Selective sampling of multiply phosphorylated peptides by capillary electrophoresis for electrospray ionization mass spectrometry analysis

Journal of Chromatography A, 1156 (2007) 101–110

Selective sampling of multiply phosphorylated peptides by capillary electrophoresis for electrospray ionization mass spectrometry analysis Jennifer N.M. Ballard a , Gilles A. Lajoie a,∗ , Ken K.-C. Yeung a,b,∗ a

Department of Biochemistry, The University of Western Ontario, London, Ontario, Canada N6A 5C1 b Department of Chemistry, The University of Western Ontario, London, Ontario, Canada N6A 5B7 Available online 26 December 2006

Abstract The ionization of phosphorylated peptides in positive ion mode mass spectrometry is generally less efficient compared with the ionization of their non-phosphorylated counterparts. This can make phosphopeptides much more difficult to detect. One way to enhance the detection of phosphorylated proteins and peptides is by selectively isolating these species. Current approaches of phosphopeptide isolation are based on the favorable interactions of phosphate groups with immobilized metals. While these methods can be effective in the extraction, they can lead to incomplete sample recovery, particularly for the most strongly bound multiply phosphorylated components. A non-sorptive method of phosphopeptide isolation using capillary electrophoresis (CE) was recently reported [Zhang et al., Anal. Chem. 77 (2005) 6078]. The relatively low isoelectric points of phosphopeptides cause them to remain anionic at acidic sample pH. Hence, they can be selectively injected into the capillary by an applied field after the electroosmotic flow (EOF) is suppressed. The technique was previously coupled with matrix-assisted laser desorption/ionization mass spectrometry (MALDIMS). In this work, the exploitation of selective sampling in conjugation with electrospray ionization mass spectrometry (ESI-MS) is presented. The transition was not immediately straightforward. A number of major alterations were necessary for ESI interfacing. These adaptations include the choice of a suitable capillary coating for EOF control and the incorporation of organic solvent for efficient ESI. As expected, selective injection of phosphopeptides greatly enhanced the sensitivity of their detection in ESI-MS, particularly for the multiply phosphorylated species that were traditionally most problematic. Furthermore, an electrophoretic separation subsequent to the selective injection of the phosphopeptides was performed prior to analysis by ESI-MS. This allowed us to resolve the multiply phosphorylated peptides present in the samples, predominantly based on the number of phosphorylation sites on the peptides. © 2006 Elsevier B.V. All rights reserved. Keywords: Phosphorylation; Multiply phosphorylated peptides; Separation; EOF suppression; Polymer capillary coatings; Non-aqueous CE

1. Introduction Protein phosphorylation is an extremely important cellular event that governs the actions of numerous processes, including gene expression, which is mediated through the phosphorylation of transcription factors by serine and threonine kinases in the nucleus [1]. Phosphorylation is carried out by a family of enzymes known as protein kinases, which are encoded by over 500 genes [2]. It is believed that one-third of all proteins present in a mammalian cell are regulated through reversible phosphorylation [3]. Given its importance, great effort has gone into better

∗

Corresponding authors. Tel.: +1 519 661 3074; fax: +1 519 661 3175. E-mail addresses: [email protected] (G.A. Lajoie), [email protected] (K.K.-C. Yeung). 0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2006.12.025

understanding phosphorylation and its connection to disease. Mass spectrometry (MS) is widely used in proteomics and is the current method of choice to detect protein phosphorylation for high sample throughput and sensitivity [4]. Unfortunately, there are several disadvantages facing the MS-based detection of phosphopeptides. Firstly, phosphorylated proteins are generally present at very low stoichiometic ratios relative to their non-phosphorylated counterparts. Secondly, acidic phosphopeptides are inherently more difficult to ionize in positive ion mode MS than less acidic, non-phosphorylated, components. As a result, phosphopeptides are subject to considerable ionization suppression, making detection difficult without first isolating them from the non-phosphorylated peptides. There are numerous techniques available to isolate phosphopeptides, many of which are chromatographic or sorptive methods based on the distinctive interactions of phosphate-containing peptides

