Determination of phospholipid regioisomers in complex biological samples using silver cation ionization in ESI-MS3

1

Determination of phospholipid regioisomers in

complex biological samples using silver cation

ionization in ESI-MS

3

Björn Berglund

May 22

2019

Degree Project C in Chemistry, 1KB010

Analytical Chemistry – Department of Chemistry BMC – Uppsala University Supervisor: Ingela Lanekoff, Uppsala University

3

Acknowledgment

I would like to thank my supervisor Ingela Lanekoff and all the members of the research group for the supportive and friendly atmosphere. A special thanks to doctoral student Johan Lilja, who during the course of the project always to took time to discuss matters with an educational sense to push me in the right direction.

4

Abstract

Phospholipid’s structural composition in cells plays an important role in the human body. Depending on the cells biological role the structural composition and ratios of phospholipids vary. An alteration of phospholipids composition within the cell membrane has been shown to relate to common diseases. Analytical methods to completely determine phospholipids structure are necessary in order to fully understand the underlying cause of diseases related to phospholipids and even how to cure them. The aim of this project is to investigate phospholipids complexed with silver in respect to the sn-1/sn-2 position of acyl chains on the glycerol backbone. Pure phospholipids samples and biological samples were analyzed with direct infusion in ESI-MS3 as the chosen method. A graphical model was established and was successfully used to estimate the ratio of regioisomers for phospholipid species in MS3 of a biological sample of rat brain homogenate.

5

Table of content

1 Abbreviations ... 6

2 Aim ... 7

3 Introduction and theory ... 8

3.1 Phospholipids ... 8

3.2 Mass spectrometry... 9

3.2.1 Electrospray ionization ... 10

3.2.2 Linear ion trap ... 12

3.2.3 Mass resolving power ... 12

3.2.4 Collision-induced dissociation ... 13

3.2.5 Isotopic pattern in mass spectrometry... 13

3.2.6 Silver adducts and analysis of phospholipids ... 14

3.2.7 Charge-remote fragmentation ... 15

4 Experimental ... 17

4.1 Chemicals ... 17

4.2 Standard stock and sample preparation ... 17

4.2.1 Standard stock solution ... 17

4.2.2 Standard samples ... 17

4.2.3 Biological sample: rat brain tissue ... 17

4.3 Instrumentation ... 18

5 Results and discussion ... 19

5.1 Phosphatidylcholine in MS2 and MS3... 19

5.2 Phosphatidylserine in MS2 and MS3 ... 22

5.3 Phosphatidylcholine in MS3: Various settings of collision energy ... 24

5.4 Phosphatidylcholine in MS3 50:50 fragment ratio ... 27

5.5 Phosphatidylcholine fragmentation plot ... 28

5.6 Phosphatidylcholine in a biological sample ... 29

5.7 Phosphatidylcholine in a biological sample - complexity ... 32

6 Conclusion and future perspective ... 37

7 References ... 38

8 Appendix... 40

8.1 Sample preparation ... 40

8.2 Instrumental settings ... 40

8.3 MS2 of pure samples of phosphatidylcholine 18:1_18:0 regioisomers ... 41

8.4 MS2 of complex samples of phosphatidylcholine... 42

8.5 MS1 of complex samples ... 44

6

1 Abbreviations

MS Mass Spectrometry

MS/MS Tandem mass spectrometry

MSn Mass spectrometry with unspecified fragmentation steps MS1 Mass spectrometry with no fragmentation step

MS2 Mass spectrometry with one fragmentation step MS3 Mass spectrometry with two fragmentation steps

m/z Mass to Charge ratio

FT-ICR Fourier Transform Ion Cyclotron Resonance

QTOF Quadrupole Time-of-Flight

ESI Electrospray Ionization

LIT Linear Ion Trap

FWHM Full Width at Half Maximum

CE Collision Energy

SID Surface-Induced Dissociation

ETD Electron-Transfer Dissociation

CRF Charge Remote Fragmentation

PE PhosphatidylEthanolamines

PG PhosphatidylGlycerol

PS PhosphatidylSerine

PC PhosphatidylCholine

Ag107 Silver atom with mass 107 Da Ag109 Silver atom with mass 109 Da

7

2 Aim

To investigate if MSn of phospholipids cationized with silver provides fragmentation information related to the sn-1/sn-2 position of the fatty acyl chains.

8

3 Introduction and theory

3.1 Phospholipids

The goal is to study phospholipids in a biological complex mixture direct injected in a mass spectrometric instrument. Phospholipids have the characteristics of amphiphilic molecules. The long nonpolar fatty acid tails are hydrophobic while the polar or charged head groups consisting of an organic compound combined with phosphate are hydrophilic. This property favors the formation of bilayer micelles which enclose the constituents of the inner cell, shielding it from the outside environment. Apart from being a structural building block of the outer cell membranes phospholipids takes part in the inner cell housing and subdividing central processes into individual partitions called organelles and plays a role in cell signaling[1]. Depending on the cells biological role the compositions and ratios of phospholipids vary and changes in composition between bilayers can have biological effects. Alternations of the ratio of both phosphatidylethanolamine (PE) and phosphatidylcholine (PC) in the human brain, plasma, and blood cells have been linked to common diseases[2][3]. Since the phospholipid composition plays such an important role in the human body, both time effective and full structural determination methods of phospholipids are necessary in order to fully understand how phospholipids are related to some disease. Fully structural composition of phospholipids includes the sn-1 and sn-2 position (Figure 1) of the acyl chains on the glycerol backbone.

The characteristic structure of phospholipids comprises of two fatty acids attached by esterification on sn-1 and sn-2 position of a glycerol backbone. Figure sn-1 shows a sn-16 carbon acyl chain at the sn-sn-1 position without any C-C double bonds (16:0) and referred to as saturated. At the sn-2 position, an 18 carbon acyl chain with one C-C double bond and is referred to as unsaturated (18:1). Summed together they can be denoted as (34:1), and then the length and unsaturation are unspecified for each of the acyl chains. The acyl non-polar chains can vary in the number of carbon and of unsaturated C-C double bonds. Sn-3 position on the glycerol backbone is occupied by one phosphate group where further attachment by phosphodiester linkage of a different polar and chargeable organic compound. Due to the different headgroups, X (Figure 1) phospholipids can be divided into different classes. For example, ethanolamine or choline is a possible attachment to the phosphate and the class is then referred to as phosphatidylethanolamine (PE) and phosphatidylcholine (PC) respectively. So if class, acyl chain length, saturation/unsaturation and position on the glycerol backbone are to be accounted for, it will be denoted as PC 16:0/18:1. If the position of acyl chains on the glycerol backbone is unknown it’s denoted as PC 16:0_18:1. The linkage may vary for the non-polar acyl chain at the sn-1 position. Therefore, the phospholipids can be divided in to further subclasses. Plasmanyl, the carbon chain is linked to the sn-1 position on the glycerol by an alkyl ether and plasmenyl by vinyl ether linkage (Figure 1). Plasmenyl is also called plasmalogen. The classes mentioned are just a few of many classes and subclasses of phospholipids. In this project, the focus lies on identifying phosphatidylcholine in complex samples.

9

Figure 1: Phospholipid general structure, different classes, and subclasses

3.2 Mass spectrometry

Mass spectrometry (MS) is an analytical method and is based on the principle of generating gas-phase ions, then separating them by the ion mass to charge ratio (m/z) under vacuum and detecting them at a detector. All mass spectrometry instruments follow these three general steps and will be mentioned and categorized as an ion source, mass analyzer, and detector respectively. Ions can be generated as both negative and positive ions at the ion source and the mass analyzer separate the ions by a constant magnetic field, time of flight or electrically by attraction and repulsion[4]. After the separation, the ions end up at the detector. The detector detects and measures the number of ions and is then graphically presented in a mass spectrum. The mass spectrum is a two-dimensional representation via computer software where the x-axis represents the m/z and the y-axis the relative abundance or intensity of the measured ions. The information generated via a mass spectrometric analysis can be used to identify compounds by their mass. Depending on the analytical goal there are a vast number of techniques to choose from in relation to the ion source and mass analyzer [4]. This has given rise to a vast variety of mass spectrometers. For this project, the ionization method used is electrospray ionization (ESI) and the mass analyzer a linear ion trap (LIT). More detailed information on these is presented in the following section.

