Analysis of phosphorylated proteins using MOAC and
LC-ESI-MS/MS with Collision Induced Dissociation
(CID) and Electron Transfer Dissociation (ETD)
Tom Panhuise (6059074) University of Amsterdam Master student Chemistry – Analytical Sciences Internship Master project, 2014-2016
Mass Spectrometry of Bio macromolecules – SILS
Professor: prof. dr. C. G. de Koster
MOAC
CID/
ETD
Supervisor: dr. L. J. de Koning Daily Supervisor: H. L. Dekker
Abstract
The analysis of Post Translational Modification’s (PTM’s) is crucial for studies on cellular signaling and other metabolic pathways. Phosphorylation is one of the most extensively studied PTM’s and is involved with many cellular functions. Both the sequence of a peptide and the localization of phosphorylation are important for phosphoproteomic research. This is the reason why a research project was set up by the Mass Spectrometry group of SILS (UvA, Amsterdam), with the goal to propose an analysis protocol for phosphorylated peptides. An extensive literature study was performed in search of relevant protocols which might be implemented and/or adapted for an analysis setup. The resulting protocol combines trypsin digestion followed by enrichment via Metal Oxide Affinity Chromatography (MOAC).The sample is then introduced to Liquid Chromatography (LC), Electron Spray Ionization (ESI) and tandem mass spectrometry (MS/MS). The utilized nanoLC-amaZon Speed ETD/ mass spectrometer (Bruker), enabled the use of both Collision Induced Dissociation (CID) and Electron Transfer Dissociation (ETD) analysis. The fact that ETD leaves labile modifications intact during fragmentation makes it uniquely qualified to be used in phosphorylation site determination. Explorative analysis of different reaction times for ETD analysis proved that the default setting (100 ms) is optimal with regard to intensity of the ions whilst minimizing the potential Proton Transfer Reaction (PTR). Enrichment proved to be most successful when using commercially obtained TiO2 columns instead of ZrO2 or Ti/ZrO2 columns. Additionally, the use
of L-glutamic acid and citric acid as additives for improved enrichment of phosphopeptides was investigated. The results indicate that the use of 70 mM L-glutamic acid yields the best
enhancement of enrichment. Elution of the peptides requires a high pH, which was supplied by a buffer containing 2M, NH4OH (pH 12). Using a categorization of the “proton mobility” of a
peptide, the likeliness to experience neutral loss of phosphoric acid during mass spectrometry analysis was assessed. The resulting workflow has been tested with both simple and more complex samples, regarding the amount of nonphosphorylated material in the sample. However, the resulting work flow might benefit from additional purification steps which are orthogonal to the MOAC enrichment, such as strong cation exchange chromatography (SCX). Additionally, optimization of the parameters of the ion trap system and subsequent data analysis could also greatly improve the overall analysis.
Keywords: Posttranslational modifications (PTM), Phosphorylation, Metal oxide affinity chromatography
(MOAC), Reversed phase liquid chromatography (RP-HPLC), Electrospray ionization (ESI), Tandem mass spectrometry (MS/MS), Collision induced dissociation (CID), Electron transfer dissociation (ETD).
Index
1. Introduction
2. Phosphoproteomic Research
2.1 Sample preparation
2.2 Phosphorylated protein enrichment
2.3 Mass spectrometry analysis of phosphoproteins
3. Materials and Method
3.1 Reagents and materials 3.2 Methods
3.2.1 Sample preparation
3.2.2 Phosphopeptide enrichment 3.2.3 LC – MS/MS Analysis 3.2.4 Data analysis
4. Results and Discussion
4.1 Optimization of enrichment 4.1.1 Different MOAC materials
4.1.2 Additives/buffers experimentation 4.1.3 Additive optimization 4.1.4 Elution optimization 4.2 Identification techniques 4.2.1 CID/ETD fragmentation 4.2.2 Proton mobility
5. Conclusion
6. References
1. Introduction
Adaptive nature of cellular machinery is continuously required in order to react to changing
conditions in the environment. In order to do so, many proteins are produced and modified in order to maintain the metabolism, structural integrity and specific functions of the cell. Amongst many different methods to cope with these changes, posttranslational modifications (PTM’s) of proteins is an often used strategy by cells.1-3 These modifications consist of either adding extra compounds or
removing a specific part of the protein, due to which physical-chemical properties of the proteins change. Some examples of these PTM’s are phosphorylation,SUMOylation, acetylation,
glycosylation, although many more exist in nature.1-3 Phosphorylation is one of the most extensively
studied PTM’s, and is responsible for protein-signaling transduction, regulation, and many other functions.1, 2, 4-9 When phosphorylation goes wrong, and the subsequent pathways and reactions
depending on this modification go awry, many kinds of pathological symptoms can be expressed, such as cancer.10 Phosphorylation of proteins can occur on the specific residues: serine, threonine and
tyrosine. These three amino acids have shown to be phosphorylated to different extent in nature. Where serine is phosphorylated frequently (86.4%), threonine (11.8%) and tyrosine (1.8%) are less common phosphorylation sites.3 However, phosphorylation on basic residues (such as histidine,
arginine and lysine) might also be phosphorylated to a certain extent.11
Figure 1. The possible workflows for proteomics research, adopted from Switzar et al. (2013).12
Here, three different classifications for proteomic research have been depicted: Top-down proteomics, Middle-down proteomics, and Bottom-up proteomics. The main difference between these three approaches is the size of the analyte, being the greatest in top-down (intact proteins) and the smallest in bottom-up proteomics (small peptides). This study would be classified as a “bottom-up” proteomics approach.
In order to study the effect of specific conditions on post translational modifications of proteins, a means of monitoring the localization, concentration levels, and specific classes of PTM’s is required. The most popular methods nowadays use some form of mass spectrometry (MS) in combination with isolation, purification and enhancement steps in order to investigate these modifications.1, 4, 5, 13, 14 This
choice of workflow is well suited for the small quantities, dynamic range and many different isoforms of phosphorylated proteins, which pose a significant challenge to analytical analysis.
Especially orthogonal purification methods have been studied extensively.1, 4, 5, 13 Depending on the
research, different approaches are defined, which are named “top-down”, “middle-down”, and “bottom-up”, see figure 1.12 These proteomic approaches use a different general set-up for the
experiment, and focus either on whole protein analysis (top-down; ≤50 kDa) or peptide analysis (middle-down; 2-20 kDa, bottom-up; 0.5-3 kDa).12, 15 The different physiochemical properties of
peptides can be exploited by using multiple purification methods into one workflow. Since the analytes are proteins, important parameters to separate on are properties such as: pH,
hydrophobic/hydrophilic retention, adsorption of specific functional groups to affinity columns, and many more.1, 4, 5, 15-17
To this day, many different analysis approaches have been established for the analysis of
phosphorylated proteins using LC-ESI-MS/MS. Thus far, no standard phosphoproteomic analysis workflow was present at the SILS Mass Spectrometry group (UvA). However, improved
understanding of these posttranslational modifications is of the utmost importance for development of new cures against diseases, as well as for the understanding of cellular mechanisms. Therefore, the aim of this research project is to implement and optimize a standard workflow for the analysis of phosphorylated proteins. Additional information regarding sample processing, modifications and analysis techniques have been reviewed, but only the relevant techniques are explored in depth.