102

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

with solid phases. Two of the more recent popular methods are immobilized metal ion affinity chromatography (IMAC) [5] and titanium dioxide (TiO2 ) chromatography [6], in which phosphopeptides are selectively isolated based on their interactions with iron or titanium, respectively. Unfortunately, the retentive characteristic of these columns can lead to sample loss particularly for the most strongly retained multiply phosphorylated peptides [4,6]. Alternatively, the relatively polar character of phosphopeptides has been used for its isolation in reversed phase chromatography, which is biased against peptides with low isoelectric points (pI) [7]. Hence, phosphopeptides generally elute earlier than their non-phosphorylated forms [8]; nevertheless, in a complex mixture, they can co-elute with other hydrophilic nonphosphorylated peptides. Likewise, the non-retentive behavior of phosphopeptides allows their isolation in the flow-through fractions from a reversed phase column [9]. However, this can contain non-volatile salts, buffers, and additives that can be deleterious to MS-based analyses. Therefore, on-target cleanup [9] or further chromatographic resolution with porous graphitic carbon chromatography [10] was necessary. A more effective, non-retentive, approach of phosphopeptide isolation has recently been developed in our laboratory using capillary electrophoresis (CE) [11]. This technique takes advantage of the acidic characteristics of phosphopeptides, which remain negatively charged even in low pH environments when most non-phosphorylated peptides have a net positive charge. When electrokinetic injection is performed without a significant electroosmotic flow (EOF), the migration of sample ions into the capillary is predominantly determined by their net charges. This allows for the selective sampling of anionic peptides, such as the phosphopeptides from a tryptic digest of phosphorylated proteins [11]. A similar approach was applied to analyze selectively sialylated glycoproteins by CE [12]. Compared with the conventional injections used in CE, selective sampling permits focusing on a sub-set of the sample and greatly reduces the complexity for subsequent separation and detection. Selective sampling of phosphorylated peptides by CE is particularly effective for isolation of multiply phosphorylated peptides. Since these species have extremely low pIs (pI < 3), they can be better distinguished from the acidic nonphosphorylated peptides, such as those rich in Asp and Glu residues. In addition, multiply phosphorylated peptides generally have higher net negative charges than singly phosphorylated peptides and are consequently injected to a greater extent during selective sampling [11]. The sample bias towards multiply phosphorylated peptides is particularly advantageous as it is complementary to the commonly used IMAC and TiO2 , both of which tend to give poor recovery of multiply phosphorylated peptides [4]. Other advantages realized by using CE include improved sample recovery since there is no binding to a stationary phase, and CE’s ability to handle extremely small sample volumes. Our previous reports of selective phosphopeptide sampling focused on the interfacing with matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS) [11,13]. In the current report, we are interested in adapting the selective sampling of phosphopeptides for detection by electrospray

ionization mass spectrometry (ESI-MS). The transition from MALDI-MS to ESI-MS for the selective injection of phosphopeptides involves a number of non-trivial alterations. It begins with the re-assessment of capillary coatings for compatibility with ESI-MS. Previously, a phospholipid coating was employed to generate the required near-zero EOF for selective sampling at acidic pH conditions. While this coating appeared to be compatible with MALDI-MS, it was nevertheless a non-covalently adsorbed coating. Desorption of coating could impose issues with ESI-MS. Instead, permanent, covalently attached polymeric coatings may be more suitable for this work. Over the years, a large collection of capillary coatings, both covalently bound and physically adsorbed, have been reported in the literature for the prevention of peptide or protein adsorption during their separations in CE. This area of development was last reviewed by Doherty et al. in 2003 [14] and briefly summarized by Huang et al. in 2006 [15]. Examples of covalently attached coatings include polyacrylamide (PA) [16], poly(vinyl alcohol) (PVA) [17], poly(vinylpyrrolidone) (PVP) [18], poly(ethylene oxide) (PEO) [18] and poly(ethylene glycol) (PEG) [19]. Some of these polymers, namely PVA [17] and PEO [20], and other neutral polymers such as hydroxypropylmethyl cellulose [21], were also used as dynamic adsorbed coatings. Other dynamic coatings included charged polymers (polybrene [22], polyarginine [23], polyethyleneimine [24], chitosan [25]), amines [26] and surfactants [27–29] including the phospholipids used in our previous work [11,13,30]. Capillary coatings previously employed for CE–MS were summarized in a review by Simpson and Smith [31]. Many of the aforementioned coatings were included. While no one technique appeared dominant, preference was given to charged coatings that could generate a significant EOF, since such a solution flow during electrophoresis is beneficial in maintaining a stable electrospray. This is illustrated by developments in CE–MS coatings within the last year, which predominantly focus on charged coatings: polyamine [32], polybrene-poly(vinyl sulfonate) bilayer [33], and ionene [34]. Unfortunately, in contrast to the traditional CE separations, the selective injection of phosphopeptides cannot be performed with a fast EOF. We will, therefore, focus on the more conventional uncharged polymeric coatings for this work. This brings us to the second focus of this work, which is to determine the best conditions to deliver the selectively injected phosphopeptides from the capillary to the ESI-MS under a nearzero EOF. The selection of phosphopeptides based on their electromigration dictates the use of capillaries with surface modifications that result in near-zero EOFs. While the injection step can be performed offline, decoupled from the detection step by ESI-MS, it is unfortunately not possible to transfer the analytes from one capillary to another without introducing significant bandbroadening. Hence, ESI-MS can only be performed in the absence of an EOF in this case. To create a flow, a make-up solution can be added near the ESI emitter using either a T-junction connector or a sheath-flow interface [35]. Unfortunately, the introduction of makeup flow can dilute the analytes and lower the sensitivity. In addition, the pressure used to deliver the make-up solution can potentially drive the capillary contents