10 3.2.1 Electrospray ionization

Electrospray ionization (ESI) is what is called a “soft” ionization method. By soft ionization, the analyte of interest is ionized from a solution to the gas phase with little or no fragmentation. Fragmentation can occur as the molecules are ionized at the ion source depending on what type of ionization method used outcome can be high or little to no fragmentation. ESI is very useful on analytes consisting of small inorganic ions to organic species that are large, non-volatile and chargeable i.e. transformed in solution with respect to acid or base characteristics or cation/anion adducts. ESI-MS is a very sensitive method and a very low concentration of the analyte is needed in the solution, concentrations ranging from 10-7 – 10-3 mol/L are adequate [5].

For ESI, the solution is channeled through a fine capillary with a high potential applied with its counter electrode attached to the small opening leading into the mass analyzer. The placement of the capillary tip can vary from 1-3 cm from the MS opening [5]. The potential determines the formation of either positive or negative ions. The following explanation is based on the formation of positive ions: If the electrospray capillary is held at a positive voltage, an electrical field is created and the charged species in the solution will move under its influence (Figure 2). The negatively charged species will move towards the capillary walls away from the surface enriching the outgoing surface of the droplet with positive ions. The charge separation or polarization of the droplets species changes the surface tension and deforms it to a cone shape pointing towards the negative potential applied from the counter electrode. If the applied potential is high enough the surface tension at the cone tip becomes unstable and ejects small droplets of the solution with an abundance of positive ions (Figure 2). This cone formation of the outgoing droplet was first described by G.I Taylor 1964 and is nowadays referred to as the Taylor cone [4].

Figure 2: Ions in solution to gas phase ions via ESI. Reprinted with permission from [9]. Copyright ©, 2009 Wiley Periodicals, Inc.

The aerosol droplets move towards the negative counter electrode at the MS by the nature of its overly positively charged species. The droplets are unstable and forced to split into even smaller droplets by the electrostatic forces at the surface created by Coulomb repulsion [5]. This phenomenon also called Coulomb fission occurs as evaporation of the solvent decreases the volume of the droplet, repulsion at the

11

surface yet again overcomes the surface tension since the charge remain constant, ejecting smaller portions into new even smaller droplets. The further shrinkage of droplets will eventually lead to gas phase ions.

The actual formations of gas phase ions from the small aerosol droplets are still debated, but two explanations are purposed. Charged residue model (CRM) was the first to be proposed by Dole et al. in 1968 [5]. This model institutes that the final droplet throughout the coulomb fission stage contains only one charged analyte molecule and complete evaporation of the solvent (Figure 3).

The latter model called ion evaporation model (IEM) were proposed by Iribarne and Thomson in 1976 [5]. When the radius of the droplets shrinks to less than 10nm the Columbus fission stops and emission of gaseous ions starts emerging (Figure 3). The result from Iribarne and Thomson and further investigation by others has strengthened the hypothesis of IEM for small inorganic analytes but can’t be applied for larger organic analytes[5]. Therefore IEM can theoretically explain ionization of smaller molecules while CRM still remains much more feasible for large organic analytes.

Figure 3: Ionization model of IEM and CRM in ESI. Reprinted with permission from [6]. Rights are managed by Taylor & Francis

When using ESI-MS interference such as ion suppression effects can occur, ultimately it negatively affects the efficiency of producing gas-phase ions produced and also signal response in the mass analyzer. Ion suppression can occur in the ionization step due to the sample matrix. Both solvent composition and co-existing compounds can affect the ionization of analyte by changes in spray droplet formation or droplet evaporation. Negative changes promoting ion suppression in the spray is bigger droplets, slower evaporation or a more stable surface tension reducing the gas-phase ions reaching the mass analyzer[7]. For phospholipids different class, higher concentration, increasing acyl length and a lower amount of unsaturation have been shown to lower the instrumental response using ESI and an Ion trap mass analyzer[8]. The ion suppression factors in both ionization and instrument response make quantitative measurements a challenge since phospholipids of equal molar concentration will give a different signal. To acquire equate data the use of internal standard for each analyte is necessary[8]. This project is focused on qualitative measurements and to reduce ion suppression solvent, phospholipid concentration in samples is considered to acquire high ionization efficiency in order to get a good signal. Instrumental settings are also optimized for the ions of interest over a specific m/z range.

12 3.2.2 Linear ion trap

The MS instrument used for this work is a linear ion trap (LIT) and it consists of a quadrupole with four hyperbolic rods (Figure 4). A quadrupole is a widely used instrument where ions are guided through in an electrically induced field by the applied potential on the four rods[4]. Two rods on the opposite side of one another are paired and in-between the pairs they have fixed opposite potential plus an added alternating potential with a specific frequency i.e. causing an attraction and repulsion of the ion between the pair of rods. For a given set of potentials over the rods it will give ions with a certain m/z a stable path across the quadrupole.

The rods in the LIT consist of three segments where the two outer segments are 12 mm and the middle section are 37 mm in length[4](Figure 4). In two of the opposite middle section rods, a 30mm slit is present for the radial ejection of ions on to a detector. The trapping of the ions is made possible by applying a higher potential on the outer shorter segments of the LIT storing them in the longer middle storage segment. To introduce the ions in the storing area of the LIT the entrance segment is held at a low potential and the back segment is held at a high potential reflecting ions back in the trapping segment. The magnitude of trapped ions is limited to the fastest ions that are now moving in the opposite direction towards the inlet. The potential of the entrance segment is then matched with the opposite segment with a high potential trapping the ions in the storage section. To eject the ions out the radial exit slit the potential of the exit slits are altered to affect the trajectory of the ion sending them out and on to the detector. Since the trajectories of the ions depend on their mass, ions of with different m/z are ejected separately.

Figure 4: Configuration of a two-dimensional linear ion trap. Reprinted with permission from [9]. Copyright© 2002, American Society for Mass Spectrometry

3.2.3 Mass resolving power

Mass resolving power is an instrument´s ability to differentiate nearby peaks of equal intensity as individual peaks in a mass spectrum. Mass resolution R, it's defined as the least difference in m/z that can be separated at a given m/z signal in a mass spectrum[4]:

13

𝑅 =

𝑚/𝑧

∆𝑚/𝑧

Example, the minimum resolution needed to separate CO+ at 27.995 m/z from ion N2+ at 28.006 m/z are calculated by dividing m/z of N2+ by the difference in m/z for CO+. This yields R= 28.006/0.011 ≈ 2500. If resolution R=2500 is achieved the resolving power is good enough to separate the two peaks. The resolution of an instrument can also be calculated by dividing peak m/z signal by the width of the peak

∆𝑚/𝑧

Low resolution is referred to were R=500-2000 is obtained and High resolution when R>5000 [4]. Quadrupole instruments such as the LIT are a low-resolution instrument where modern FT-ICR instruments can achieve resolution as high as R=105-106. For the pure sample of phospholipids used in this project, the LIT resolution will suffice in order to resolve productions. However, with a complex matrix such as a biological sample, the phospholipid species are vast and product ions might not be resolved properly from each other due to small mass differences.