2. Phosphoproteomic research
A typical workflow (see figure 2) consists of isolation of the proteins followed by fractionation, possible enhancement, and finally LC-MS/MS. Obtained spectra are then subjected to database searching and specific software in order to identify the peptides and localize the phosphorylation sites.4, 5, 15-17 Many phosphoproteomic studies start their workflow with extraction of proteins from
cells or tissue followed by digestion. This procedure is often applied before analysis, and many different methods exist for digestion.12
Figure 2. The typical workflow of a “bottom-up” phosphoproteomic research, as illustrated by C. Yang et al. (2014).15
In this workflow the cells are first broken down after which the protein is harvested. Afterwards, an enzymatic digestion and subsequently fractionation are used to obtain specific protein mixtures. These mixtures are enriched (one of many possible methods) for phosphopeptides and subjected to LC-MS/MS, after which a database search and subsequent analysis with specific software is performed.15
2.1 Sample preparation
After selection of the appropriate part of the organism under investigation (cellular membranes, cytosol, or other) the harvested cells are lysed. The choices of homogenization buffer and
subsequently denaturing buffer are important, as was shown by Hoffer et al. (2008), whom reviewed the effects of different buffers.4 Here, they showed the increased identification of phosphopeptides
when using a 50 mM ammonium bicarbonate buffer, compared to the use of 6 M guanidine, or 1% SDS. Interestingly, the sample had not been denatured beforehand, indicating that protease digestion does not need a denaturing agent. However, the reproducibility of these samples was less than when a solubilized sample was used.
Digestion of the isolated protein mixture can be performed in many different ways. A detailed review of the recent developments and available applications on protein digestion was performed by Switzar et al. (2013).12 Possible ways of digestion consist of enzymatic digestion, and non-enzymatic
digestion, which are performed either in gel or in solution. The most common approach is enzymatic digestion of proteins using trypsin, which has been perfected in the past years, resulting in a highly efficient, low cost protease.18 This enzyme is capable of cutting peptide bonds C-terminal to the basic
amino acids lysine and arginine, unless followed by a proline.19 Upon digestion with this enzyme,
medium sized peptides (~14 residues) are obtained, which have a charge states of at least 2+.12
been developed.4, 5, 12 For example, Lys-C, Arg-C, Asp-N, and Glu-C are some of the different
endoproteases, which can be used under different conditions than trypsin.12 For example, Lys-C and
Arg-C are able to cleave lysine and arginine N-terminally, respectively, and provide high efficiency and specificity. As mentioned by Switzar et al., Lys-C is a recent addition to the list of proteases, and is able to perform under intense denaturing conditions (high temperature, 8 M urea, or 80%
acetonitrile).12 Peptides obtained with these kind of enzymes can vary in length (longer) and charge
state (higher), making them ideal choices when “middle-down” analysis is performed.12
Besides using proteolytic enzymes, proteins can also be digested by using chemical cleavage or instrumental techniques, which have also been reviewed by Switzar et al.12 Usually acids are used
such as formic acid (FA) or acetic acid (HAc), but more are possible. Other examples of chemicals used for protein digestion are cyanogen bromide (CNBr) and 2-nitro-5-thiocyanobenzoate (NTCB). These kinds of chemicals often have a very specific cleavage site, which differs per chemical, resulting in the same level of specificity of digestion as when enzymes are used. The obtained
peptides are usually of the size that is useable in middle-down proteomics. The application of a certain chemical for the use of digestion is therefore strongly dependent on the aim of the experiments. Besides chemicals, instrumental applications have been developed, such as electrochemical oxidation, or the acceleration of digestion (enzymatically) via ultrasound, microwave, infrared, or solvent applications.12, 15, 16, 20 Additional techniques such as immobilized enzymatic digestion, in which the
digestion enzymes have been covalently linked to a support, have also been explored.4, 5, 12 The main
reason for the development of these non-enzymatic digestion methods is automation. By creating a solid supporting material, or technique with the same resulting digestion as with enzymes, online applications have been realized. Additionally, these techniques require significantly less time to obtain the same results as those obtained whilst using enzymes. Another important aspect of non-enzymatic digestion is the behavior of the sample under the applied conditions (highly acidic or basic). Denaturation is readily achieved when the pH is dropped below 7.4, and might influence affinity-based purification, which is based on specific properties of the target analyte. Therefore, the use of digestion methods other than enzymatic digestion should only be used with a suitable, adapted workflow. Depending on the application and institute, the digestion of samples can be achieved in many different ways. Whereas a high-throughput workflow (non-enzymatic, online applications) might be required in settings such as hospitals or factories, research facilities might benefit more from the use of labor-intensive, but highly specific protocols.
2.2 Phosphorylated protein enrichment
After creating the peptides from the proteins, some pre-fractionation or isolation steps may be performed.4, 5, 12, 15-17, 20, 21 This is usually the case when only a small, select part of the entire sample is
of interest (i.e. phosphorylated protein analysis from a whole cell lysate). Many different applications are used, depending on the specific properties of the analyte of interest, see figure 3.16 The potential
ionic and Lewis base interactions that phosphate groups are capable of, are important factors for enrichment method development. Besides the charge-based methods, alternative chemical
modifications, such as β-elimination with a Michael addition,15 have been developed. However, these
kind of techniques are prone to introduce incomplete-, and side-reaction products, resulting in sample loss and further complication of subsequent analysis. Therefore, chemical modifications will not be reviewed here, but can be explored in extensive reviews.4, 5, 12, 14-17 Immuno-precipitation is another
enrichment strategy, but is mainly applicable to phosphopeptides containing phosphotyrosine. Both phosphorylated serine and threonine are not readily enriched with this technique, due to lack of specific antibodies.16 Therefore, additional information on this enrichment route can be found
elsewhere.15-17
Figure 3. Possible enrichment applications for phosphate containing molecules, by Leitner et al.16
Structural recognition can be used by applying specific antibodies to phosphate groups. Phosphate is able to retain charge, depending on the pH, and is a polar group, which can be exploited by ion exchange chromatography, antibodies and partially metal affinity techniques. Another target of enrichment is the chemical reactivity of the phosphate group.16 In this research, only the Metal Affinity (MOAC) enrichment has been explored.