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

backwards. Alternately, we choose to apply pressure directly to the sampling capillary to mobilize the contents to the ESI-MS. While band broadening from laminar flow is expected from such a pressure-assisted setup, high separation efficiency is less critical when the primary focus of this work is the selection or isolation of phosphopeptides. Instead, we aim to maintain good sensitivity, while achieving moderate electrophoretic resolution to separate the selectively injected components prior to ESI-MS. A final consideration in obtaining optimal operating conditions for selective injection CE–ESI-MS is the use of organic solvents, which is common in ESI-MS for assisting solvent evaporation and sample ionization. Our approach is to incorporate organic solvents, such as methanol and acetonitrile, into the sampling and separation buffers, anticipating that the presence of the solvents would not impose major negative effects on the integrity of the capillary coating, the electroosmotic flow, the electrophoretic mobilities of the peptides, and, ultimately, the selective injection of peptides. If successful, this approach will facilitate transfer of selectively sampled peptides, along with the appropriate concentration of organic solvent, to the ESI-MS source via direct infusion. 2. Experimental 2.1. Apparatus All CE experiments were performed on an Agilent 3D CE instrument (Palo Alto, CA, USA) using fused silica capillaries (50 ␮m I.D. × 360 ␮m O.D.) from Polymicro Technologies (Phoenix, AZ, USA) with a total length of 48.5 cm. Capillaries were maintained at 25 ◦ C for all experiments. An on-capillary UV-absorbance detector was used in some experiments to monitor the injection of peptides and to measure the EOF. The EOF measurements were performed using the three-injection method introduced by Williams and Vigh [36] using mesityl oxide as a neutral marker. Mesityl oxide, 20 mM, was introduced into the capillary at 20 mbar for 5 s and monitored at 254 nm. Mass spectral data was collected using a Q-TOF mass spectrometer (Global, Micromass/Waters, Manchester, UK) by ESI in the positive ion mode. Capillaries containing selectively sampled peptides were removed from the CE instrument, and one end of the capillary was then fitted to the nanoemitter of the ESI mass spectrometer via a micro-tight capillary LC fitting (Upchurch Scientific, Oak Harbor, WA, USA). The other end of the capillary was connected to a syringe filled with running buffer. Capillary contents were infused to the ESI-MS with the onboard syringe pump of the mass spectrometer. 2.2. Reagents HPLC-grade acetonitrile and water were purchased from Fisher Scientific Inc. (Hampton, NH, USA). Methanol, sodium hydroxide, formic acid, and acetic acid were purchased from Merck KGaA (Darmstadt, Germany). Ammonium hydroxide, calcium chloride, ammonium hydrogencarbonate, mesityl oxide, adrenocorticotropic hormone (ACTH) fragment 18–39, neurotensin, leucine enkephaline, ammonium persulphate