3.2.4 Collision-induced dissociation

Fragmentation of a specific ion can provide partial or absolute information about the structural composition of the ion by identifying the molecular mass and their molecular composition of the fragmented ions constituents. Fragments of an ion are called product ion while the originating ion is called precursor ion. Collision-induced dissociation (CID) is a fragmentation technique where stable ions are accelerated by an increase of the potential field on to a neutral gas present in the mass analyzer[10]. A parent ion is first selected with a specific m/z. Before acceleration, all other ions are ejected from the LIT. The acceleration of the ion is caused by increasing the potential over the outer segments of the LIT. The increased speed and motion through the LIT the ions have gained kinetic energy. Colliding with the neutral gas the kinetic energy is transformed into internal energy and with enough internal energy gain the ion dissociate into daughter ions. [4]

There are many different fragmentations methods and depending on the analytical goal and on what type of mass analyzer used they may vary. In Ion trap instruments other than CID, Electron-transfer dissociation (ETD) provides fragmentation by a reaction of precursor ion with radical anions as a source of electrons. The electron transfer causes a reaction and bond cleavage dissociation of the precursor ion into product ions[11]. Surface-induced dissociation (SID) compared to CID does not induce collision with neutral background gas but rather with a collision with specially treated surfaces. This is not used in a LIT instrument but rather in QTOF/FT-ICR mass analyzers[11].

3.2.5 Isotopic pattern in mass spectrometry

Atoms are categories in the periodic table by the number of protons present in the nucleus. In the nucleus, the protons are accompanied by neutrons. If the numbers of neutrons differ from the number of protons they are termed isotopes and give rise to a different mass of that element. In the periodic table, the relative atomic mass is presented, which is the average weight of the elements naturally occurring isotopes[4]. Isotopes and their natural abundance vary from element to element and in mass spectrometry, the existent of isotopes for each element must be considered. As mentioned, mass spectrometry separate ions of different m/z. Thus, the existents of isotopes the m/z of the same compound may vary and will result in an isotopic pattern which can be a useful tool to identify the elemental composition of the compound. To

14

identify a compound in a mass spectrum calculating the nominal mass is preferred. Nominal mass is referred to as the calculated mass where the mass of the most abundant isotope is accounted for. In a mass spectrum, the nominal mass will correlate to the most abundant peak present with its lower abundant isotopic peaks close by. Silver, for example, has two naturally occurring isotopes, Ag107, and Ag109 with the isotopic mass of 106.905094 and 108.904756 respectively. Ag107 is the most natural abundant isotope and if normalized to 100% the isotope Ag109 correspond to 92.9% to its lower mass isotope [4]. The isotopic pattern in figure 5 represents the visual appearance that would be expected in a mass spectrum.

Figure 5: Representation of the expected isotopic pattern of Ag in MS.

3.2.6 Silver adducts and analysis of phospholipids

Silver can be used to specifically ionize molecules with C-C double bonds. Silver ions are prone to form complexes with alkenes with no changes to the characteristics of the double bond. The empty 5s-orbital of the silver ion forms a σ-bond with the filed π-bond of the C-C double bond and a π-bond is formed with the empty anti-π-orbitals of the carbons with the filled d-orbitals of the silver [12]. Similar bonding properties are shown with elements of Cu and Au, where Ag is the weaker complex [13].

There are several techniques for detailed analysis of phospholipids, but the most common are liquid chromatography combined with mass spectrometry (LC-MS). The complex forming property of silver is also used to separate phospholipids in LC-chromatography by their order of unsaturation by using silver as stationary phase[14]. LC is an effective separation method but requires time in order to get fully separated phospholipids. Direct infusion-MS called “shotgun lipidomics” is a newly grown method of lipid analysis that was made possible by the development of ESI[15]. In the “shotgun” approach separation step is excluded and rely only on the separation of ions m/z. Without a separation step, a directly injected sample can be more time efficient and experimental steps are removed from the method at whole. The separation is achieved within the ESI-MS/MS method and has been proven efficient when determining phospholipid content in crude samples[16]. The development of mass spectrometric methods to differentiate the sn-1/sn-2 position of the acyl chains has been done by looking at an abundance ratio of product ion in MS/MS analyses with the use of metal adducts[17][18]. Combined with LC separation sn-1/sn-2 position of the acyl chains have successfully been determined by ionization of phospholipids complexed with lithium [19]. In this project, the shotgun approach will be applied on standards and a rat brain homogenate

15

complexed with silver in order to differentiate different classes of phospholipids and determine the sn-1/sn-2 position of the acyl chains on the glycerol backbone.

For phospholipids formed in ESI are typically adducts with a proton or metal cation and denoted as [M + H] +, [M + Ag] + and [M + Na] + etc. where M is the intact molecule. Other notations containing Rx refers to the unsaturated acyl chain and if R1 and R2 are used it refers to the unsaturated acyl chain with its known position on the glycerol backbone. R1 refers to sn-1 glycerol position and R2 refers to sn-2.

Adducts of divalent, alkali and alkaline earth metals have been used to study the fragmentation of different classes of phospholipids with ESI-CID-MS/MS in the search of fragmentation pattern which can give qualitative information of their structural formation[18][17]. The result of Ho et al. show that the ratio of produced fragmentation ions [M-R1OOH + (Co+2- H+)]+ and [M-R2OOH + (Co+2- H+)]+ for PE, PG and PS can be used to determine the sn-1/sn-2 position on the glycerol backbone. However, the ratio was dependent on the different classes of phospholipids and of the acyl chain composition [17]. Cobalt did not form cation adducts with PC, so Hus et al were able to determine sn-1/sn-2 of PC with lithium as an adduct. They found a correlating ratio of product ions [M -R1OOH-(C3H9N) +Li

+

]+ and [M –R2OOH-(C3H9N) +Li

+

]+ from parent ion [M +Li+]+, where [M –R1OOH-(C3H9N) +Li +

]+ is the more abundant product ion[18]. So to achieve an accurate structural determination of sn-1/sn-2 position on the glycerol backbone parameters such as phospholipid class, the structural character of the acyl chains and metal cation adduct need to be considered since this will affect the abundance and ratio of product ions. Yoo et al. showed that silver complexed with phospholipids of classes PE, PS and PC are not reliant on the phospholipid class nor do different metals have to be used in order to get suitable ions [20]. Three characteristic product ions originating from silver complexed with phospholipids were distinguished to be [RxOOH +Ag], [RxOOH –H2O +Ag] and [RxOOH – CO2 – H2 +Ag], where Rx is the unsaturated acyl chain. The ratio of [RxOOH +Ag] and [RxOOH –H2O +Ag] are similar for all tested phospholipids and [RxOOH –H2O +Ag] is favored when the unsaturated acyl chain is at the sn-2 position. However, the ratio changes with an abundance of [RxOOH +Ag] ions when the unsaturated acyl chain is at the sn-1 position. Different level of Collision energy (CE) used did not appear to change the fragmentation patterns and with higher CE only ion fragmentation efficiency increased but the relative abundance of product ions did not change[20].

3.2.7 Charge-remote fragmentation

Charge remote fragmentation (CRF) is the phenomena where gas phase ions dissociate into fragments where the cleaving of a bond is located and seemingly not affected by the charge site of the parent ion[21]. The fragmentation mechanism is independent and not affected by the actual charge site. Phospholipids, when ionized, can have multiple charge sites depending on the phospholipid class and metal adduct used, but usually, they have an overall charge of plus one. As suggested by Hsu et al. the product ions used to determine the sn-1/sn-2 position of the acyl chains of PC is first the loss of trimethylamine followed by either loss of one acyl chain[18]. The loss of trimethylamine is seemingly induced by the negative charge of the phosphate (Figure 6, first mechanism stage). The loss of either acyl chain is not located near the charge site and the mechanism suggested indicates CRF (Figure 6, third mechanist stage). PC complexed with lithium when dissociated with CID undergoes both CRF but also fragmentation induced by the charge sites. Phospholipids complexed with silver in comparison have its charge site located at the unsaturated double bond. Yoo et al. studied phospholipids complexed with silver and fragmentation showed an abundance of product ions [RxOOH +Ag], [RxOOH –H2O +Ag] and [RxOOH – CO2 – H2

16

+Ag][20]. These product ions consist of the unsaturated acyl chain with a difference in a small neutral loss. Losses of both the polar headgroup and second acyl chain indicate that fragmentation of phospholipids complexed with silver undergo CRF. The charge site of phospholipids complexed with lithium is physically positioned at the headgroup compared with silver where its positioned at the C-C double bond and this as shown will give rise to very different fragmentation patterns.