Phosphorylated peptides are often enriched by means of metal-based materials.22-49 Metal-oxides or
immobilized metal-oxide-matrices are able to bind the phosphate groups in a selective way, see figure 4.23 When the pH is chosen appropriately, the free oxygen atoms on the phosphate group can
coordinate to the metal, whilst (most) other peptides are not able to do so.43 This gives a powerful way
of selectively enriching the sample before applying it to subsequent LC chromatography and mass spectrometry. Many different adaptations of this concept have been explored, such as different kind of metals (titanium, zirconium, germanium, etc.) as well as different kind of matrices (columns, filters, porous beads, etc.). 22-49 Most of these methods show selective behavior towards the binding of
phosphopeptides in varying effectiveness. Titanium (Ti), iron (Fe), zirconium (Zr), and germanium (Ge) are some of the more popular choices and have been applied in immobilized-metal-affinity chromatography (IMAC) or metal-oxide-affinity chromatography (MOAC). 15, 22, 25, 28, 36 Evaluating
experiments on these different metals showed a difference in the enrichment efficiency, depending on the complexity of the sample and the preparation of the MOAC material, either ZrO2 or TiO2 can be
However, many experiments do not rely solely on metal-based enrichment applications and often incorporate additional separation steps, such as strong cation exchange chromatography (SCX) or antibody-based enrichment, but these kind of approaches have not been considered for this study.7, 8, 15, 17, 21, 49, 50
Figure 4. Coordination of phosphate and chelating bidentates when adsorbed to the TiO2 surface. Other metal-based materials are prone to the same kind of coordination. This specific coordination of both phosphate and carboxylic acids is utilized during additive enhancement of MOAC enrichment. Figure adapted from Larsen et al.23
Besides the enhancement using metal-based materials, many additives to the loading and washing buffer compositions have been investigated for the further improvement of the isolation. These kinds of additives are used to flush out retained molecules which have adsorbed to the material through acidic groups that are also able to coordinate to the metal-based materials if the pH is suitable.27 Most
additives are either pH regulators or inhibitors that replace unwanted, retained molecules by competitive binding, such as depicted in figure 4. For the enhancement of phosphorylated peptides using MOAC-type of applications, carboxylic acids have been studied extensively.23, 34 Of these, the
substituted aromatic carboxylic acids (2,5-DHB, salicylic acid, and phthalic acid) outperformed the other types of acids tested (such as trifluoroacetic acid, acetic acid, cyclo-hexane-carboxylic acid and benzoic acid) in the prevention of non-phosphorylated peptide retention.23 From these examples, the
aromatic compound 2,5-DHB has been studied in great detail and proved to be an effective way to further improve the selectivity of the enhancement of phosphorylated peptides.23 However, the use of
this additive is limited to a setup using a MALDI ionization source, since it is also part of the MALDI sample solutions. Usage of this specific additive in other type of ionization source (for example ESI) might lead to severe contamination or damage. Moreover, glutamic acid (0.1 M solution) has shown to be more effective than 2,5-DHB when used in combination with TiO2 for phosphopeptide
2.3 Mass spectrometry analysis of phosphoproteins
Following the enhancement step an often applied separation method is Reversed Phase High Pressure Liquid Chromatography (RP-HPLC) or a closely related system.4, 5, 13, 15-17, 20, 24, 38, 40, 41, 51, 52, 53 Liquid
chromatography (LC) systems can be directed to an ElectroSpray Ionization (ESI) needle from which the liquid phase is evaporated and gas phase ions remain.41, 52, 54 Ionization using ESI depends on the
Rayleigh limit principle, which is concerned with the amount of electrostatic charges per volume density. When a droplet evaporates to such an extent that the charges contained within start to repel each other, the droplet explodes into many tiny droplets. These small droplets can subsequently reach the Rayleigh limit again and explode, leaving charged gas phase ions in the end. By applying ESI to LC fractions, separated analytes are ionized in a “soft” way that keeps them intact and applies a relative high charge state. This is especially useful for low-efficiency fragmentation methods (ETD) or more robust applications (CID). The obtained ions are suitable for injection into the MS system and are subjected to a series of focusing lenses and quadrupoles, which direct the ions to the ion trap as can be seen in the example given in figure 5.55 When reaching the ion trap, fragmentation is either
induced by adding an inert collision gas (usually helium) for CID or other means of fragmentation are initiated.
Figure 5. The setup of the AmaZon Speed / ETD mass spectrometer as purchased from Bruker, Daltonics. This mass spectrometer is used in combination with nano-LC-ESI inlet. In this model an optional ETD chemical ionization source is present, see enlarged section.55
The fragmentation of the peptide ions is not exclusively limited to collisional activation, since there are many other ways to introduce energy into the compounds being analyzed. Most of the techniques require a specific setup or modification to a standard mass spectrometer in order to perform the required fragmentation. The fragmentation techniques explored here are: Collision Induced Dissociation (CID), Electron Transfer Dissociation (ETD) and a hybrid mode in which CID is followed up by ETD or vice versa.14, 55-63
Whereas the CID technique is straightforward, the ETD technique requires a specific ion source that is capable of producing radical anions (usually fluoranthene 13 is used as reactant). These two
fragmentation methods produce different kind of fragment ions, due to differences in the methods of introducing energy and the resulting fragmentation reactions.
The CID technique produces b- and y-ions which are the result of a peptide bond being broken (Ccarbonyl-N, cleavage activation energy required ~40 kcal/mol)64, see figure 6. However, the energy
supplied during the collisions is more than enough to also break the weaker phosphate bond (O-P bond, cleavage activation energy required <20 kcal/mol)65, which results in either neutral losses of
phosphate related fragments (H3PO4, 98 Da loss; HPO3 or HPO3 + H2O, 80 Da loss) or charged losses
(PO2-, 63 Da loss; PO3-, 79 Da loss; H2PO4-, 97 Da loss). Electron transfer dissociation however,
produces c- and z-ions, see figure 6.56, 57, 66-71 These ions are formed after breaking the N-C
α bond of
the peptide backbone and are the result of radical reactions.67 Unlike CID, ETD fragmentation does
not target the labile bonds between phosphate groups and the peptide, making it an ideal method to study the localization of the phosphorylation.57, 68, 69, 72, 73 However, ETD does require high charge
densities on the peptides of interest, due to the formation of non-dissociated charge reduced radical cations.14, 57 These kinds of ions are not effective in the investigation of amino acid sequence,
phosphorylation site localization or peptide identification. Multiple methods exist in order to make sure that ETD results in highly efficient ionization and fragmentation, such as the application of CID after ETD (ETcaD).56 Other methods consist of using different digestion enzymes that produce higher
charge states peptides, or the use of additives which increase the charge state of the peptides in solution, as has been described previously.4, 5, 12-14 However, upon digestion with trypsin,
phosphorylated peptides will usually carry a low charge state due to the deprotonated phosphogroups.
Figure 6. Nomenclature of the fragmentation ions of the peptide backbone.
Traditional CID fragmentation yields b- and y-ions, whilst ETD fragmentation yields c- and z-ions. The numerals denote the amount of amino acid residues contained within the ion. The amino acid residues (the specific groups) are denoted with R.14
Many different mechanisms have been suggested for the neutral loss of phosphate related fragments upon ionization (H3PO4, 98 Da loss; HPO3 or HPO3 + H2O, 80 Da loss).14, 59, 60, 63, 74 Although it was
thought that these reactions occurred via β-elimination with the charge remotely placed, Palumbo et al demonstrated that this is not the case.74 Controversially, the evidence supported a “charge-directed”
mechanism in combination with a SN2 participation reaction, which led to the creation of a cyclic
product ion, as can be seen in scheme 1. In this study they also promoted the relations between proton mobility and the gas-phase fragmentation reaction pathways of modified peptides, as will be
Scheme 1. Three potential reactions for the loss of a neutral group (H3PO4) as demonstrated from a phosposerine containing peptide.
Pathway A has been widely accepted and supposedly supported by empirical data. Whereas, both pathway B and C have been obtained by molecular orbital calculations and experimental data. A: Charge-remote β-elimination reaction. B: Charge-directed E2 elimination reaction. C: Charge-directed SN2 neighboring group participation reaction.74
Since the development of ETD several different mechanisms have been proposed for the transfer of an electron to a peptide. The Cornell-mechanism, as theorized by McLafferty and colleagues, is based on the uptake of an electron at a protonated site, see scheme 2 (usually a basic amino acid residue or the N-terminally amine group).57 This results in a radical species which is typical for these kinds of
mechanisms. The excited electron would then undergo relaxation upon which a proton is transferred to the oxygen located in the peptidebond. In turn, this would result in an intermediate species
containing an aminoketyl radical, as can be seen in the third section of scheme 2. The last step in this mechanism is the cleavage of the N-Cα bond situated to the right of the radical site. This reaction
mechanism is charge-directed in the sense that a protonated site initiates the fragmentation. The site of protonation therefore dictates which kind of ions are observed; c- or z-ions.
Scheme 2. The Cornell mechanism as described by McLafferty et al.
This mechanism shows the fragmentation of the N-Cα bond upon electron transfer dissociation with a charge directed from a C-terminal amine group. Note that this could also be initiated from the N-terminus resulting in differently charged c- and z-ions.57
Charge-remote fragmentation has also been investigated and different mechanisms have been
suggested, such as the Utah-Washington mechanism, see scheme 3.57 Here, the initial electron uptake
is thought to happen directly on the amide group, specifically in the π* antibonding orbital. This forms the aminoketyl radical intermediate which was also found in the previous mechanism indicating the importance of the creation of this intermediate species. Depending on the specific mechanism, the anionic charge is negated with the attraction of a proton and the amide bond is broken (Washington mechanism, Utah supposes the reverse order). Depending on the source of the proton used in the neutralization of the anion, either a c- or z-ion can be observed.
