orifice of ~10 μm) were used to obtain the anodic corrosion of two metals (cobalt and copper) in a single experiment (Fig. 1). One barrel of the theta tip was loaded with a lipid sample spiked with 0.1% formic acid and the other barrel was loaded with ACN with 0.01%HCl. Cobalt wire was introduced into the barrel containing the lipid sample and the copper wire was introduced into the barrel containing ACN solvent. Voltage was then applied to one electrode at a time. Optimal voltages of 3.0 – 3.5 kV and 1.0 – 1.5 kV were applied to the cobalt and copper electrodes respectively to generate one metal-adducted lipid product at a time. Fragmentation of the cobalt and copper-adducted lipid products via tandem MS generated characteristic ion patterns denoting the sn-position of the FAs on the glycerol backbone of the lipid and the C=C bond position in the FA chain, respectively.

3. Result and Discussion

3.1. Electrolytic transition metal-lipid interactions for isomeric lipid analysis

The formation of metal-lipid complexes by adding metal salts (e.g., alkali salts) into the lipid solution enhances lipid structure characterization either by improving ionization efficiencies⁵³ or by triggering specific fragmentation patterns upon CID to yield distinctive diagnostic fragmentation ions that aid in lipid identification.^39–44 To circumvent the introduction of unwanted anions and their adverse effects on the analysis, we in situ generate metal cations through electrolysis of nESI electrode wires in this study. This approach effectively forms metal-lipid adducts without the need for externally added salts. We focus on transition metals to investigate their role in assisting lipid isomer identification and to determine the distinct lipid isomers that can be distinguished with specific electrolytic transition metals.

In this work, transition metals including iron (Fe), cobalt (Co), nickel (Ni), copper (Cu), and silver (Ag) were systematically studied for anodic corrosion to form metal-adducted lipid complexes [M+X]ⁿ⁺ followed by tandem MS analysis of the metal-adducted lipid product to generate characteristic ions for lipid isomer identification (Table 1, Fig. S2). PC 16:0/18:1(Δ9) was used as a model lipid for studying electrolytic transition metal-lipid interactions.

Table 1.

Metal electrodes in nESI and their applications in lipid isomer identification. Characteristic ions for isomer identification upon CID of metal-adducted lipids are shown with a lipid standard PC16:0/18:1(Δ9)

Metal	Optimal Voltage (kV)	Lipid isomer that can be analyzed	Metal-adducted lipid species	Metal-adducted lipid product m/z	Characteristic Ions of PC 16:0/18:1(Δ9)
					Major m/z	Minor m/z
Fe	3.5 – 4.0	sn- position	[M+Fe]2+	407.76	337.18	311.17
					478.33	504.34
Co	3.0 – 3.5	sn-position	[M+Co]2+	409.25	340.18	314.17
					478.33	504.34
Ni	2.0 – 2.5	sn-position	[M+Ni]2+	408.76	339.18	313.17
					478.33	504.34
Cu	1.0 – 1.5	C=C position	[M+Cu]+	822.51	201.07
Ag	3.0 – 4.0	C=C position	[M+Ag]+	866.48	245.05

Upon the application of voltage to the nESI electrode wire in the lipid solution, the metal-adducted lipid complex ions [M+X]ⁿ⁺ were observed at m/z 407.76 for [M+Fe]²⁺, m/z 409.25 for [M+Co]²⁺, m/z 408.76 for [M+Ni]²⁺, m/z 822.51 for [M+Cu]⁺, and m/z 866.48 for [M+Ag]⁺ respectively. Due to varied redox potentials of these metals, the optimal voltages required to initiate the electrolysis of metal electrodes differed significantly, ranging from 1kV to 4kV (Table 1). Notably, Cu exhibited the highest ease of electrolysis, while Fe proved to be the most challenging electrolysis.

Two major fragmentation pathways upon CID are found to occur by fragmenting the metal-adducted lipid complexes [M+X] ⁿ⁺, which can be used to assist the lipid characterization of either sn-positions or C=C bond positions. Fig. 2 illustrates the two possible fragmentation pathways and corresponding isomeric structure determination.

Fig. 2.