Figure 6: Suggested mechanism of lithiated adducts of PC. Reprinted with permission from [18]. Copyright © 1998, American Society for Mass Spectrometry

17

4 Experimental

4.1 Chemicals

Silver Nitrate (AgNO3) was purchased from Merck. Water used was of MilliQ™ grade and made in the

lab. Liquid-chromatography grade methanol and formic acid (concentration 98-100%) used for solvent buffer were purchased from Merck and Sigma Aldrich respectively. The pure phospholipids of phosphatidylserine (PS 16:0/18:1) and phosphatidylcholine (PC 18:0/18:1 & 18:1/18:0) were purchased from Avanti Polar Lipids Inc. (PS 16:0/18:1) and PC (18:1/18:0) in chloroform and PC 18:0/18:1 in powder form.

4.2 Standard stock and sample preparation

Silver standard solution of 20 mL was made by weighing Silver Nitrate (AgNO3) 0.06270g and 20 mL

MilliQ™ measured and added. Actual silver weight was equal to 0.0398g and the calculated concentrations consist of 1990 ppm. Solvent standard of 20 mL was made by measuring 20 mL of liquid-chromatography grade Methanol (Merck) and by adding 20 µL formic acid (Sigma Aldrich) corresponding to 0.1%.

The pure lipids used were already prepared in a standard stock solution. The pure lipid PC 18:0/18:1 had been weighted and diluted with methanol (Merck) to a concentration of 100 µg/mL. (PS 16:0/18:1) and PC (18:1/18:0) solved in chloroform had been further diluted with methanol (Merck) to 151 µg/mL and 100 µg/mL respectively.

4.2.2 Standard samples

Separate standard sample of a total volume 1 mL was made from silver standard solution, solvent standard and from lipid standard stock solutions of PC 18:0/18:1, PC 18:1/18:0 and PS 16:0/18. Dilution was made in order to achieve 10ppm of silver, 10 µg/mL PC and 50 µg/mL PS. Calculated volumes for each sample (Appendix, Table 1) were pipetted into 2 mL sample flasks and shaken for 10 seconds on a vortex shaker. A 50:50 sample mix of PC 18:1/18:0 and 18:0/18:1 was also prepared by pipetting 400 µL of each prepared standard sample into a sample flask. Dilution corresponds to 5 µg/mL of each isomer and a total concentration of 10µg/mL phosphatidylcholine and 10ppm silver.

4.2.3 Biological sample: rat brain tissue

Animal experiments were approved by the Uppsala animal ethics committee and were performed in

accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Rat

brain extract was made by taking a whole rat brain, add 50 mL of methanol (Merck) and by hand triturated. The sample was sonicated for 40 minutes before centrifuged for 5 min at 3000 rpm. The supernatant was collected and stored in -20 °C. A total volume of 1 mL of rat brain sample was made from silver standard solution, solvent standard and from rat brain extract. Dilution was made in order to achieve 10ppm of silver, 1:500 dilution of extract. Calculated volumes for the sample (Appendix, Table 2) were pipetted into 2 mL sample flask and shaken for 10 seconds on a vortex shaker.

18

4.3 Instrumentation

The instrumentation used was Finnigan™ LTQ™ MS detector from Thermo Fisher Scientific Inc. All experiments are done in positive ion mode with a flow rate of 10 µl/min. General settings (Appendix, Table 3) were used for all experiments. Furthermore, specific settings (Appendix, Table 4) in addition to general settings vary for different experiments and are listed separately.

19

5 Results and discussion

5.1 Phosphatidylcholine in MS

and MS

The two regioisomers PC 18:1_18:0 were used for the first experiment to determine fragmentation patterns and to see similarities/differences with the main goal to identify the acyl chains positions on the glycerol backbone in both MS2 and MS3.

Separate Standard samples each containing pure PC 18:0/18:1 and PC 18:0/18:1 were injected by a 500µl syringe mounted to a syringe pump at the mass spectrometer. Specific settings 1 (Appendix, Table 4) were used for these experiments. Chemical structures and calculated nominal mass were done in Chemical Draw. Ion [M +H] + of PC 18:1_18:0 has a nominal mass of 788.129. When forming a complex with silver to the unsaturated C-C double bond the ion [M + Ag] + nominal mass are 894.4. Ions between m/z 894.4 and 896.4 were isolated for both standard samples and CID of 27 applied in order to induce fragmentation of product ions in MS2. Figure 7 and Figure 8 shows the MS2 mass spectrum of the precursor ion 894.4- 896.4 m/z for regioisomers PC 18:0/18:1 and PC 18:1/18:0 respectively. For both PC isomers and their corresponding silver isotope, the most abundant production ions are at m/z 835.25, 837.25, 711.42 and 713.42. Both isomers show the same fragmentation pattern and in the MS2 mass spectrum, no product ions indicate an identification of the acyl chains sn-1/sn-2 position on the glycerol backbone. The calculated neutral loss correspond to [C3H9N] = 59.07 Da and [PO4NC5H10] =183.07 Da respectively. Which correspond to the loss of parts of or the whole polar head group (Figure 9).

20

Figure 8: MS2 mass spectrum of PC 18:1/18:0, precursor ion m/z 894.4. ±7 Da ions capture width.

Figure 9: Fragmentation scheme of PC 18:0/18:1 complexed with silver. Dotted lines are drawn to indicate bond breakage as PC fragments together with the new product ion m/z (found in Figure 7 and 8) and the corresponding neutral loss.

Next, ions with m/z 711.4-713.4 were isolated in MS3 for both samples of PC 18:1/18:0 and PC 18:0/18:1 and CID of 35 applied in order to induce product ions to further investigate fragmentations patterns. Figure 10 and 11 shows MS3 mass spectrum of PC 18:0/18:1 and PC 18:1/18:0 respectively with the selected precursor ion with m/z 711.42-713.42. The induced fragmentation revealed new product ions, first product ion of m/z 603 which does not have corresponding silver isotope, the loss of silver was confirmed as the mass loss corresponds to [AgH]. The small fraction of m/z 445.17 and 447.17 match a neutral loss of the saturated acyl chain (Figure 12). Most abundant ions together with their corresponding silver isotope are m/z 389.17, 371.17 and 343.25. These ions correspond to a major loss of the saturated acyl chain together with the glycerol backbone. The largest of the three fragments 389.17 consists of a carboxylic acid complexed with silver and will be denoted as [R18OOH +Ag] +. Product ion of m/z 371.17 relative to [R18OOH +Ag] + match the loss of one water molecule and will be denoted as [R18OOH –H2O +Ag] +. Third product ion relative to [R18OOH –H2O +Ag] + proceeds to lose carbon monoxide and will be denoted as [R18OOH –H2O -CO +Ag] +. The different isomers did not show any specific product ion unique to the position of the unsaturated acyl chain on the glycerol backbone in ether MS2 and MS3. However, the fragmentation pattern in MS3 of PC 18:0/18:1 shows a higher abundance of the [R18OOH – H2O +Ag] + compared to [R18OOH +Ag] + (Figure 10). The opposite is noticed for PC 18:1/18:0 where a higher abundance of [R18OOH +Ag] + in respect to [R18OOH –H2O +Ag] + (Figure 11). This indicates that an abundance of [R18OOH –H2O +Ag] + are more favored when the unsaturated acyl chain 18:1 is positioned at the sn-2 position on the glycerol backbone. Opposite, the abundance of [R18OOH +Ag] + are more favored when the unsaturated acyl chain 18:1 is positioned at the sn-1 position on the glycerol backbone.