103

(APS), N,N,N ,N -tetramethylethylenediamine (TEMED), N,N methylenebisacrylamide (bis-acrylamide), acrylamide, and 3-(trimethoxysilyl)propyl methacrylate were received from Sigma-Aldrich (St. Louis, MO, USA). Bovine ␣-casein (Sigma– Aldrich) was digested with modified sequence grade trypsin (Promega Corp., Madison, WI, USA). Digests were performed in 50 mM ammonium bicarbonate at pH 8 with 20 mM calcium chloride and at a 1:100 (w/w) enzyme to protein ratio. Digestions were allowed to proceed overnight at 37 ◦ C and the digest solutions were vacuum concentrated to dryness. The peptides were then redissolved in water to a concentration of 100 pmol/␮L and stored at −20 ◦ C until use. All reagents were used as received without further purification. 2.3. Capillary modification for selective sampling The capillaries used in this work were coated with polyacrylamide based on a protocol first introduced by Hjert´en in 1985 [16] and later modified by Gao and Liu [37]. In this method, the capillary was pretreated by rinsing with a 1 M NaOH solution for 45 min, followed by water for 15 min, and finally acetonitrile for 15 min. The capillary was then dried with argon gas or air. Next, bi-functionalization of the pretreated capillary surface with 3-(trimethoxysilyl)propyl methacrylate was performed by flushing the capillary with a solution containing 0.4% (v/v) 3(trimethoxysilyl)propyl methacrylate and 0.2% (v/v) acetic acid in acetonitrile at a rate of 50 ␮L/min for 1 h using a syringe pump. This step was followed by rinsing with acetonitrile and drying with air. Polymerization of the acrylamide was carried out in the absence of oxygen as described by Gao and Liu [37]. This reaction was done inside a sealed vessel kept under a continuous stream of argon at 20 psi. After optimization, the polymerization solution contained 2% (w/v) acrylamide and 0.004% bis-acrylamide. The reaction was initiated by the addition of 1 ␮L 10% ammonium persulphate solution and 1 ␮L of TEMED. The 2 mL reaction mixture was immediately flushed through the capillary for 8 min at 20 psi. Finally, the crosslinked polyacrylamide coated capillary was flushed with water for 2 min, and was stored filled with water until use. 2.4. Selective sampling of peptides Selective sampling of peptides was performed at either pH 3.0 or 5.25. For the pH 3.0 experiments, an ammonium formate solution was prepared by adding dilute ammonium hydroxide to a formic acid solution until the desired pH was attained. The solution was then diluted with the appropriate amount of methanol to achieve a final formate concentration of 20 mM. For the pH 5.25 experiments, an ammonium acetate solution was prepared in the same manner as above, but by adding ammonium hydroxide to acetic acid. Prior to each run, the cross-linked polyacrylamide capillary was rinsed with water for 2 min and then the running buffer for 2 min by applying 1 bar of pressure. The peptide sample, prepared in the running buffer, was placed in a vial at the cathodic end of the capillary, while another vial

104

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

containing only buffer was place at the anodic end. To perform selective injection, −20 kV was applied across the capillary for varying intervals of time. When subsequent electrophoretic separation of the selectively injected peptides was desired, additional voltage (−20 kV) was applied for up to 18 min, after replacing the sample vial with another buffer vial at the cathode. 2.5. Detection of selectively sampled peptides by ESI-MS Capillary contents were pushed at a rate of 0.5 ␮L/min through the capillary with an applied ESI voltage of 4.0 kV. Ions in the m/z range of 400–1800 were observed in each mass spectrum. The cone voltage was set at 60 V and the collision cell voltage at 10 V during MS scans. Peptide identity was confirmed by MS/MS in a data-dependent acquisition (DDA) mode as described previously [38]. Peak assignments were made by using MassLynx (Micromasss/Waters, Manchester, UK) and were considered valid if were within ±0.1 Da of the theoretical value. Tandem MS data was processed with MassLynx “peptideauto” using default settings. 3. Results and discussion 3.1. EOF suppression for selective sampling and ESI-MS Selective sampling based on the net charges of the sample molecules is only possible when electrokinetic injection is performed in the absence of any significant EOF. While the EOF of an untreated silica capillary is relatively low at acidic pH conditions, the desirable EOF for selective injection (<5 × 10−5 cm2 /V s) is only achievable at very acidic pH (e.g., pH <4). The use of this technique only at extremely low pH greatly limits its versatility. Hence, capillary modification is used to suppress the EOF, and in addition, to prevent any peptide adsorption on the capillary wall. In our earlier reports, double-chained zwitterionic phospholipids, 1,2-dilauroyl-sn-phosphatidylcholine (DLPC) and 1,2-dimyristoyl-sn-phosphatidylcholine (DMPC), were used as semi-permanent capillary modifiers to control the EOF and to prevent wall adsorption of peptides to the capillary walls [11,30]. A suppressed EOF of <5 × 10−5 cm2 /V s was obtained between pH 4 and 7 with the modified capillaries. This facilitated the selective injection of standard peptides, which generally exhibited effective mobilities of roughly 1 × 10−4 cm2 /V s. The DLPC or DMPC coatings were also found to be compatible with MALDI mass spectrometry. While signals corresponding to the [M + H]+ of these phospholipids was detected in the MALDI mass spectra, significant effects on the peptide signals were not observed, and low attomole-level detection of peptides was demonstrated [11,30]. We also examined the compatibility of DLPC coatings with ESI mass spectrometry. A test peptide solution of 0.1 mg/mL ACTH, prepared in ammonium acetate at pH 5.25, was infused to the nano-ESI-MS at 1 ␮L/min via a DLPC-coated capillary. The DLPC signal at m/z 622.4 dominated the ESI mass spectra, as the signal of ACTH was greatly reduced to less than 10% in comparison with signal obtained with an uncoated capillary