Types of fragmentation pathway of metal-adducted lipid structures on application of CID. Type I fragmentation pathway forms characteristic ions specific to sn-positional isomers. Type II fragmentation pathway forms characteristic ions specific to C=C bond positional isomers. X= metals that form doubly charged metal-adducted lipid products and follow Type I fragmentation pathway (Fe, Co, and Ni); Y = metals that form singly charged metal-adducted lipid products and follow Type II fragmentation pathway (Cu, and Ag)

Among the transition metals investigated, Fe, Co, and Ni displayed the first type of fragmentation pathway, and Cu and Ag underwent the second type of fragmentation pathway. For example, upon application of an optimal voltage of 3.5 – 4.0 kV, doubly charged Fe-adducted lipid product [PC+Fe]²⁺ was formed at m/z 407.76 (Fig. S2.1). CID of [PC +Fe]²⁺ resulted in the singly charged five and six membered cyclic phosphoesters (dioxaphospholane and dioxaphosphorinane) which corresponded to the loss of FA at sn-2 (FA2) and sn-1 (FA1), respectively. Formation of the cyclic phosphosester was attributed to the oxygen of the phosphate group in the [PC+Fe]²⁺ complex attacking carbon atoms in the glycerol backbone at either the sn-1 or sn-2 positions, followed by removal of protons by the ester carbonyl. The α-hydrogen of the FA at the sn-2 position exhibits greater lability compared to the FA at the sn-1 position, making it more prone to cleavage.^{29, 39, 41, 54, 55} As a result, [PC+H-FA2]⁺ and [FA2-H+Fe]⁺ showed more dominant peaks at m/z 478.33 and 337.18 than [PC+H-FA1]⁺ and [FA1-H+Fe]⁺ at m/z 504.34 and 311.17 indicating FA 18:1(Δ9) at the sn-2 position and FA 16:0 at the sn-1 position (Fig. S2.1).

MS² of the doubly charged nickel-adducted lipid product [PC+Ni]²⁺ m/z 408.76 generated major characteristic ions of [FA2-H+Ni] ⁺ at m/z 339.18 and [PC+H-FA2] ⁺ at m/z 478.33, indicating FA 18:1(Δ9) at the sn-2 position and FA 16:0 at the sn-1 position (Fig. S2.2).

On the contrary, MS² of the singly charged Ag-adducted lipid and Cu-adducted lipid exhibited the second type of fragmentation. According to the hard and soft acid and base (HSAB) principle, Cu⁺/Ag⁺ has a soft acidic property that preferably interacts with the π-electrons of C=C bonds in FA chains because of their soft basic property. Therefore, Cu⁺/Ag⁺ preferentially coordinates to form singly charged ions with unsaturated FA chains of the lipid, and follows charge remote fragmentation to reveal the location of the C=C bond position in the FA chain when applied with collision energy.^{29, 40, 44, 56} For example, upon application of an optimal voltage of 3.0 – 4.0 kV, a singly charged Ag-adducted lipid product [PC+Ag] ⁺ is formed (Fig. S2.3).

Initial fragmentation of the [PC+Ag]⁺ resulted in the choline headgroup loss generating a major peak of [PC-183+Ag] ⁺. Further fragmentation of the [PC-183+Ag] ⁺ ion generated the dominant peak of [R₁COOH+Ag] ⁺ and peaks originating from neutral losses such as [R₁COOH-CO₂-H₂+Ag]⁺ and [R₁COOH-H₂O+Ag] ⁺. CID of [R₁COOH+Ag]⁺ resulted in γ-hydrogen rearrangement and allylic mechanisms⁵⁷ cleaving at the carbon beside the double bond of the FA chain to form the most intense ion of the resultant MS⁴ spectra.^56–58 MS² of the singly charged silver-adducted lipid product [PC 16:0/18:1(Δ9)+Ag] ⁺ at m/z 866.48 generated peaks of [PC 16:0/18:1(Δ9)-183-CO₂+Ag]⁺ m/z 683.42, [FA18:1(Δ9)+Ag] ⁺ m/z 389.16, [FA18:1(Δ9)-CO_2-H₂+Ag] ⁺ m/z 343.15, and [FA18:1(Δ9)-H₂O+Ag] ⁺ m/z 371.15. The most intense ion peak for the MS⁴ of [FA18:1(Δ9)-CO_2-H₂+Ag] ⁺ m/z 343.15 was seen at m/z 245.05 due to the cleavage of the bond between C10-C11 to indicate C=C double bond at Δ9 position (Fig. S2.3).