21

Figure 10: MS3 mass spectrum of PC 18:0/18:1, precursor ion m/z 711.5, ±7 Da ion capture width.

Figure 11: MS3 mass spectrum of PC 18:1/18:0, precursor ion m/z 711.5. ±7 Da ion capture width.

Figure 12: Fragmentation scheme of ion m/z 711.14 of PC 18:0/18:1 complexed with silver. Dotted lines are drawn to indicate bond breakage as PC fragments together with the new product ion m/z (found in Figure 10 and 11).

22

In order to test if product ions of 389, 371 and 343 could be detected in MS2 standard sample of both PC isomers were injected and specific settings 3 (Appendix, Table 4) were used. The three product ions for PC 18:1_18:0 could be distinguished in MS2 as 0-1% of the relative abundance of the overlooked spectrum 200-1000 m/z when the precursor ion was depleted with 90-95% (Appendix, Figure 1 and 2). For the case of PC 18:1/18:0 product ion of [R18OOH –H2O -CO +Ag] + is not even detectable (Figure 1). The respective isomer follows the same trend as seen where 389and 371 are either lower or higher as seen in MS3 (Figure 10 and 11). So, in MS2 the intensity ration of the two product ions do follow the same trend as in MS3. But fragmentation efficiency was poor and ion signal unstable. In the time the spectrum was recorded these product ions are not continuously showing. As an average of the intensity is recorded and as the product ion fluctuates in abundance this probably affects the reproducibility and ratio of the two product ions will differentiate. The poor ionization efficiency of the smaller product ions could be due to inefficient fragmentation of the more stable product ions 835 m/z and 711 m/z. More useful information can thus be obtained from an MS3 mass spectrum.

5.2 Phosphatidylserine in MS

and MS

PS is a different class of phospholipid and so the purpose is to investigate any similarities/differences to fragmentation patterns of PC. Standard sample of PS 16:0/18:1 was injected by a 500µl syringe mounted to a syringe pump at the mass spectrometer. Specific settings 2 (Appendix, Table 4) were used for these experiments. Ion [M +H] + of PS 16:0/18:1 has a nominal mass of 761.52. Silver when forming a complex to the unsaturated C-C double bound the ion [M + Ag] + nominal mass is 868.43. Ions of m/z 868.43-870.43 were isolated for PS 16:0/18:1 and CID of 30 applied in order to induce fragmentation of product ions in MS2. Figure 13 shows the MS2 mass spectrum of PS 16:0/18:1 with precursor ion with m/z 868.43 – 870.43.

The most abundant production ions in the MS2 mass spectrum in Figure 13 are 781.25, 783.25, 683.42 and 685.42. Calculated neutral loss corresponds to [C3H5NO2] and [C3H8NO6P] respectively. These neutral losses in MS2 correspond to the loss of parts of or the whole polar head group (Figure 14). Compared with PC species the initial Fragmentation patterns MS2 are very similar, two product ions are abundant and correspond to the loss of parts of or the whole polar head group. For PS and PC ions complexed with silver, the MS2 product ions can be used to determine the phospholipid class by identifying the neutral loss of the headgroup.

23

Figure 14: Fragmentation scheme of PS 16:0/18:1 complexed with silver. Dotted lines are drawn to indicate bond breakage as PS fragments together with the new product ion m/z (found in Figure 13) and the corresponding neutral loss.

Further experiments in MS3 were conducted with precursor ion of m/z 683.42 and 685.42 with CID 40 applied. The purpose is to conclude if the three characteristic fragments of [R18OOH +Ag] +, [R18OOH – H2O +Ag] + and [R18OOH –H2O -CO +Ag] + are as consistent as for PC and if similar ratio patterns of product ions are visible. Precursor ion of m/z 683.42 as shown in Figure 14 contains the glycerol backbone and the two fatty acyl chains positioned at the sn-1/sn-2 position. Figure 15 shows the MS3 mass spectrum of PS 16:0/18:1 with the selected precursor ion with m/z 683.42-685.42. In a similar manner to PC, the most abundant of product ions are m/z 575.42. No silver isotope is present and neutral loss is calculated to [AgH]. Product ion of m/z 445.17 and 447.17 correspond to the loss of the saturated 16:0 acyl chain positioned at the sn-1 position (Figure 16). Last 389.17, 371.17 and 343.25 m/z with their corresponding silver isotope are present. In the case of PS 16:0/18:1 the product ion of 371.17 m/z were more abundant in relation to 389.17 m/z. The origin of these fragments is the unsaturated acyl chain positioned at the sn-2 position of the glycerol backbone as in the case of PC 18:0/18:1. The higher abundance of m/z 371.17 for PS strengthens the hypothesis that this product ion is more favored when its position on the glycerol is sn-2 and seems to be independent of the phospholipid class.

24

Figure 16: Fragmentation scheme of ion m/z 683 of PS 16:0/18:1 complexed with silver. Dotted lines are drawn to indicate bond breakage as PS fragments together with the new product ion m/z (found in Figure 15).

5.3 Phosphatidylcholine in MS

: Various settings of collision energy

& MS3” product ions 381 and 371 seems to have a characteristic ratio depending on the sn-1/sn-2 attachment to the glycerol backbone. If fragmentation ratio is independent of collision energy applied it might be useful in order to determine the PC species sn-1/sn-2 structure. To investigate if the collision energy has an impact on the ratio of product ions m/z 389.17 and 371.17 in MS3, experiments were done with both PC species. Specific settings 3 (Appendix, Table 4) were used. First Standard sample of PC 18:0/18:1 and PC 18:1/18:0 were first tested at what CID yield the most precursor ion 711.42 m/z in MS2. The highest abundance of precursor ion for both isomers in MS2 was achieved at CID 30-40. Throughout the experiments, CID in MS2 was set to 40. Next, characteristic fragmentation in MS3 was investigated in respect of the ratio of fragments 389.17 and 371 m/z at different CID applied. Table 1 shows the sum for peak intensity for 371 and 389 for CID 23-45 and the percentage of the total intensity was calculated for each product ions. In Graph 1 the ratio of each fragment for both isomers is plotted as their ratio percentage of the sum of their total intensity over CID 23-45. Different levels of collision energy applied for both isomers showed that the ratio product ions of 371.37 and 389.17 stay linear (Graph 1). For PC 18:0/18:1 67.8-71.1% of the total raw intensity comprises of ion 371.17 m/z and 389.17 m/z of the remaining 28.9-32.2% (Table 1). PC 18:1/18:0 product ionsas shown in Figure 11, has the opposite product ion ratio and ion of m/z 389 is present as the higher abundant product ion compared to 371.17. 77-80.7% of the total intensity was ion 389.17 m/z and 19.3-23.1% were 371.17 m/z. Table 1: Sum of the peak intensity of product ions 371/389 and the percentage of the total intensity for CID 23-45

PC 18:0/18:1 PC 18:1/18:0

CID Tot Intensity of 371 & 389 m/z

371 m/z (%) 389 m/z (%) CID Tot Intensity of 371 & 389 371 m/z (%) 389 m/z (%) 23 65.2 68.6 31.4 23 476.7 21 79 25 100.9 71.1 28.9 25 526.5 19.3 80.7 27 297.3 68.7 31.3 27 1536.1 20.7 79.3 30 482.3 67.8 32.2 30 2784.1 23 77 35 544 69.5 30.5 35 2404.9 23.1 76.9 40 657.5 68.9 31.1 40 1697.8 22.2 77.8 45 440.6 70.4 29.6 45 1223.9 22.8 77.2