Scheme 3. The Utah-Washington mechanism as proposed by Simons et al and Turecek et al.
This mechanism shows the fragmentation of the N-Cα bond upon electron transfer dissociation without charge initiation . Note that this could also be initiated from the N-terminus resulting in differently charged c- and z-ions.57
Another mechanism suggested by Tsybin et al uses the “enol”-like properties and differs greatly from the previous described mechanisms.57 The first difference is the location of the fragmentation of the
N-Cα bond, which occurs N-terminally rather than C-terminally, see scheme 4. Moreover, the amide
bond is cleaved heterolytically in contrast to the previously used homolytical cleavage. This results in the highly basic c-fragment and the z+•-ion as seen in the last panel of scheme 4. The resulting charge and composition of the fragments are independent of the initial hydrogen donation site, which is also different in the previously mentioned mechanisms.
Scheme 4. The “enol” mechanism of ETD as suggested by Tsybin et al.
Here the mechanism is shown for the fragmentation of the N-Cα bond upon electron transfer dissociation. The c-fragments often abstract a proton from the z+•-ion in order to form the c’-ion.57
Fragmentation obtained by using either CID or ETD techniques results in mass-to-charge ratios in a MS spectrum. The fragments can be directly analyzed (MS) or can be subsequently filtered for specific target ions (MS/MS, MSn), resulting in cleaner spectra of those ions.14, 56-59, 62, 75-77 Using a
peak picking acquisition method for additional rounds of fragmentation of a specific ion is an often used approach. This is especially useful in combination with CID, by scanning for signals that result from the neutral or charged losses of phosphate groups. From the subsequent fragmentation of the isolated ion, the peptide on which the phosphate group was present can be identified.
Although historically the obtained spectra were analyzed by hand, nowadays large databases and extensive software packages exist to help organize and analyze the obtained results.77-83 Firstly, the
obtained raw spectrum is subjected to software to clean up and prepare the data for being used in database searches. Some of these programs can also annotate sequences from the raw data, when appropriate methods have been used. Matching the peaks in the spectrum to specific peptides is performed by search engines such as MASCOT,3 which uses statistics and probability based scoring
methods.81-83 After the link has been made between peptide and protein, specific parameters can be
monitored in order to optimize the entire method. These include protein coverage, which states how much of the whole protein has been covered by the obtained peptides, and protein scoring. Usually a probability based method that states the expectancy values of obtaining the same scoring using random entries is used to gauge the significance of the match.83
3. Materials and Methods
3.1
Reagents and materialsDithiothreitol (DTT), iodoacetamide (IAA), citric acid, MS PhosphoMix 2 Light (MSP2L), β-casein (from bovine milk), ammoniumbicarbonate (NH4HCO3) and L-glutamic acid were obtained from
Sigma Aldrich (St. Louis, MO, USA). Water (H2O, ULC-MS grade), acetonitrile (ACN, ULC-MS
grade), trifluoroacetic acid (TFA, ULC-MS grade) and formic acid (FA, ULC-MS grade) were purchased from BioSolve (Dieuze, France). Hydrochloric acid (HCl, 7.7 M) and ammonia solution (NH3, 25% solution) were procured from Merck (Kenilworth, NJ, USA). Bovine serum albumin (from
bovine pancreas) was obtained from Boehringer Ingelheim (Ingelheim am Rhein, Germany). Trypsin (Trypsin Gold-Mass Spec Grade) was purchased from Promega (Madison, WI, USA). OMIX C-18 pipette tips (100 μL, 80 μg capacity) were obtained from Agilent Technologies (Santa Clara, CA, USA) and OMIX C-18 pipette tips (10 μL, 8 μg capacity) were bought from Varian (Palo Alto, CA, USA). TopTip micro-spin columns (TT1) packed with MOAC media (TiO2, ZrO2 and a mixed
material of 50/50 Ti/ZrO2) were purchased from Glygen (Columbia, MD, USA).
3.2
Methods3.2.1 Sample preparation
Samples from proteins were digested before analysis with the use of trypsin (Gold, Promega). Both bovine serum albumin and β-casein were digested using 50 μg of protein. The digestion buffer for BSA consisted of 0.1 M ammonium bicarbonate and 10% acetonitrile and water. Reduction of the protein was performed by adding 5 mM dithiothreitol to the mix and letting it react at 55˚ C for 30 minutes. Subsequent alkylation of the cysteine residues was performed using 15 mM iodoacetamide, which was left to react for 20 minutes in the dark at room temperature. After reduction and alkylation, digestion was performed by adding 1:50 (w/w) trypsin at 37˚ C overnight. Afterwards, the digestion mix was lyophilized using liquid nitrogen and a freezedryer (Thermo, HETO PowerDry LL1500 Freeze dryer). The dried sample was then resuspended in 20 μL of a solution containing 50% ACN and 2% formic acid and frozen with liquid nitrogen and stored at -26˚ C until further use. The final concentration of these aliquots was 0.04 nmol/μL.
β-Casein is a phosphorylated protein and was used because of the mixture of phosphorylated and nonphosphorylated peptides obtained upon digestion. This protein was digested using a similar method as before, although it contained 20% ACN in the digestion buffer and no reduction or
alkylation was necessary prior digestion. The end concentration of the β-casein digestion aliquots was 0.1 nmol/μL.
A synthetic phosphopeptide kit was used (MSP2L) which is a mimic tryptic digestion of HeLa cell proteins, and no digestion was therefore necessary. This kit contained a total of 200 pmol of peptides, 20 pmol per peptide, see supplemental information. Upon arrival, the kit was dissolved in 40 μL of a solution containing 20% ACN and 0.1% FA as was suggested in the manufacturer’s notes. Aliquots were made of 2 μL, which were snap frozen using liquid nitrogen and were stored at -80˚ C until use. These aliquots contain a concentration of 5 pmol/μL of total peptides, or 0.5 pmol/μL for the separate peptides. Upon arrival a preliminary test run was performed to check the visibility and behavior of the peptides after running a 30 minute LC gradient and CID fragmentation in a MS/MS analysis. From this data it was determined that not all the peptides (especially 2.2) were readily observable, see supplemental information.