(mass spectrum not shown). The effect of DLPC appeared to be less significant when higher concentrations (e.g., 0.5 mg/mL) of ACTH were used. However, it was clear that the use of this phospholipid coating would deleteriously affect the sensitivity of the phosphopeptide analyses. Hence, covalently bonded permanent capillary coatings were considered as an alternative. A large number of permanent capillary coatings based on polymers have been reported in the literature [14,39]. Although most coatings were designed primarily to prevent analyte adsorption onto silica, many polymers possess neutral net charges and were, therefore, potentially useful in generating a suppressed EOF. We were particularly interested in the crosslinked polyacrylamide coating reported by Gao and Liu [37]. They demonstrated the operation of their coating at acidic conditions as low as pH 3.25, while others [39] mainly focused their evaluations at neutral or weakly alkaline conditions where analyte adsorption and coating stability were most critical. In this work, selective sampling of phosphorylated peptides required the use of a highly acidic environment to avoid the introduction of low pI non-phosphorylated peptides. Hence, the EOF from a cross-linked polyacrylamide coated capillary was determined at pH 3.0 and pH 5.25. Under both pH conditions, a very low EOF with mobility ranging from 10−6 to 10−5 cm2 /V s was measured. It should be noted that the suppression of EOF was not possible at such a low pH using DLPC or DMPC. While these semi-permanent coatings generated low EOFs at neutral and weakly alkaline conditions, a significant reversed, anodic, EOF of over 10−4 cm2 /V s was observed below pH 4 [11]. Thus, the polyacrylamide coating appeared to be superior in reducing the EOF at low pH conditions. 3.2. Selective sampling in the presence of organic solvents An additional benefit of using permanently coated capillaries for selective injection is their compatibility with organic solvents, whose presence enhances the desolvation and ionization of the sample in subsequent ESI-MS analyses. An organic solvent content of 20–50% in peptide samples is typical for ESIMS. In this work, the incorporation of organic solvent directly into the sample and the running buffer is proposed, as it eliminates the need of additional hardware (e.g., flow splitters, flow sheaths, etc.) for solvent introduction. As selective injection was always performed in an aqueous environment in previous work, the focus of this experiment is to determine the effect of organic solvents on the electroosmotic mobility and the peptide mobilities. The effect of organic solvents was previously studied in non-aqueous CE by Schwer and Kenndler [40]. Organic solvents were found to affect all three parameters that govern the electroosmotic mobility: μeo = εζ/η, where ε is the dielectric constant, η is the viscosity and ζ is the zeta potential. Generally, a decrease in dielectric constant was observed with increasing content of organic solvent in a binary system. The viscosity could increase or decrease depending on the organic solvent. For example, viscosity was found to decrease upon addition of acetonitrile; however, the viscosity was increased by up to 50% methanol, after which point the viscosity began to decrease upon