25

Graph 1: Ratio percentage for PC18:0/18:1 and PC 18:1/18:0 over CID 23-45

In the comparison of both Isomers, one can differentiate that the total intensity varies over the rising collision energies applied (Table 1). The total intensity for both isomers differs both in magnitude and at what CID the highest intensity is achieved. In Graph 2 CID 23-45 is plotted versus the total intensity of product ions 371 and 389 m/z of each isomer. PC 18:0/18:1 generally show a lower intensity of product ions throughout MS3 compared with its isomer PC 18:1/18:0. The peak intensity for fragments 389 and 371 of PC 18:0/18:1 is highest at CID 40. For PC 18:1/18:0 the different collision energies applied yields higher intensity throughout the experiment compared to its isomer and the highest abundance of fragments was achieved at a CID of 30 in MS3. The intensity of precursor ions m/z 711 in MS2 for PC 18:0/18:1 is around 1500. As MS3 is investigated the intensity of the product ions is noticed to be around 500, meaning a third of the precursor ion are fragmented into respective product ion. The intensity of precursor ions m/z 711 in MS2 for PC 18:1/18:0 is around 3000. As MS3 is investigated the intensity of the product ions is noticed to be around 2500, meaning a much larger portion is fragmented into product ions. This indicates that it’s harder to fragment acyl chain 18:1 complexed with silver positioned at the sn-2 position as a lower intensity of product ions is produced. Also, the amount of precursor ion of 711 is twice as much in PC 18:1/18:0 and could be due to concentration difference between the two samples.

0 10 20 30 40 50 60 70 80 90 23 25 27 30 35 40 45 R at io % CID PC 18:0/18:1 371 PC 18:0/18:1 389 PC 18:1/18:0 371 PC 18:1/18:0 389

26

Graph 2: Total intensity over CID 23-45 for PC 18:0/18:1 and PC 18:1/18:0

The MS3 mass spectrum (Figure 15) in the section “PS MS2 and MS3” showed a similar ratio pattern where 371 are more abundant to 389 similar to PC 18:1/18:0 (Figure 10). Indicating that product ion 371 is more favored if the acyl chain is at the sn-2 position. However, the calculated ratio for PS 16:0/18:1 showed 61% of the total raw intensity comprises of ion 371 and 389 of the remaining 39% and compared to the ratio of PC 18:0/18:1, it is not within the same ratio range (Table 1). For PC 18:0/18:1 the ratio for ion 371 was 67.8-71.1% and 28.9-32.2% for ion 389. The precursor ion used in MS3 to achieve the fragments in both cases comprises of both acyl chains and glycerol backbone. This can conclude that the ratio difference seen is independent of the polar head group. For both PC and PS species the double bond is positioned between C9 –C10. The only difference consists of the size of the saturated acyl chain positioned at the sn-1 position, 16:0 for PS and 18:0 for PC. This indicates that the ratio of product ions 371 and 389 are affected by the configuration and size of nearby sn-1 acyl chain when fragmented in MS3. In order to investigate this further ratio of different acyl chain compositions in case of size, double bond position and different headgroups need to be studied.

27

5.4 Phosphatidylcholine in MS

50:50 fragment ratio

The purpose is to investigate how the ratio changes if the sample contains a mixture of isomers as it is more likely to be the outcome of a biological sample. As ion intensity does not change the product ion ratio (see section “PC MS3: Various Collision-induced Dissociation”), CID 40 for both MS2 and MS3 were used to give consistent data and yield as much of precursor and product ions as possible. The hypophysis is that when a mixture is used, the calculated ratio will differ from both individual standard samples investigated. Standard 50:50 mixture of PC was injected by a 500µl syringe mounted to a syringe pump at the mass spectrometer. Specific settings 4 (Appendix, Table 4) were used. Figure 17 shows the MS3 mass spectrum of isolated precursor ion 711.17 with CID 40. Table 2 shows the sum for peak intensity for 371 and 389 for CID 40 and the percentage of the total intensity was calculated for each product ions. For the 50:50 sample mixture of PC the more abundant of the two characteristic fragments are m/z 389 followed by 371 (Figure 17). 62.4% of the total intensity of the two fragments was ion m/z 389 and the remaining 37.5% ion 371 (Table 2). The ratio of the two fragments is changed compared to the ratio of individual investigated standard samples of PC 18:0/18:1 and PC 18:1/18:0 in section “PC MS2

and MS3” (Table 1). The rigorous ratio pattern seen for the standard samples is different as expected for the 50:50 mixtures. The conclusion can only be that the sample is, in fact, a mixture of isomers. The change in product ion ratio would at first gaze be thought to be more of PC 18:1/18:0 cause of the higher abundance of ion 389. The higher ion efficiency and potential concentration difference for PC 18:1/18:0 shown and discussed in section “PC MS3

: Various settings of Collision-induced Dissociation” could be the cause of the high abundance of ion 389. 389 ions originating from sn-1 position are more efficiently fragmented generating more 389 ions and a difference in concentration could be a causing factor as well.

28

Table 2: Sum of the peak intensity of ion 371/389 and the percentage of the total intensity 50:50 sample mixture PC

CID Tot Intensity of 371 & 389 m/z

371 m/z (%) 389 m/z (%)

40 160.7 37.5 62.5

5.5 Phosphatidylcholine fragmentation plot

As previously shown the fragments of 371 and 389 are of different ratio depending on its isomeric appearance (Figure 10 and 11) and are independent of the intensity of CID applied (Table 1). The ratio changes if a mixture of isomers is present (Table 2). In order to try to find a qualitative way of determining the sn-1/sn-2 regioisomer ratio of PC fragmentation, a linear relationship between fragment ratios is mathematically investigated. Data from Table 1 of standard PC species at CID 40 and Table 2 of 50:50 mixtures were used. Data used is not perfect due to different intensities of each regioisomer samples analyzed. Intensity may vary because of the difference in concentration and fragmentation efficiency which will influence the final result in this section. But this shows the basic principle that it could be used. The intensity of peak 371(I371) was subtracted by the intensity of peak 389 (I389). As previously shown in Table 1, ion 371 is of higher abundance in respect of 389 for PC 18:0/18/1 and the reversed for PC 18:1/18:0. For the standard PC samples, this yields a positive absolute value for PC 18:0/18:1 and a negative absolute value for PC 18:1/18:0. The values are normalized towards the remaining intensity of precursor ion 711.42(I711) in respective MS3 experiment to give the plot an intensity/concentration independent use (Equation 1). From the 50:50 mixture of PC one intermediate value was calculated. The absolute positive value was given value x = 1, 50:50 mix x = 0.5 and absolute negative x = 0. Table 3 shows the calculated values for each sample.

𝑦 =

𝐼

𝐼

(1)

Table 3: Calculated values for PC 18:1/18:0, PC 18:0/18:1 and 50:50 mixture with Equation 1 Sample Y value X value

PC 18:1/18:0 -5.02613 0 50:50 mixture -2.07772 0.5 PC 18:0/18:1 2.701087 1

The values from Table 3 were plotted in a coordinate system in Office Excel (Graph 3). A linear trendline was chosen and the function obtained were Y= 7.7272x – 5.3312. The function will be used as the biological samples are investigated. Intensity data obtained from MS3 mass spectrum of precursor and product ions a Y value will be calculated with Equation 1 and X value from Function 1: Y= 7.7272x –

29

5.3312. Depending on the X value obtained a conclusion of the ratio between sn-1/sn-2 regioisomer species can be estimated. X value 0 represent a 100% abundance of sn-1 position phospholipids, X = 0.5 a 50:50 mix and X= 1 100% abundance of sn-2 phospholipids.

Graph 3: Data from Table 3 plotted with trendline and associated function.

To further strengthen the linear correlation in Graph 3 more mixed isomer samples are needed, for example, ratios of 40:60, 30:70, 20:80 and 10:90 for both isomers. Also, as noted in section “PC MS2 and MS3” standard sample of PS 16:0/18:1 behaved slightly different in respect of the ratio of product ions 371 and 389. As different acyl chain composition and unsaturation seem to alter the ratio this linear correlation might just be correlating for acyl chains of 18:0_18:1.