3.2.2 Phosphopeptide enrichment
Phosphopeptide enrichment was carried out using the TopTip micro-spin columns (TT1) with different packing material (TiO2, ZrO2 and Ti/ZrO2). The enrichment relies upon the coordination of
the phosphorylated peptides to the material and several washing steps, as described before. A standard enrichment workflow consists of activation of the MOAC material by subjecting it to the loading buffer (without introducing the sample). Afterwards, the sample (diluted in the loading buffer) is loaded on the material and subsequently washed with more loading buffer. This is followed by additional washing with a solution containing at least 65% ACN and 2% TFA to remove non-coordinated peptides. The next step is an elution with a very basic solution which needs to be acidified quickly afterwards in order to retain the phosphategroups on the peptides. A general workflow can be seen in scheme 5. During this experiment the following samples were obtained before, and during enrichment: Negative control, Flow through plus wash fraction, and Eluted fraction. The control sample was obtained by splitting the untreated sample in equal portions and storing one of these portions as control, the other half was used for subsequent enrichment. For example, when using the kit (MSP2L), samples were prepared by adding 2 μL of the kit into 18 μL loading buffer and splitting this sample in 10 μL enriched and 10 μL negative control. Flow through and wash fraction were obtained during loading and washing of the sample and were pooled together. The eluted fraction was obtained after elution and acidification of the retained (phospho)peptides. Due to the chemicals in the loading buffer, samples should be desalted and lyophilized before analysis. The resuspension buffer which was used to dissolve the samples after desalting and freeze drying, contains 2% ACN and 0.1% TFA. It was also important to choose the right capacity for the desalting pipette tips, since some samples may contain a large amount of peptides (i.e. wash fractions should contain all the nonphosphorylated peptides). A specific loading buffer was used when additives were used during the enrichment, in which the sample was dissolved. The different loading buffers used during this study were:
o 80% ACN, 2% TFA o 65% ACN, 2% TFA
o 65% ACN, 2% TFA, L-Glutamic acid (35 mM, 70 mM, and 0.14 M) o 65% ACN, 2% TFA, Citric acid (50 mM, 0.5 M, and 1 M)
From these loading buffers, only the last two contained an additive. These same loading buffers were used to activate the column and wash the loaded sample. The sample was then washed using a different washing buffer. The used washing buffers were:
o 80% ACN, 2% TFA o 65% ACN, 2% TFA
The elution buffer that was used during these experiments has been used with two different pH values:
o NH4OH (2 M (~2% solution), pH 10.5 and pH 12)
This solution was directly acidified after elution, using 1:1 (v:v) of a 10% TFA solution. This was done to prevent the loss of phosphoric acid due to the basic environment.
General workflow of phosphopeptide enrichment using MOAC columns: Dissolve the sample in the loading buffer. Keep in mind that you will split the sample in two:
one for enrichment, one for control.
Gently tap the TopTip column on a flat surface so that the material settles at the bottom. Remove the red cap on the top and hold the column over an Eppendorf container. Activate the material with 10 μL of the loading buffer and use a syringe to push the liquid
through (you can’t use negative air pressure as it will disturb the material). (3x) Switch the column to a new Eppendorf marked as the Flow through + wash sample. Load half of the prepared sample on the column. Store the rest of the sample as Control. The flow through could be loaded again, to make sure that all phosphopeptides are bound to
the material.
Wash the column with 10 μL of the loading buffer (3x). Wash the column with 10 μL of the washing buffer (3x).
Switch the column to a new Eppendorf marked as the Eluted sample. Elute the sample with 10 μL of elution buffer (3x).
Quickly acidify the eluted fraction with 30 μL of 10% ACN solution.
If an additive was used, choose a tip with appropriate desalting capacity, and desalt.
Afterwards, freeze dry the sample and resuspend in such a volume that the concentration is equal to the concentration you have in the control sample.
Store the samples at -20° C until analysis.
Scheme 5. A general workflow for the enrichment of phosphopeptides using MOAC columns.
Depending on the composition of the loading-, washing-, and elution-buffers, steps may vary between experiments. Modified from Glygen manual as provided with purchase of the TopTip.
When enrichment was performed on the phosphopeptide kit (MSP2L) the amount of sample loaded onto the columns was 1 μL mixed into 9 μL loading buffer (50 fmol/ μL) the same amount was set aside as a negative control sample (untreated sample).
The same amount of sample was loaded for analysis for each separate experiment. This means that results from different experiments cannot be readily compared to each other, but comparison per experiment should be able. However, scaling of the data with regard to the mass balance between the samples was not performed due to lack of a suitable method and the small amounts of the samples. Unfortunately, all the experiments performed during this study have been done only a single time due to time constraints. This means that all the data presented represents single measurements and can only be used to give indications about certain trends. These results therefore represent the first steps into developing a phosphopeptide specific enrichment workflow for the mass spectrometry group at SILS (UvA).
3.2.3 LC – MS/MS Analysis
After enrichment of the phosphopeptides, the samples were analyzed using a nanoLC (Easy – nLC II, Bruker) coupled to an Advance Captive Spray (ESI) in combination with the amaZon Speed / ETD mass spectrometer from Bruker. The ACS electrospray source is a modification from the original Apollo II Electrospray Source which is supplied with the amaZon Speed mass spectrometer. The Advance Captive Spray source is easier to operate and tune in contrast to the other source.
The nanoLC was set to undergo a linear gradient using two solutions: solution A: water, 0.1% formic acid solution, and solution B acetonitrile, 0.1% formic acid solution. The linear gradient was
completed in 25 minutes with 5 minutes delay at the start, making the LC-run 30 minutes total. The gradient starts at 5% B and 95% A, which was ramped up to 12% B and 88% A in the first 5 minutes. During the next 15 minutes the gradient was slowly increased to 30% B and 70% A, which was the highest concentration used in peptide/protein analysis. After this functional gradient the column was flushed with 95% B and 5% A over 4 minutes and reset to 5% B and 95% A during the last minute. The nanoLC setup was operated at a flow rate of 300 nL/min.
The amaZon Speed / ETD mass spectrometer used for this study has been used with the standard settings as supplied by the manufacturer, unless specifically stated otherwise (i.e. ETD reaction time experiment). Only the “positive mode” was utilized during this, the “negative mode” has not been explored. However, the “negative mode” might be worth looking into because of the negative charges phosphopeptides carry. Especially for multiple phosphorylated peptides the resulting charges after protonation might not be suitable for “positive mode” detection. When using MS/MS analysis, the first round used “enhanced resolution” whereas the second used “Xtreme resolution”. Additionally, “enhanced resolution” used the average of 5 measurements for every spectrum, whereas “Xtreme resolution” used averaging of 2 measurements. The resolution obtainable when “Xtreme Resolution” was applied, was 2+ ions with a scan speed of 52,000 u/sec. When using the “Enhanced Resolution” these parameters changed to a resolution of 4+ ions with a scan speed of 8100 u/sec.55 These rounds of
fragmentation use helium as collision gas and require a fragmentation time of 28 milliseconds. The mass range analyzed with this setup ran from 300 m/z to 1400 m/z.
Figure 7. A schematic representation of the ETD reaction during ETD analysis, from the brochure of the AmaZon Speed / ETD mass spectrometer as purchased from Bruker, Daltonics.
In this simplified reaction scheme the produced radical anion reagent from fluoranthene reacts with the peptide of interest after which the N-Cα bond is broken.55
For ETD measurements the specific chemical ionization cartridge had to be activated, see figure 5. This cartridge uses fluoranthene and methane gas to create an electron donating radical anion species which enables the cleavage of the N-Cα bond in a peptide, see figure 7. The reactions towards the anionic radical species start with the transfer of high energy electrons (~60-80 eV) from filaments to the mediator methane. After this interaction, two lower energy electrons are available for a reaction with the fluoranthene, which will result in C16H10●─, as can be seen in scheme 6. However, this
specific chemical is also used for the Proton Transfer Reaction (PTR), which differs from ETD in the sense that it donates a proton instead of creating a radical species. This PTR reagent is easily created when a proton is absorbed by the radical anion, creating a C16H11─ carbanion, see scheme 7. Both
reactions can be carried out using the same equipment and with only minor tweaking of chemical ionization reaction time and heating of the fluoranthene. This very flexible setup gives a choice to the user, but also poses an inherent problem. Due to the dual nature of the chemical used for PTR/ETD reactions, both reactions take place to some extent.71 Although the manufacturers have optimized the
parameters for each kind of reaction, some side reactions are to be expected.71
Scheme 6. The reactions leading to the creation of the radical anion species used during ETD analysis. The reaction is started with a filament that is heated up to release high energy electrons (~60-80 eV) which can be captured by the methane. Absorbing part of the energy of this high energy electron, together with the release of an electron from methane itself, this results in two electrons of lower energy. These low energy electrons are then able to interact with fluoranthene to form the radical anion species that is used during the ETD reaction inside of the ion trap.