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

the introduction of additional methanol. Finally, the zeta potentially was generally lowered by the presence of organic solvent in untreated silica capillaries. The effect on the zeta potential in coated capillaries was, however, less straightforward [41]. An experiment was conducted to determine the EOF of the cross-linked polyacrylamide coated capillary in the presence of 10–50% (by volume) acetonitrile or methanol. For both solvents, the μeo remained low, 10−8 to 10−6 cm2 /V s, from 10–30%, but increased to 10−5 to 10−4 cm2 /V s at 50%. While the cause of the EOF increase by the addition of organic solvents in crosslinked polyacrylamide coating capillaries is unclear, it is suffice to say that lower organic solvent contents appear to provide a more suitable EOF for selective injection. Next, the electrokinetic injection and separation of peptides was tested in the presence of methanol and acetonitrile. The 25% organic solvent content was chosen to maintain a low EOF. Neurotensin (pI 8.7) and leucine enkephalin (pI 5.8) were electrokinetically injected in a pH 5.25 acetate buffer for 5 min at +20 kV. Following the injection, the voltage application was resumed to induce electrophoresis. The electropherogram obtained in 25% methanol showed resolution closely resembling what was obtained in pure aqueous buffers, whereas significant deterioration in resolution was observed in 25% acetonitrile. This finding agreed with previous reports of methanol producing better peak efficiency [42]. It is noteworthy that the peptide mobilities appeared slightly slower in the presence of methanol and acetonitrile, most likely due to the increased viscosity, from 0% to 25% organic content. For example, the effective mobility

105

of ACTH was slowed from 1.2 × 10−4 cm2 /V s at 0% methanol to 8 × 10−5 cm2 /V s at 25% methanol in 20 mM ammonium formate buffer at pH 3. Nevertheless, the effect of organic solvents on peptide mobility is universal [43], and thus we do not anticipate any unexpected results in selective sampling, other than the extended injection times required to mobilize the peptide molecules into the capillary. Peptide samples containing various concentrations of methanol were independently analyzed by ESI-MS. It confirmed that 25% methanol resulted in a significant improvement in ESIMS compared with pure aqueous samples. Higher contents of methanol did not result in significant additional enhancement to counterbalance the negative effect of the higher EOF.

3.3. Selective sampling of phosphopeptides from tryptic digests of α-casein Selective sampling on a tryptic digest of ␣-casein at 10 pmol/␮L was performed using a cross-linked polyacrylamide coated capillary and sampling/running buffers containing 25% methanol as depicted in Fig. 1. ␣-Casein is a commonly used phosphorylated protein standard. The commercial source of ␣casein (Sigma) contains two variants, S1 and S2. Up to 10 and 12 phosphorylation sites were reported, respectively for these variants according to the NCBI database [44]. When completely digested with trypsin, peptides with one to five phosphorylation sites are generated (Table 1). Selective sampling was performed

Fig. 1. An illustration of selective sampling of phophorylated peptides at acidic pH conditions. Phosphopeptides have relatively low pIs so they remain as anions and are selectively introduced into the capillary under a near-zero EOF. Peptides with higher net negative charges, which are most likely multiply phosphorylated, are injected to a greater extent. Most non-phosphorylated peptides have relatively high pIs and, therefore, are cationically charged and not injected.

106

Table 1 List of relative ESI-MS signal intensities of peptides from a tryptic digest of ␣-casein treated with various procedures Number of phosphate groups

pI

[M + nH]n+ , (m/z)