To move forward, we chose Co for sn-positional isomer identification and Cu for C=C bond positional isomer characterization. The Fe electrode required an optimal voltage of 3.5 – 4.0 kV to generate the iron-adducted lipid product. The high voltage resulted in a higher flow rate and higher rate of sample consumption. Ni wire required an optimal voltage of 2.5 – 3.0 kV but generated the sn-position characteristic ions only at a very specific isolation window of 2.5 – 2.6 Da in MS². Hence, Co was chosen for sn-isomer identification in the method, that required a moderate sample consumption and generated sn-position characteristic ions irrespective of the isolation window in MS².

Ag wire required an optimal voltage 3.0 – 4.0 kV to generate the Ag-adducted lipid product. Characterization of C=C bond position with Ag had drawbacks such as the high voltage leading to high sample consumption rate, and MS⁴ of the [FA18:1(Δ9)-CO₂-H₂+Ag] ⁺ producing very low ion intensity spectra of ~ 1.7E0. Whereas, Cu wire required a low optimal voltage of 1.0 – 1.5 kV that led to low sample consumption rate and generated a MS⁴ spectra with ion intensity of 5.0E1 or more. Therefore, Cu wire was used in the method for C=C bond position characterization in the lipid sample.

3.2. Achieving dual transition metal electrolysis in a theta capillary for multi-level isomeric lipid analysis

With the selected transition metal nESI electrodes, we are able to distinguish lipid sn-positions and C=C bond positions. To achieve the lipid analysis at multiple isomer levels in a single experiment, we initiated the electrochemical corrosion of both Co and Cu in two separate barrels of a theta nESI emitter, allowing for characterization of lipids at the sn-positional and C=C bond positional isomer levels, respectively. Voltage was applied to one electrode at a time creating a potential difference between the two barrels. This potential difference generated an electrochemical cell and served four purposes: 1) it produced a stable electrospray, 2) it caused anodic corrosion of the electrodes 3) it allowed electro-osmosis⁵¹, and 4) allowed time separation between the formation of the Co and Cu-adducted lipid product. Moreover, the distances between the electrodes and nESI emitter tips vary for Co and Cu, enabling the formation of a single type of metal-lipid complex at a time.

A control experiment was performed by inserting both electrodes in a single barrel nESI tip. However, it resulted in a Cu-adducted lipid product signal dominating the mass spectrum over the Co-adducted lipid product. Separation of the two metal-adducted lipid products was achieved by 1) using a theta emitter; 2) loading the barrels with Co and Cu electrodes with lipid solution and ACN (lipid:ACN = 1:2 (v:v)) 3) spiking ACN with 0.01% HCl; and 4) placing the copper electrode further away from the orifice than the Co electrode.

It is worth noting that using two electrodes in separate barrels of the theta nESI emitter alone resulted in a similar mass spectrum as observed in the case of a single barrel nESI emitter with the inability to separate the metal-adducted lipid products or achieve lipid isomer identification. Application of 3.0 – 3.5 kV to the Co electrode generated a [M+Co]²⁺ that was transferred into the MS inlet. However, this voltage also resulted in the transfer of the lipid solution in the barrel with the Co electrode into the barrel with the Cu electrode via electroosmosis⁵⁹ (Fig. S4). The transfer diluted the lipid sample in the Cu barrel preventing [M+Cu]⁺ from suppressing [M+Co]²⁺ signal.

To further increase the Co and Cu-adducted product separation, ACN in the barrel with copper electrode was spiked with 0.01% HCl. Protons from HCl suppress the anodic corrosion of Cu electrode and the generation of [M+Cu]⁺. This allowed [M+Co]²⁺ to be the dominant peak when voltage is initially applied to Co electrode.