5.6 Phosphatidylcholine in a biological sample

A biological sample will contain different phospholipid classes and a variety of acyl chain compositions for each class. By adding silver ions, [M+Ag] + will be present in the mass spectrum. The purpose is to identify PC species in MS2 and further MS3 experiments in order to differentiate acyl chain composition, number of unsaturation and determine the sn-1/sn-2 regioisomer ratio. Standard biological sample containing rat brain homogenate was injected by a 500µl syringe mounted to a syringe pump at the mass spectrometer. Specific settings 4 (Appendix, Table 4) were used for these experiments.

Figure 18 shows the MS1 mass spectrum of the biological sample containing rat brain homogenate in a narrow m/z range of 865-905 to help with identification of silver isotopes. Silvers two naturally occurring isotopes, Ag107 and Ag109 have the isotopic mass of difference of 2 Da. Where Ag107 is the most abundant isotope and if normalized to 100% the isotope Ag109 correspond to 92.9% to its lower mass isotope. Silver isotopic peaks are identified at 866, 874, 882, 888, 894 and 898 (Figure 18). Appendix, Figure 5 shows the MS1 mass spectrum of the biological sample with a wider range of m/z 700-930. The isotopic patterns for silver can be identified within m/z range 860-920 and correspond to phospholipids as [M + Ag] +. Lower m/z are the same corresponding phospholipids ionized as [M + H] +, [M + Na] +. Phospholipid complexed with silver found at m/z 894 is the most abundant ion compared as ionized with Na or H. The calculated

y = 7.7272x - 5.3312 R² = 0.9816 -6 -5 -4 -3 -2 -1 0 1 2 3 4 0 0.2 0.4 0.6 0.8 1 1.2 I ( 37 1−3 89 )/ I ( 71 1) Ratio of regioisomers

30

mass difference from Ag to Na, the phospholipid ionized with Na is found at m/z 810 and ionized with H the phospholipid is found at m/z 788.

Figure 18: MS1 mass spectrum of the biological sample, range 865 – 905 m/z. Highlighted circles indicating potential phospholipid species complexed with silver.

In order to thoroughly identify a phospholipid over a specific m/z the following checklist 1-4 were used: 1. Calculated nominal mass for the peak of interest at 866 m/z without the silver atom. From

literature m/z for [M+H] + = 788 are very likely to be PC (34:1) [22].

2. In MS2 confirm the neutral loss to be 183.07 and 59.07 which verify it to be of PC species. 3. Anticipate potential acyl chain composition. PC (34:1) generates a few different structures such as

14:1_20:0, 20:1_14:0, 16:1_18:0 and 18:1_16:0

4. MS3 of precursor ion [M + Ag107+ - (183.07)] + and [M + Ag109+ - (183.07)] + to confirm the existence of anticipated acyl chain compositions generated.

5. Through intensity data from MS3 predict the regioisomer ratio with the graphical model established in section “PC fragmentation plot”.

The peak of m/z 866 is the far most abundant silver adduct (Figure 18), in literature m/z of 866 are expected to be PC (34:1)[22]. The peak investigated in MS2 show an abundance of product ions of m/z 807.25 and 683.3 (Figure 19). The calculated neutral loss corresponds to the characteristic product ions of [M + Ag+ - (59.07 u)] + and [M + Ag+ - (183.07)] + which matches the loss of the PC headgroup and it can be identified as a PC species (Figure 20).

31

Figure 19: MS2 mass spectrum of the biological sample containing rat brain homogenate, precursor ion of m/z 866.4.

Figure 20: Fragmentation scheme of precursor ion PC of m/z [M+ Ag]+ = 866.48. Dotted lines are drawn to indicate bond breakage as PC fragments together with the new product ion m/z (found in Figure 19) and the corresponding neutral loss.

In order to appoint acyl chain composition in respect of size and number of unsaturation’s MS3

of m/z 683.3 were investigated. Figure 21 shows the MS3 mass spectrum with 683 m/z as the precursor ion. The product ions of interest are m/z 389, 371 and 343 which are recognized as [R18OOH +Ag] +, [R18OOH – H2O +Ag] + and [R18OOH –H2O -CO +Ag] +. The unsaturated acyl chain is identified as (18:1). The neutral loss corresponds to the loss of the other acyl chain comprising of acyl chain 16:0 and remaining glycerol backbone. However, PC 16:1_18:0 where the unsaturation appears on the opposite acyl chain would have the same mass as the precursor ion. If PC 16:1_18:0 were present in the sample, product ions of [R16OOH +Ag] +, [R16OOH –H2O +Ag] + and [R16OOH –H2O -CO +Ag] + would also be present. No product ions could be identified to confirm the existence of (16:1). Thereby PC 16:0_18:1 is the phospholipid corresponding to ion 866 in this sample.

32

Figure 21: MS3 mass spectrum of Complex sample, precursor ion m/z 683.4.

In the biological sample, one cannot exclude the presence of both isomers PC 16:0/18:1 and 18:1/16:0. At first gaze, ion of m/z 371 is the most abundant in respect of 389.17 (Figure 21). The ratio pattern and the abundance of ion 371 are similar to ratio pattern seen in standard sample PC 18:0/18:1 (Figure 10). This indicates a higher portion of PC 16:0/18:1 rather than PC 18:1/16:0. In order to confirm this hypothesis the linear correlation between product ion ratio and their sn-1/sn-2 position in section “PC fragmentation plot” is used to determine the regioisomer ratio. Intensity data from the MS3

mass spectrum of precursor ion of m/z 683.3 (Figure 21) were used to calculated Y and X value from Equation 1 and Function 1 (Appendix, Table 5). The values Y and X values are plotted and showed in Appendix, Graph 1 together with the original trendline from Graph 3. X value of 0.769 indicated that the ratio of regioisomer of PC 16:0_18:1 is 75:25, meaning 75% of PC 16:0/18:1 and 25% of PC 18:1/16:0 respectively.

5.7 Phosphatidylcholine in a biological sample - complexity

In MS1 mass spectrum of the biological sample, silver peaks are distinguished at m/z 892, 894 and 896 (Figure 18). The predicted PC species are (36:2), (36:1) and (36:0) respectively[22]. Since LIT instrument are labeled a low-resolution instrument it can only resolve ions of a mass difference of one Da. The nominal mass difference between PC (36:2) and (36:1) are 2 Da and silver isotope mass difference are also 2 Da. This means that peak at m/z 894 is intermixed with Ag109 adducted PC (36:2) species and might give false information about the actual PC (36:1) species present. In order to determine PC (36:1) species within 894 m/z MS3 experiments must be done to nearby peaks at m/z 892 in order to exclude product ions of PC (36:2) origin and also investigate peak of m/z 896 to confirm PC (36:1) by its heavier silver isotope. Standard biological sample containing rat brain homogenate was injected by a 500µl syringe mounted to a syringe pump at the mass spectrometer. Specific settings 4 (Appendix, Table 4) were used for these experiments. The checklist presented in section “PC in biological sample” is applied herein flowing text. Shown in the result discussed in this section the low resolution obtained in the 200-400 m/z range product ion overlap in the MS3 mass spectrum preventing regioisomer ratio determination.