Scheme 7. Schematic representation of the creation of the PTR reagent from the radical anion species.
After creation of the radical anion species in the chemical ionization cartridge, another reaction may take place. During this follow up reaction a proton is absorbed by the radical anion which will produce a carbanion. This compound is readily able to donate a proton to a peptide when introduced to the ion trap.
In addition to direct data dependent analyses, ETD – MS/MS analyses were also performed during which ETD was triggered by the detection of CID loss of neutral fragments with a nominal mass of 98 and 80 Da. These measurements used “enhanced resolution” for the primary CID fragmentation and used the average of 5 measurements. The subsequent round of ETD fragmentation upon neutral loss detection utilized “Xtreme resolution”, which required 100 ms of reaction time in the iontrap and used the average of 3 measurements.
3.2.4 Data analysis
The raw MS/MS data was processed with the Data Analysis program version 4.2 (Bruker Daltonics, Bremen Germany). With this program the results were deconvoluted and searched for compounds using the AutoFind (MS/MS) function. Full compound lists were extracted containing the CID and ETD data (if acquired), in order to use in subsequent database searching. The search engine used for the database searching was the Matrix Science server, MASCOT. An in house operated server (version 2.5.1) was used over the available online servers, because this allowed for larger data files to be uploaded. The used settings for the MS/MS Ions Search were dependent on the specific
experiment, but the overall settings used were: MSP2L and BSA analysis:
Database: Swissprot (550,116 reviewed entries, UniProtKB). Digestion: trypsin (1 missed cleavage allowed).
All taxonomy allowed.
Fixed modifications: carbamidomethyl (C).
Variable modifications: oxidation (M); phosphorylation on (ST) and (Y) . Peptide tolerance: ± 0.5 Da.
MS/MS tolerance: ± 0.4 Da.
Peptide charges: 1+, 2+, 3+ (For future analyses; 2+, 3+,4+ is recommended). Error tolerant on.
Instrument: CID: ESI-Trap; ETD: ETD-Trap.
The kit (MSP2L) was analyzed using the swissprot database instead of an in house created phosphopeptide kit database, due to the ability to also detect BSA when this was spiked in. β-casein analysis:
Database: Contaminants (in house).
Digestion: trypsin (1 missed cleavage allowed). All taxonomy allowed.
Fixed modifications: none.
Variable modifications: oxidation (M); phosphorylation on (ST). Peptide tolerance: ± 0.5 Da.
MS/MS tolerance: ± 0.4 Da.
Peptide charges: 1+, 2+, 3+ (For future analyses; 2+, 3+,4+ is recommended). Error tolerant on.
Instrument: CID: ESI-Trap; ETD: ETD-Trap.
When measuring β-casein a contaminants database was used (matrix science, EMBL) since this protein is an often encountered contaminant. The use of the contaminant database is strictly useful when preforming optimization experiments with a compound which is listed in this database (for example β-casein). The use of this database is not recommended when analyzing other samples, because it is very limited in the amount of entries and might produce false positive results.
4. Results and Discussion
1.1 Optimization of enrichment 1.1.1 Different MOAC materials
The approach to set up a standard workflow for the analysis of phosphopeptides combines the use of the amazon Speed / ETD mass spectrometer from Bruker Daltronics 55 and commercially obtainable
MOAC tips. To this end, three different MOAC tips were bought (TopTip) containing three different materials: TiO2, ZrO2 and Zr/TiO2. Due to the ambiguity in literature which material performs best, an
assessment of the obtained materials was needed. The performance of these tips on phosphopeptide enrichment has been tested using a phosphopeptide kit (MSP2L, see supplemental information) and the following conditions:
Loading buffer: 80% ACN, 2% TFA
Washing buffer: 80% ACN, 2% TFA
Elution Buffer: 2M NH4OH, pH 10.5 Acidification: 1:1 (v/v) with 10% TFA
For each round of enrichment, the following samples were taken: control (50% of the starting
mixture), flow through and wash (pooled), and the eluted fraction. The samples were freeze dried and dissolved after enrichment, after which they could be desalted. The desalting step is not absolutely necessary at this stage, since the samples contain no additives or salts and are freeze dried before analysis. However, due to the experimentation with additives later on, the workflow was kept the same for all the samples. Subsequently, these samples were analyzed upon the mass spectrometer and CID data was obtained. The obtained data was processed and presented to a database search using the in-house MASCOT server, from which the following ion scores were collected, see table 1.
Table 1. Ion score and presence of the phosphopeptides from MSP2L after MOAC enrichment using different materials.
MOAC : Sample type 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.1 0 TiO2 Control 49 - 31 59 32 55 10 15 - 68 FT + wash - - - 69 a - - - -Eluted 46 - 31 79 29 61 13 16 - -ZrO2 Control 51 - 28 64 33 65 22 12 - -FT + wash - - - -Eluted 46 - 27 19 - - - -Ti/ZrO 2 Control 51 - 29 60 42 63 15 12 - -FT + wash - - - 32 a - - - -Eluted 44 - 21 51 - 11 - - -
-The MASCOT ion scores of the identified phosphopeptides (MSP2L) after enrichment on different Metal Oxide Affinity Chromatography materials (TiO2 ZrO2, and Ti/ZrO2). For each enrichment experiment three samples were collected; not enriched negative control (Control), a pooled fraction of the flow through and wash fractions (FT + wash), and the eluted fraction (Eluted). Green: The assigned ion score from MASCOT. - : Not observed. a: Only observed with neutral loss of phosphoric acid.
From these results, it becomes evident that more peptides are found in the TiO2 eluted fraction than in
both the eluted fraction from ZrO2 or Zr/TiO2 enrichment. However, the loss of peptide 2.10 in the
titanium enrichment shows, as well as the loss of other peptides from the zirconium- and mixed-enrichment, that the enrichment is prone to sample loss. Additionally, the ion scoring of the peptides is not increased when compared to the control run as was expected, since the phosphopeptide kit was not mixed with any other peptides. This indicates the potential for enrichment, since phosphopeptides can be retained by the materials during washing and eluted from the material with little loss.
After the potential for enrichment with the TopTips was confirmed, enrichment from a complex mixture was performed. This was done in order to simulate a more realistic sample, in which both phosphorylated and nonphosphorylated peptides will be present upon tryptic digestion. In this experiment the MSP2L phosphopeptide kit was mixed 1:50 (molar ratios) with tryptic Bovine Serum Albumin (BSA) digest. Again the three kinds of MOAC tips were tested for enrichment, using the same experimental conditions as used in the previous experiment although with a slight modification to the washing buffer. This change in concentration of acetonitrile in the washing buffer was done because of optimization experiments (which will follow shortly) from which it became evident that this is also a suitable concentration. Besides, this concentration was required in later experiments, due to the need for a more hydrophilic solution when solvating the additives.
Loading buffer: 80% ACN, 2% TFA
Washing buffer: 65% ACN, 2% TFA
Elution Buffer: 2M NH4OH, pH 10.5 Acidification: 1:1 (v/v) with 10% TFA
Again, samples were freeze dried and resuspended in (2% ACN, 0.1% TFA), after which they were desalted using OMIX tips. In the same fashion as before, three kind of samples were taken (control, flow through + wash, and eluted) which were also analyzed the same way. Again, only the CID data was collected during the mass spectrometry analysis, which was used to obtain ion scoring after database searching.
Table 2. Ion score and presence of the phosphopeptides (MSP2L) after enrichment from a complex mixture using different MOAC materials.