The distance of the electrodes from the orifice impacted the time separation between the two-metal adducted products. With the distance of the Co electrode fixed at 2.5 mm from the orifice, distances of 4.5 mm, 6.5mm, and 10mm of the Cu electrode from the orifice gave a time separation of about 0.5 min, 2 min, and 9 min between the generation of Co and Cu-adducted lipid product in the mass spectra (Fig. S5). A distance of 10 mm and 2.5 mm of Cu electrode and Co electrode from the orifice, respectively, gave the best separation between the abundance of the ion intensities of the two metal products and were hence used in experiment (Fig. 3). Switching the voltage application to the Cu electrode resulted in the formation of Cu-adducted lipid product at an optimal voltage of 1.0 – 1.5 kV, which gradually dominated the mass spectrum over the Co-adducted lipid product.

Fig. 3.

Selected extracted chromatogram and full mass spectra showing Co and Cu-adducted product separation when Co electrode is placed at 2.5mm and Cu electrode is placed at 10mm from the orifice of nESI (a) Extracted ion chromatogram of PC[16:0/18:1(Δ9)+Co]²⁺m/z 409.3, (b) extracted ion chromatogram of PC[16:0/18:1(Δ9)+Cu]⁺ m/z 822.5, (c) full mass spectrum from time t1 where PC[16:0/18:1(Δ9)+Co]²⁺ (m/z 409.3) dominates over PC[16:0/18:1(Δ9)+Cu]⁺ (m/z 822.5) for 9 minutes (d) full mass spectrum from time t2 where PC[16:0/18:1(Δ9)+Cu]⁺ (m/z 822.5 dominates over PC[16:0/18:1(Δ9)+Co]²⁺ (m/z 409.3) after 9 minutes. 0 to 8 minutes was used to perform tandem MS of the metal-adducted lipids to confirm the presence of the characteristic ions (not shown).

3.3. Characterization of PC sn-positional isomer via tandem MS of Co-adducted lipid complexes using dual transition metal electrolysis in a theta capillary

After successfully identifying the specific transition metal electrolysis for lipid isomer identification and accomplishing dual transition metal electrolysis in a theta capillary, we were eager to apply this method for lipid analysis at multiple isomer levels in a single experiment.

One barrel of the theta capillary was loaded with 50μM lipid solution (spiked with 0.1%FA) and the other barrel was loaded with ACN solvent (spiked with 0.01%HCl). Co and Cu wires were inserted into the barrels containing lipid solution and ACN solvent respectively. Optimal voltage of 3.0 – 3.5 kV was first applied to Co nESI electrode to form [M+Co]²⁺, which was then fragmented via CID to characterize lipids at the sn-positional isomer level (at MS²). Next, an optimal voltage of 1.0 – 1.5 kV was applied to Cu nESI electrode to form [M+Cu]⁺, which was then fragmented via CID to characterize lipids at the C=C bond positional isomer level (at MS⁴).

CID of [M+Co]²⁺, generated a pair of characteristic ions for sn-postion isomer identification. This was demonstrated by PC 18:1 (Δ11)/16:0 and PC 16:0/18:1 (Δ9) (Fig. 4). CID of [PC18:1(Δ11)/16:0+Co]²⁺ at m/z 409.25 generated major characteristic ions of [FA 16:0-H+Co] ⁺ at m/z 314.17 and [PC+H-FA 16:0]⁺ at m/z 504.34, indicating FA 16:0 at the sn-2 position and FA 18:1(Δ9) at the sn-1 position. Whereas CID of [PC16:0/18:1(Δ9)+Co]²⁺ at m/z 409.25 generated major characteristic ions of [FA18:1(Δ9)-H+Co] ⁺ at m/z 340.18 and [PC+H-FA 18:1(Δ9)] ⁺ at m/z 478.33, indicating FA 18:1(Δ9) at the sn-2 position and FA 16:0 at the sn-1 position.

Fig. 4.