Untreated tryptic digest

Peptides selectively sampled at pH 5.0

Peptides selectively sampled at pH 3.0

QMEAEpSIpSpSpSEEIVPNpSVEAQK (␣-S1-(74–94)) NANEEEYpSIGpSpSpSEEpSAEVATEEVK (␣-S2-(61–85)) NANEEEYSIGpSpSpSEEpSAEVATEEVK (␣-S2-(61–85)) NTMEHVpSpSpSEESIIpSQETYK (␣-S2-(17–36)) DIGpSEpSTEDQAMEDIK (␣-S1-(58–73)) EQLpSTpSEENSK (␣-S2-(141–151)) TVDMEpSTEVFTK (␣-S2-(153–164)) QFYQLDAYPSGAWYYVPLGTQYTDAPSFSDIPNPIGSENSEK (␣-S1-(167–208)) VPQLEIVPNpSAEER (␣-S1-(121–134)) VNELpSK (␣-S1-(52–57)) EDVPSER (␣-S1-(99–105)) ITVDDK (␣-S2-(86–91)) LTEEEK (␣-S2-(168–173)) EPMIGVNQELAYFYPELFR (␣-S1-(148–166)) HQGLPQEVLNENLLR (␣-S1-(23–37)) TTMPLW (␣-S1-(209–214)) FPQYLQYLYQGPIVLNPWDQVK (␣-S2-(107–128)) ENLCSTFCK (␣-S2-(48–56)) FFVAPFPEVRGK (␣-S1-(38–49)) YLGYLEQLLR (␣-S1-(106–115)) QEK (␣-S2-(37–39)) EVVR (␣-S2-(57–60)) ALNEINQFYQK (␣-S2-(96–106))

5 5 4 4 2 2 1 0

0.72 0.72 0.90 1.37 2.16 2.28 3.52 3.60

907.6, +3 1030.0, +3 1003.4, +3 873.6, +3 964.4, +2 706.3, +2 733.8, +2 1573.7, +3

+ − − − ++ − ++ +

+ − − − ++ − + +

++ ++ ++ ++ ++ ++ − −

1 1 0 0 0 0 0 0 0 0 0 0 0 0 0

3.62 3.94 4.09 4.10 4.20 4.20 5.60 5.70 6.10 6.30 6.40 6.40 6.40 6.40 6.40

830.9, +2 769.4, +1 831.4, +1 690.4, +1 748.4, +1 772.7, +3 880.5, +2 748.4, +1 903.8, +3 1044.5, +1 1384.7, +1 1267.7, +1 404.2, +1 502.3, +1 684.4, +2

++ − ++ + ++ ++ ++ ++ ++ − ++ ++ + ++ ++

++ − ++ ++ ++ + − ++ − − − − − − −

− − − − − − − − − − − − − − −

Peptides with pIs above 7.0 were omitted for simplicity. Peptides are denoted with: (−) when not observed, (+) when observed under 1000 ion counts, and (++) when observed above 1000 ion counts. The concentration of sample was 10 pmol/␮L, other conditions are as described in the experimental section.

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

Peptide sequence

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

Fig. 2. Total ion signal from the infusion of selective sampled peptides from a tryptic digest of ␣-casein. Conditons: buffer, 20 mM pH 5.25 ammonium acetate in 25% methanol; sample injection, 10 min at −20 kV; infusion flow rate, 0.5 ␮L/min; total ion signal range: m/z 400–1800.

at −20 kV for an extended period of 20 min to fill most of the capillary with the selected phosphopeptides. A sampling pH of 5.25 was initially attempted in order to include all of the phosphopeptides in the injection. Identification of the injected peptides was conducted by infusing the selectively injected peptides from the capillary into the ESI-MS. The total ion signal (m/z 400–1800) obtained during infusion of the selectively sampled components in Fig. 2 clearly confirmed the injection of peptides even under the previously determined near-zero EOF. At a flow rate of 0.5 ␮L/min, the signal duration is approximately 3 min. This corresponds to a quantity slightly larger than one capillary volume of approximately 1 ␮L. The increased length of the total ion signal is likely explained by band broadening due to the laminar flow and the dead volume in the union connector to the ESI emitter. Nevertheless, this confirms that most of the capillary was filled with injected peptides. Most importantly, the peptides identified from the selectively sampled fraction are noted in Table 1. For comparison, the peptides identified from the original digest without any selective sampling are also shown here. This clearly illustrates the ionization suppression effect of the high pI peptides on the low pI peptides. It should be noted that peptides with pI above 7 were also detected in the untreated sample, but were omitted from the table as none of them were phosphorylated. A number of observations regarding the selective sampling experiments conducted at pH 5.25 can be made from the data in Table 1. Compared with the results from the original digest, it is apparent that peptides with pI above the sampling pH of 5.25 were not observed, which confirmed their exclusion by selective sampling. Signals from the acidic non-phosphorylated peptides were observed along with the phosphopeptides, which validated that the peptide selection was strictly based on pI. However, most of the multiply phosphorylated peptides were still not detected with significant intensities, suggesting further isolation from the higher pI peptides was needed to overcome the ionization suppression.