MOAC: Sample type 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.1
0 TiO2 Control - - - -FT + wash - - - -Eluted 48 - 14 72 32 28 2 b 3 20 29 ZrO2 Control - - - -FT + wash - - - -Eluted 46 - - 27 - - - -Ti/ZrO2 Control - - - 17 - - - -FT + wash - - - -Eluted 48 - - 46 - - - 14 -
-The obtained MASCOT ion scores of the identified phosphopeptides after enrichment from a complex 1:50 (molar ratios) mixture (containing Phosphopeptides amd BSA digest, respectively). Green: the assigned ion score from MASCOT. - : Not observed. b: Only observed with transfer of the phosphogroup to another position.
From table 2, it directly becomes clear that (almost) no phosphopeptides have been detected in the control and washing samples. However, this is to be expected from a complex sample, since the phosphopeptides will be suppressed during ionization by the overwhelming amount of
nonphosphorylated peptides. Furthermore, during the washing (and loading) we expect no
phosphopeptides to be detected since they should coordinate to the MOAC material. During elution we hope to find all the phosphopeptides back, as is the case with the titanium dioxide material (peptide 2.2 has never been detected, see supplemental information). Sadly, we don’t see the same kind of recovery with the zirconium containing columns. Controversially, enrichment of the complex mixture with the TiO2 column made it possible to detect the 2.10 peptide, which was not detected in
the enrichment using only the MSP2L kit. This could be a result of the change in washing buffer (from 80% ACN to 65% ACN), which would indicate the necessity of appropriate washing
conditions. However, it has to be noted that all of the eluted samples contained peptides originating from BSA, with very high protein scoring (ranging between 2497 – 6193) and large protein
fragmentation coverage (38%-74%). This indicates incomplete enrichment of the phosphopeptides from the mixture, and might suggest additional washing or purification steps are necessary. These results also support the necessity of adding additives to the loading and washing buffers, which would compete with the nonphosphorylated peptides for binding to the MOAC material.
1.1.2 Additives/buffers experimentation
The effect of different additives and buffer conditions on the enrichment of phosphopeptides was tested using the three different MOAC materials as well as two acidic additives; L-glutamic acid and citric acid. These acids, and their concentrations, have been chosen after evaluating literature for appropriate additives and due to their ability to coordinate to the materials, see figure 8. Due to the previous results, only the trials with TiO2 are shown below. The approach of the experiments was very
much similar to the previous experiments in regard to obtaining different fractions (control, flow through + wash, eluted) but with different buffer compositions. Both simple and more complex samples (β-casein tryptic digest, 1:50 molar ratios of MSP2L: BSA tryptic digest) have been analyzed using this workflow.
Figure 8. Molecular structures of the two additives used during the MOAC enrichment.
Both compounds were chosen after study of recent literature as well as consideration of the suitability for the experimental setup. Left: L-glutamic acid, right: citric acid.
Comparison of the enrichment of the phosphopeptides has been performed by comparing ion scores as well as the scoring and coverage for the BSA protein or β-casein. Ideally, the eluted fraction would yield only phosphopeptides identifications with high ion scores and low protein scores for the used nonphosphorylated protein (both BSA and β-casein).
In order to test the effect of the percentage of acetonitrile in the washing buffer on the enrichment of phosphopeptides, an experiment was performed using β-casein tryptic digest, of which the peptides can be seen in the supplemental information. In this test, two different concentrations of acetonitrile were used during the washing steps, 65% and 80%. However, both the samples were subjected in a loading buffer containing 80% ACN. The following buffer compositions have been used in these experiments:
Acetonitrile buffer:
Loading buffer: 80% ACN, 2% TFA
Washing buffer 1: 80% ACN, 2% TFA Washing buffer 2: 65% ACN, 2% TFA Elution Buffer: 2M NH4OH, pH 10.5 Acidification: 1:1 (v/v) with 10% TFA
Before enrichment control samples were collected, during the enrichment both a wash + flow through and eluted sample were collected. Due to mechanical problems with the vacuum dryer, small aliquots (3 μL) of the samples were dried using a vacuum centrifuge (alcohol mode, 35 ˚C, 10min). The dried samples were then dissolved in six microliter of a 2% ACN, 0.1% TFA solution. These samples have been analyzed by using CID and a 30 minute LC gradient, after which protein scores and coverage as well as presence of phosphopeptides of β-casein was determined.
1 2 3 4 5 6 0 250 500 750 1000 1250 1500 1750 0% 25% 50% 75% 100%
Figure 9. Protein scoring and coverage of β-casein after enrichment on titanium dioxide with different concentrations of acetonitrile in the washing buffer.
The filled bars represent the MASCOT protein score (arbitrary units) for β-casein on the left y-axis, the faded bars represent the protein coverage in percentages on the right y-axis. The three left bars indicate the results from samples using a 65% acetonitrile washing buffer; control, wash + flow through and eluted. The three bars on the right indicate the results obtained when washing was performed with a 80% ACN washing buffer; control, wash + flow through and eluted.
Since titanium proved superior to zirconium and the mixed material, only the enrichment of
phosphopeptides on TiO2 material was assessed here. Observing the protein scoring in figure 9 shows
the indication that washing the column with 65% acetonitrile yields similar results as when 80% is used for the eluted sample. A decrease in the scoring and coverage can be seen when comparing the the control and the eluted samples from the 65% acetonitrile washing buffer experiment. This lowering of the protein score could be an indication that less nonphosphorylated peptides are present after washing and eluting the peptides from the MOAC material. However, it is to be noted that both eluted samples from both experiments yield a similar score and coverage after enrichment. Along with the knowledge that in both cases the same sets of phosphopeptides could be found, this gives another strong argument that changing the percentage of acetonitrile during the washing step is not influential.
1.1.3 Additive optimization
After establishing that changing the percentage of acetonitrile in the washing buffer will not benefit the enrichment much, new experiments were performed in which additives were used. According to literature, both L-glutamic acid and citric acid proved to be promising additives for enrichment of phosphopeptides. Therefore the following buffers were prepared for the enrichment of
phosphopeptides from a commercially available kit in a complex mix (MSP2L:BSA digest | 1:50):
L-Glutamic acid buffer:
Loading buffer: 65% ACN, 2% TFA, 0.14M L-glutamic acid
Washing buffer: 65% ACN, 2% TFA
Elution Buffer: 2M NH4OH, pH 10.5 Acidification: 1:1 (v/v) with 10% TFA
Citric acid buffer:
Loading buffer: 65% ACN, 2% TFA, 50 mM Citric acid
Washing buffer: 65% ACN, 2% TFA
Elution Buffer: 2M NH4OH, pH 10.5 Acidification: 1:1 (v/v) with 10% TFA
The obtained samples (control, wash + flow through and eluted) were then analyzed using a 30 minute LC gradient prior to CID fragmentation. After processing of the data, the ion scores of the specific peptides were recorded and are shown below, see table 3.
Table 3. Ion score and presence of the phosphopeptides (MSP2L) after enrichment using different additives to the loading buffer.
Additive Sample type 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10
80% ACN
Control - - - -Wash - - - -Eluted 48 - 14 72 32 28 2 b 3 20 29L-Glutamic
acid
Control - - - -Wash - - - -Eluted 46 - 30 57 33 78 12 11 - 57Citric acid
Control - - - -Wash - - - -Eluted 42 - 26 54 29 51 27 24 - 52Enrichment of phosphopeptides from a complex sample containing (1:50 | Phosphopeptides: BSA digest), performed on titanium dioxide material with different kind of additives in the loading buffer. The three different loading buffers contained either no additive (80% acetonitrile), 0.14 M L-glutamic acid, or 50 mM citric acid. The concentrations of the additives are derived from recent publications. The scores indicate the MASCOT ion scores for the identified phosphopeptides. b: Only observed with transfer of the phosphogroup to another position.