Identification of lipid sn-position isomer from CID of Co-adducted lipid. (a) Full mass spectrum of [PC 18:1(Δ11)/16:0 + Co]²⁺ (b) Full mass spectrum of [PC 16:0/18:1(Δ9)+ Co]²⁺ (c) MS² of [PC 18:1(Δ11)/16:0 + Co]²⁺ at m/z 409.3 (d) MS² of [PC 16:0/18:1(Δ9) + Co]²⁺ at m/z 409.3 (e) characteristic ions of sn-position identification and their structures for [PC 18:1(Δ11)/16:0+Co²⁺] (f) characteristic ions of sn-position identification and their structures for [PC 16:0/18:1(Δ9)+ Co]²⁺

The method when applied on lecithin from soybean could also identify PC lipids with multiple unsaturations in the FA chain showing potential of the applicability of the method on a more diverse FA chain for sn-positional isomer identification (Fig. S6)

3.4. Characterization of PC C=C bond isomer via tandem MS of Cu-adducted lipid complexes using dual transition metal electrolysis in a theta capillary

Upon application of an optimal voltage between 1.0 – 1.5 kV, [PC+Cu] ⁺ is formed. MS⁴ of this [PC+Cu]⁺ revealed the position of the unsaturation in the FA chain of the lipid. This has been demonstrated with PC 18:1 (Δ11)/16:0, PC 16:0/18:1(Δ9), and PC 16:0/18:1(Δ8) (Fig. 5). [PC 18:1 (Δ11)/16:0+Cu] ⁺ is seen at m/z 822.51 in the full mass spectrum (Fig. 5a), CID of which results in headgroup loss resulting in [PC 18:1 (Δ11)/16:0–183+Cu] ⁺ m/z 639.44 (Fig. 5b). Mass selection and fragmentation of the [PC 18:1 (Δ11)/16:0–183+Cu] ⁺ ions in MS² resulted in the peaks of [FA18:1(Δ11)+Cu]⁺ m/z 345.18, [FA18:1(Δ11)-CO₂-H₂+Cu]⁺ m/z 299.18, [FA18:1(Δ11)-H₂O+Cu] ⁺ m/z 327.17, the [PC 18:1(Δ11)/16:0–183-CO₂+Cu]⁺ m/z 595.45, [PC 18:1(Δ11)/16:0–183-CO₂-H₂O+Cu]⁺ m/z 577.44 in the MS³ spectrum (Fig. 5c). Further CID of [FA18:1(Δ11)-CO₂-H₂+Cu]⁺ m/z 299.18 in MS⁴ generated a series of ions as we reported before⁴² with equal losses of mass difference of 14.02 Da relating to the -CH₂ loss. The C-C bond between C12-C13 was more favorably cleaved to generate the characteristic most intense ion peak at m/z 229.18 for C=C double bond at Δ11 position (Fig. 5d). [PC 16:0/18:1 (Δ9)+Cu]⁺ (Figure S3.2), and [PC 16:0/18:1(Δ8)+Cu] ⁺ generated same peaks as [PC 18:1 (Δ11)/16:0+Cu] ⁺ for full MS, MS², and MS³. However, on applying CID to [FA18:1(Δ9)-CO₂-H₂+Cu]⁺ m/z 299.18 and [FA18:1(Δ8)-CO₂-H₂+Cu]⁺ m/z 299.18 generated the most intense peak at m/z 201.07 and m/z 187.05 that are characteristic peaks of C=C double bond at (Δ9) and (Δ8), respectively (Fig 5e, and 5f).

Fig. 5.

Identification of lipid C=C bond isomer from CID of Cu-adducted PC lipid. (a) Full mass spectrum of [PC 18:1(Δ11)/16:0 + Cu]⁺ (b) MS² of [PC 18:1(Δ11)/16:0 + Cu]⁺ at m/z 822.5 (c) MS³ of [PC 18:1(Δ11)/16:0 + Cu]⁺ at m/z 639.4 (d) MS⁴ of [PC 18:1(Δ11)/16:0 + Cu]⁺ at m/z 299.2 (e) MS⁴ of [PC 16:0/18:1(Δ9) + Cu]⁺ at m/z 299.2 (f) MS⁴ of [PC 16:0/18:1(Δ8) + Cu]⁺ at m/z 299.2 (g) characteristic ions of C=C bond position identification and their structures

However, the method has some limitations. In case of FA chains with multiple unsaturations, the C=C bond position could not be correctly identified from the characteristic ion patterns of [M+Cu]⁺ in MS⁴.