Ions at m/z 894 can be identified as silver adduct by the characteristic silver isotope pattern (Figure 18). [M +Ag] + are expected to be PC (36:1). The peak investigated in MS2 show an abundance of product ions of m/z 835.25 and 711.42 (Appendix, Figure 3). The calculated neutral loss corresponds to the characteristic product ions of [M + Ag+ - (59 u)] + and [M + Ag+ - (183)] +, which can be identified as a PC species (Figure 9). In the same manner MS2 mass spectrum for m/z of 892 show an abundance of product ions corresponding to PC species (36:2) and the presence of PC species of (36:3) complexed with Ag109 cannot be excluded to exist as well(Appendix, Figure 4). MS2 mass spectrum for 896 m/z show an abundance of product ions corresponding to PC species (36:1) adducted with silver isotope Ag109

33

(Appendix, Figure 5). In order to identify the individual PC species at m/z 894 MS3 mass spectrum, each of the three peaks 892, 894 and 896 are compared, starting with 892 followed by 894 and then 896. The peak at m/z 892 corresponds to ions of PC species (36:2) and PC (36:3) complexed with Ag109. Nominal mass for potential acyl chains of both PC (36:2) and PC (36:3) as ions [RxOOH +Ag] +, [RxOOH –H2O +Ag] +

and [RxOOH –H2O -CO +Ag] + were calculated and matched to MS3 spectrum of 892 m/z (Figure 22). Table 4 shows the possible PC (36:2) Ag107 adduct and PC (36:3) Ag109 adduct species based on the characteristic product ions and their expected m/z. One PC (36:3) matched the mass spectrum and it was [PC (16:0_20:3) +Ag109] +. Two species were matched as PC (18:1_18:1) and (16:0_20:2). These product ions complexed with silver isotope Ag109 are expected intermixed with PC species (36:1) at peak of m/z 894. Product ions of [PC (16:0_20:3) +Ag109] + and PC (16:0_20:2) +Ag107] + have the same m/z. Because of the low resolution, the small mass difference for the product ions does not show as separate peaks. If separated, we could differentiate them and easier determine the actual PC species present. As it is now we cannot exclude the peaks to be [PC (16:0_20:2) +Ag107] + and therefore it must be accounted for to intermix in MS3 mass spectrum at m/z 894 as [PC (16:0_20:2) +Ag109] +.

Table 4: Proposed PC (36:2) species PC species (36:2) for 892 m/z [RxOOH +Ag] + m/z [RxOOH –H2O +Ag] + m/z

[RxOOH –H2O -CO +Ag] + m/z [PC (18:1_18:1) +Ag107] + 389 371 343 [PC (16:0_20:2) +Ag107] + 415 397 369 [PC (16:0_20:3) +Ag109] + 415 397 369

34

Figure 23 shows the MS3 mass spectrum using precursor ion with m/z 894 for MS2 and 711.4 as precursor ion in MS3. First, the characteristic production ions of PC (18:1_18:1) and (16:0_20:2) found in Figure 22 were identified in respect of their heavier silver isotope of plus 2 Da (Table 5). Next, Nominal mass for potential acyl chains of PC (36:1) as ion [RxOOH +Ag] +, [RxOOH –H2O +Ag] + and [RxOOH –H2O -CO +Ag] + were calculated and matched to MS3 spectra of 894 m/z (Table 5). Some characteristic fragments originating from PC (36:2) and PC (36:1) are found to overlap. Product ions [PC (16:0_20:1) +Ag107] + and [PC (16:0_20:2) +Ag109] + have the same nominal mass (Table 5). Product ion [RxOOH –H2O -CO +Ag] + of the two species mentioned also overlap with product ion [RxOOH –H2O +Ag] + of [PC (18:0_18:1) +Ag107] +. Next, MS3 of precursor ion of m/z 896 are evaluated to confirm the potential existence of PC (36:1) species found in Figure 23.

Figure 23: MS3 mass spectrum of Complex sample, precursor ion 711.4 m/z.

Table 5: Proposed PC (36:2) and PC (36:1) species in MS3 mass spectrum of 894 m/z

PC species [RxOOH +Ag]

+ m/z

[RxOOH –H2O +Ag] + m/z

[RxOOH –H2O -CO +Ag] + m/z [PC (18:0_18:1) +Ag107] + 389 371 343 [PC (16:0_20:1) +Ag107] + 417 399 371 [PC (18:1_18:1) +Ag109] + 391 373 345 [PC (16:0_20:2) +Ag109] + 417 399 371

Figure 24 shows the MS3 mass spectrum of the peak with m/z 896 and 713.5 as the precursor ion. The characteristic production ions corresponding to PC (18:0_18:1) and (16:0_20:1) listed in Table 5 were identified in respect of their heavier silver isotope of plus two Da (Figure 24). Table 6 shows the characteristic product ions of PC (36:1) species and their respective m/z. Here no overlapping can occur from PC species of (36:0) due to no unsaturation on the acyl chains. As Silver form adducts with the C-C double bond and with no such bond, no product ions are to be found. However, some of the characteristic product ions overlap in between the three PC species (Table 6). For PC (18:0_18:1) product ion [RxOOH – H2O +Ag] + overlap with [RxOOH –H2O -CO +Ag] + of PC (16:0_20:1). As regioisomer ratio

35

determination (see section “Pc fragmentation plot”) is based on the peak intensity of product ions [RxOOH +Ag] + and [RxOOH –H2O +Ag] +, overlap in the mass spectrum will alter ion intensity and therefore the ratio and the graphical model previously established is of no use. As overlap in mass spectrum occurs for one product ion for [PC (18:0_18:1) +Ag109] + the graphical model determining the ratio of regioisomers cannot be performed as it will give inconclusive results. However, Ag109 adducted PC (16:0_20:1) do not have any discovered intensity mass spectrum overlap and regioisomer ratio can be estimated. Intensity data from MS3 mass spectrum Figure 24 were used to calculate Y and X value from Equation 1 and Function 1(Appendix, Table 5). The values Y and X values from Appendix, Table 5 are plotted and showed in Appendix, Graph 2 together with the original trendline from Graph 3. A value X = 0.604 indicated that regioisomer ratio between PC (16:0/20:1) and PC (20:1/16:0) is around 60:40 respectively.

Figure 24: MS3 mass spectrum of Complex sample, precursor ion 713.5 m/z.

Table 6: Identified PC (36:2) species in MS3 mass spectrum of 896 m/z

PC species [RxOOH +Ag]

+ m/z

[RxOOH –H2O +Ag] + m/z

[RxOOH –H2O -CO +Ag] + m/z

[PC (18:0_18:1) +Ag109] + 391 373 345

[PC (16:0_20:1) +Ag109] + 419 401 373

Analyzing a biological sample with the shotgun method in a single LIT can become rather complex, due to the low resolution achieved. In Figure 23 product ion [RxOOH +Ag] + originating from PC 16:0_20:1 complexed with Ag107 and the other PC 16:0_20:2 complexed with Ag109 are overlapping in the mass spectra. The mass difference of acyl chain (20:2) compared to (20:1) are minus two Hydrogen = 2.0157 Da and for the Ag109 isotope an addition of 1.9997 Da. This concludes a product ion mass difference of 0.016 Da. This is a far too small mass difference to separate in a LIT instrument. To calculated the actual mass resolution achieved at around m/z 400 MS3 mass spectrum in Figure 23 was used. Peak 371 m/z had a width at half height of 0.4 Da which gives R ≈ 930. The resolving power of the instrument used are seemingly not able to separate product ions of [PC (16:0_20:1) +Ag107] + and [PC (16:0_20:2) +Ag109] +

36

(Figure 25). Product ion [RxOOH +Ag] + of [PC (16:0_20:1) +Ag107] + and [PC (16:0_20:2) +Ag109] + have a nominal mass of 417.1917 and 417.1757 respectively. As shown in the section “Mass resolving power” the estimate resolution needed to separate two ions can be calculated by dividing m/z difference of two ions to m/z peak signal. In order to separate the two product ions at m/z of 417 a resolution of R ≈ 26 000 is needed. Therefore in order to resolve the product ions an instrument with a higher mass resolving power is need. This would help to differentiate product ions directly and exclude ions of different PC species originating from Ag109 adducts. No peak intensity overlap would also mean that regioisomer ratio determination can be estimated.