One of the most apparent aspects of these results is the recovery of almost all the phosphopeptides in all of the enrichment experiments (peptide 2.2 has never been detected). Additionally, no
phosphopeptides are detected in control or washing fractions, which is to be expected. The most intriguing difference between the different additives is the reduction in protein score and protein coverage for BSA in the eluted fraction, in which L-glutamic acid has been used as additive to the loading buffer, see figure 10. Whereas both acetonitrile and citric acid retain approximately the same scoring and coverage, the coverage is reduced about 2-fold and the protein scoring seems four times smaller than control or washing fractions. This suggests that the addition of L-glutamic acid (an amino acid) is a useful tool for eliminating unwanted coordination of nonphosphorylated peptides to the MOAC material. However, since there are still nonphosphorylated peptides present in the eluted fraction, additional washing and clean up steps might be needed.
Citric acid has been used in a quantity associated with monophosphorylated peptide enrichment, and might provide different results when higher concentrations are used.
0 2500 5000 7500 0% 20% 40% 60% 80% 100% 0 2500 5000 7500 0% 20% 40% 60% 80% 100% 0 2500 5000 7500 0% 20% 40% 60% 80% 100%
Figure 10. The protein scoring and coverage of Bovine Serum Albumin (BSA) after titanium oxide enrichment with different additives.
For each enrichment experiment three samples were collected; not enriched negative control (Control), a pooled fraction of the flow through and wash fractions (FT + wash), and the eluted fraction (Eluted). Each filled bar represents the protein score (arbitrary units) of BSA on the left y-axis, each fading bar represents the protein coverage in percentages on the right y-axis. From left to right: Enrichment with 80% acetonitrile in the loading buffer; L-Glutamic acid (0.14M) as additive to the loading buffer; Citric acid (50 mM ) additive to loading buffer.
In contrast, the loading buffers containing acetonitrile and citric acid both have increased or similar scoring and coverage of BSA after enrichment on TiO2 material. This indicates that using these
additives is not as effective as using L-glutamic acid at competitively displacing nonphosphorylated peptides from titanium. However, due to limited capacity of the MOAC columns, less
nonphosphorylated peptides could be co-eluting and end up in the enriched sample. This “dilution” of nonphosphorylated peptides could explain why the phosphopeptides are still observed, whereas they could not be detected in the starting mixture, see “control” table 3. Multiple rounds of enrichment with these kinds of additives could perhaps obtain the same level of enrichment obtained by a single round of L-glutamic acid assisted enrichment. Another possibility is to combine multiple additives into one washing buffer, or develop a workflow with multiple different washing steps.
In order to observe the effect of concentration of the additives on enrichment properties, a series of experiments was performed on tryptic digested β-casein with different concentrations of both
L-glutamic acid and citric acid. The benefit of using β-casein over the phosphopeptide kit is that the tryptic digestion of this protein contains both phosphorylated and non-phosphorylated peptides, as can be seen in the supplemental information. Moreover, this specific β-casein (purchased from Sigma Aldrich, St. Louis, MO, USA) also contains κ-casein, α-S1-casein and α-S2-casein. These other casein proteins also contain phosphorylation sites, which makes them eligible for phosphopeptide
enrichment. However, some of the obtained peptides have a long sequence and/or contain multiple phosphorylation sites of up to five sites, see supplemental information. These two facts render some of the possible phosphopeptides unfit for analysis using our specific setup. These large peptides require a charge of 4+ or higher to reduce the m/z ratio within the detection range of our ion trap mass spectrometer. For phosphopeptides, with their negatively charged phosphate groups, higher positive charges are very unfavorable. Moreover, obtaining such a charge requires multiple basic sites along the sequence, which is often not the case. Multiple phosphorylation sites are responsible for more negative charge, which makes it difficult to observe these kinds of peptides during mass spectrometry analysis. Additionally, multiple phosphorylation sites could result in tighter binding to the MOAC material, which could result in loss during elution. Because of these reasons, only the peptides from β-casein and α-S1-β-casein were analyzed during this experiment.
Using L-glutamic acid as an additive, three different concentrations were tested: 35 mM, 70 mM and 140 mM. The last concentration (140 mM) lies closely to the saturation limit in water for this
compound. Although the buffers also contain a high percentage of organic solvent (acetonitrile, 65%) and acid (2% TFA), this saturation limit seems to still be near this point, because the acid did not dissolve readily. Also for the citric acid additive, three different concentrations were used: 50 mM, 0.5M and 1M. These concentrations span the range as is used during Citric Acid-Assisted Two-Step Enrichment (CATSET) protocols.48 The other buffers (washing, eluting, and resuspension buffers)
were kept the same as used before.
The enriched samples were analyzed with CID neutral-loss-triggered ETD mass spectrometry and the protein scoring as well as the protein coverage was determined, see figure 11. Also the presence of the specific phosphopeptides was evaluated, in combination with the charge state of the specific ion; see both tables 4 and 5.
1
2
3
0 250 500 0% 25% 50% 75% 100%1
2
3
0 250 500 0% 25% 50% 75% 100%Figure 11. The protein scoring and coverage of β-casein from titanium oxide enrichment with different concentrations of additives to the loading buffer.
For each enrichment experiment three samples were collected; not enriched negative control (Control), a pooled fraction of the flow through and wash fractions (FT + wash), and the eluted fraction (Eluted). Each filled bar represents the protein scoring on the left y-axis, each fading bar represents the protein coverage in percentages on the right y-axis. Left: Enrichment with L-glutamic acid as additive to the loading buffer (35 mM, 70 mM, 140 mM). Right: Enrichment with citric acid as additive to the loading buffer (50 mM, 0.5 M, 1 M).
Table 4. Identification of specific phosphopeptides from both β-casein and α-S1-casein after L-glutamic acid enhanced MOAC enrichment.
Conc. L-glutamic acid (mM) β-casein α-S1-casein 2060.8 2 2431.0 4 1659.8 4 1926.7 6 1950.9 4 35 CID 2+ - 2+ - 2+ NL-ETD 2+ - 2+ 2+ 2+ 70 CID 2+ 3+ 2+ 2+ 2+ / 3+ NL-ETD 2+ 3+ 2+ 2+ 2+ 140 CID 2+ 2+ / 3+ 2+ 2+ 2+ / 3+ NL-ETD - 2+ / 3+ 2+ 2+ 2+
The observed ions (either 2+ or 3+ ions) from a selection of the phosphopeptides from both β-casein and α-S1-casein after enrichment with different concentrations of L-glutamic acid in the loading buffer. The specific mass of the observed peptides is shown underneath the protein name, see supplemental information for sequence. The enrichment was performed on titanium dioxide columns. Both the CID and Neutral-Loss-triggered ETD results have been listed per concentration.
Table 5. Identification of specific phosphopeptides from both β-casein and α-S1-casein after citric acid enhanced MOAC enrichment.
Conc. citric acid (M) β-casein α-S1-casein 2060.82 2431.04 1659.84 1926.76 1950.94 0.05 CID 2+ 2+ / 3+ 2+ 2+ 2+ / 3+ NL-ETD 2+ 2+ 2+ 2+ 2+ 0.5 CID 2+ 2+ / 3+ 2+ - 2+ NL-ETD 2+ 3+ - - -1 CID 2+ 2+ / 3+ 2+ 2+ 2+ NL-ETD 2+ - - -
-The observed ions from a selection of the phosphopeptides from both β-casein and α-S1-casein after enrichment with different concentrations of citric acid in the loading buffer. The enrichment was performed on titanium dioxide columns. The specific mass of the peptides is stated underneath the protein names, for more sequence see