Relative quantification of sn-positional isomers could be achieved based on the ion intensities of the sn-position characteristic ions on MS² of [M+Co]²⁺. Mixture of two lipid standards, PC 18:1(Δ9)/16:0 and PC 16:0/18:1(Δ9), were used to obtain a linear relationship (R² = 0.9994) between the ion intensities of (I₃₄₀ + I₄₇₈)/(I₃₄₀ + I₄₇₈ + I₃₁₄ + I₅₀₄) over percentage of PC 16:0/18:1(Δ9) in the lipid standard mixture.

Relative quantification of C=C bond positional isomers could also be achieved based on the ion intensities of the characteristic ions on MS⁴ of [M+Cu]⁺. Mixture of two lipid standards, PC 16:0/18:1(Δ8) PC 16:0/18:1(Δ9), were used to obtain a linear relationship (R² = 0.9966) between the ion intensity I₂₀₁/(I₂₀₁+I₁₈₇) over percentage of PC PC 16:0/18:1(Δ9) in the lipid standard mixture.

To investigate the fragmentation pathways induced by lipid adduction with electrolytic Co and Cu, we applied the dual metal electrolysis method to a range of lipid species, including PC, PE, PS, PA, and PG lipids. For anionic glycerophospholipids, the formation of metal-lipid complexes leads to a polarity inversion of the lipids which can be readily detected in the positive mode. The method when tested on PE, PS, PG, and PA lipids could reveal C=C bond position on the FA chain in the lipid. For example, MS⁴ of [PE 16:0/18:1(Δ9)+Cu]⁺ generated C=C bond positional characteristic ions.

All Cu-adducted lipid species follow fragmentation pathway II when applied with CID (Fig. S7). However, MS² of the Co-adducted PE, PS, PG, and PA lipids might not be able to reveal sn-positional characteristic ions. It was observed that [PE+Co]²⁺ lipid follows fragmentation pathway I when applied with CID, however, the relative intensities of the characteristic peaks are not sn-position specific. For lipids PS, PG, and PA, the Co-adducted peaks did not follow fragmentation pathway I as shown by PC lipids.

When tested for the limit of detection (LOD) with the lipid standard PC 16:0/18:1 (Δ9), clear characteristic ions for sn-positional and C=C bond positional isomer was found at a concentration of 10μM (Fig. S8).

The dynamic range for isomer detection obtained from the relative quantification calibration plot was determined to be 50μM (the lowest concentration of a particular lipid isomer at which distinctive characteristic peak for both sn-position and C=C bond positional isomers could be obtained) to 100μM.

3.5. Application of dual transition metal electrolysis to the analysis of egg yolk PC lipids

To demonstrate the feasibility of the dual transition metal electrolysis method, we applied it to analyze PC lipid extract from egg yolk. The protonated and metal-adducted lipid products were observed in full mass spectrum within a range of m/z 200–1000. 7 sn-positional isomers and 11 C=C bond positional isomers were identified with the characterization from the Co and Cu-adducted lipid products respectively (Table 2).

Table 2.

Identification of sn-positional and C=C bond positional isomers of L-α-phosphatidylcholine (95%) (Egg, Chicken) via dual metal electrolysis in theta capillary

Lipid	[M+H]+	sn-positional isomer	Characteristic ions of sn-positional isomer		C=C bond positional isomer	Characteristic ions of C=C bond positional isomer
			Major m/z	Minor m/z
PC 32:1	732.55	PC 16:0/16:1	478.33	476.31	PC 16:0/16:1(Δ7)	173.04
			312.15	314.17
					PC 16:0/16:1(Δ9)	201.07
		PC 14:0/18:1	450.30	504.34	PC 14:0/18:1(Δ9)	201.07
			340.18	286.13
PC 34:2	758.57	PC 16:1/18:1	476.31	504.34	PC 16:1(Δ9)/18:1(Δ9)	201.07
			340.18	312.15
		PC 16:0/18:2	478.33	502.33
			338.17	314.17
PC 34:1	760.59	PC 16:0/18:1	478.33	504.34	PC 16:0/18:1(Δ9)	201.07
			340.18	314.17
					PC 16:0/18:1(Δ11)	229.10
PC 36:2	786.60	PC 18:1/18:1	504.34		PC 18:1(Δ9)/18:1(Δ9)	201.07
			340.18
					PC 18:1(Δ11)/18:1(Δ11)	229.10
					PC 18:1(Δ12)/18:1(Δ12)	243.12
PC 36:1	788.62	PC 18:0/18:1	506.36	504.34	PC 18:0/18:1(Δ9)	201.07
			340.18	342.20
					PC 18:0/18:1(Δ11)	229.10

As an example, [PC 36:1+H]⁺ was observed at m/z 788.62. The Co-adducted product [PC 36:1 + Co]²⁺ was observed at m/z 423.27, which upon fragmentation via CID simultaneously revealed the FA chain lengths through the characteristic ion masses and the FA chains sn-position through the characteristic ion intensities in MS². CID of [PC 36:1 + Co]²⁺ generated major characteristic ions of [FA18:1-H+Co] ⁺ at m/z 340.18 and [PC+H-FA 18:1] ⁺ at m/z 506.36, indicating FA 18:1 at the sn-2 position and FA 18:0 at the sn-1 position (Fig. S9).

The Cu-adducted product [PC 18:0/18:1+Cu]⁺ was seen at m/z 850.54 in the full mass spectrum, CID of which led to headgroup loss resulting in [PC 18:0/18:1–183+Cu]⁺ m/z 667.47. Mass selection and fragmentation of the [PC 18:0/18:1–183+Cu]⁺ ions in MS² resulted in the peaks of [FA18:1+Cu] ⁺ m/z 345.18, [FA18:1-CO₂-H₂+Cu]⁺ m/z 299.18, [FA18:1-H₂O+Cu] ⁺ m/z 327.17, the [PC 18:0/18:1–183-CO₂+Cu]⁺ m/z 623.46, [PC 18:0/18:1–183-CO₂-H₂O+Cu]⁺ m/z 605.45 in the MS³ spectrum. Further CID of [FA18:1-CO₂-H₂+Cu] ⁺ m/z 299.18 in MS⁴ generated the characteristic most intense ion peak at m/z 201.07 for C=C at Δ9 thus characterizing the lipid as PC 18:0/18:1(Δ9). Identification of the positional isomers of C=C bonds within PC lipids is limited to lipids with a monounsaturated fatty acid (FA) chain in the lipid extract from egg yolk.

We performed relative quantification of the identified lipid isomers within the PC lipid extract from egg yolk (Fig. 7). PC 16:0/18:2 and PC 18:0/18:1 were found to be the most abundant among the sn-positional isomers (Fig. 7a), whereas the C=C bond positional isomers were found to have been majorly constituted by C=C bond at Δ9 position in the FA chain (Fig. 7b).

Fig. 7.

Relative quantities of a) sn-positional and b) C=C bond positional isomer in L-α-phosphatidylcholine (95%) (Egg, Chicken). Each PC lipid isomer was normalized via the formula (I_{characterisitc ion}/I_{TIC in MS}ⁿ) prior to relative quantification

4. Conclusion

In this work, the electrochemical nature of nESI was used to achieve electrolysis of transition metal electrodes and form metal-adducted lipid products. The fragmentation of these products follow two types of pathways that aid in either sn-positional or C=C bond positional isomer in lipids. Fe, Co, and Ni were found to aid in lipid characterization at the sn-positional isomer level, whereas Cu and Ag aid in characterization of lipids at the C=C bond positional isomer level. Eventually, a modified nESI was developed to successfully achieve dual transition metal (Co and Cu) electrolysis in a theta capillary to aid simultaneous characterization of lipids at multiple isomer levels. The preferential loss of the FA at sn-2 position on CID of the cobalt-adducted lipid product allowed for sn-positional isomer identification. The π-Cu⁺ interaction enabled the formation of characteristic patterns in MS⁴ on application of CID to the copper-adducted lipid product, thus allowing C=C bond positional isomer identification. With this method, lipid standards and PC lipid extract from egg yolk were successfully characterized at multiple isomer levels under single experimental conditions without the need of extra apparatus.

Fig. 6.

Calibration curve for relative quantification of a) lipid sn-positional isomers PC 16:/18:1(Δ9) and PC 18:1(Δ9)/16:0; and b) lipid C=C bond positional isomers PC 18:1(Δ8)/16:0 and PC 18:1(Δ9)/16:0