107

In order to enhance the selective sampling of the multiply phosphorylated peptides, the sampling pH was lowered from 5.25 to 3.0. This should facilitate the injection of only peptides with two or more phosphorylation sites with pI from 0.72 to 2.28: ␣-S1-74–94, ␣-S2-61–85 (with five phosphorylations and with four phosphorylations), ␣-S2-17–36, ␣-S1-58–73, ␣-S2141–151. Once again, after selective sampling was performed for 20 min, the capillary was reconnected with the syringe pump and the ESI emitter for MS analysis. In this experiment, the ESI emitter was connected to the capillary inlet where sampling was previously conducted (Fig. 1). In other words, the capillary contents were infused backward into the MS with respect to the initial peptide migration direction under selective sampling. In this case, signals from all of the previously mentioned six phosphopeptides were detected (Table 1), illustrating the effectiveness of selective sampling for phosphopeptide analysis. The extracted ion signals and the resulting mass spectra are shown in Fig. 3. The extracted ion signal onsets occurred at approximately the same time in all cases, with the delay most likely due to the initial period needed for the syringe pump to build up pressure and the elution of the volume in the ESI emitter capillary. Importantly, a significant increase in signal intensity was observed for all six phosphopeptides when analyzed at pH 3 compared with pH 5.25. The signal duration was observed to increase with decreasing pI of the peptides. Peptides with lower pIs potentially exhibited higher mobilities, and thus, were injected further into the capillary than the high pI peptides during the same period of voltage application. This preferential injection of the low pI, highly phosphorylated, peptides is extremely advantageous as these peptides typically ionize poorly in positive ion mode MS. Such sample bias towards the most highly phosphorylated peptides is unique and complementary to existing chromatographic techniques such as IMAC and TiO2 , which both report the most difficulty with recovery of multiply phosphorylated peptides [4]. Another interesting result from the selective injection technique was the resulting “partial separation” of the injected peptides. While the signal onsets of each phosphopeptide overlapped with one another, the tail regions were free from the signals of the peptides with higher pIs. This turned out to be advantageous as the effect of ionization suppression from higher pI peptides was generally more significant. Indeed, such tail regions (marked in the extracted ion signals, Fig. 3) were primarily used to obtain the mass spectra shown. Interestingly, a significant amount of sodium- and potassiumadducts were observed in the mass spectra, particularly for the multiply phosphorylated peptides. The source of sodium and potassium ions in our experiment was unknown. While selective sampling disfavored the injection of sodium and potassium ions, they could still be introduced as adducts with the peptides, as this favorable interaction between salt cations and phosphate groups has been previously reported [45]. In addition, salt contamination may have originated from the instruments, CE and/or MS, as well as the reagents and solvents used. Solvents and reagents of highest purity available commercially were used. Special care was exercised in rinsing all glassware and Teflon bottles with deionized water to minimize the presence of salts.

108

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

Fig. 3. Extracted ion signal and mass spectra from the infusion of selectively sampled phosphopeptides: (A) ␣-S1-74–94 (m/z 907.6), (B) ␣-S2-61–85 (m/z 1030.0), (C) ␣-S2-61–85 (m/z 1003.3), (D) ␣-S2-17–36 (m/z 873.6), (E) ␣-S1-58–73 (m/z 964.3), and (F) ␣-S2-141–151 (m/z 706.3). The labeled masses refer to the monoiosotopic peaks. Marked regions were used to generate MS spectra (|——|). Conditions: buffer, 20 mM pH 3.0 ammonium formate in 25% methanol; sample, tryptic digest of 10 pmol/␮L ␣-casein prepared in buffer; sample injection, 20 min at −20 kV; infusion flow rate, 0.5 ␮L/min.

3.4. Selective sampling and subsequent separation Even though the preferential injection of selective sampling resulted in a partial separation of the phosphorylated peptides, it is still advantageous to perform a more comprehensive electrophoretic separation following selective sampling. This would be especially important for a more complex mixture of peptides. To accommodate such a separation, the selective sampling step was performed as described previously at pH 3.0 with 25% methanol, with the exception of a shortened sampling time of 10 min. Following injection, running buffer was placed at the inlet and electrophoresis was performed for a further 18 min using a voltage of −20 kV. Finally, the selectively sampled and separated peptides were infused to ESI-MS as above (reverse in relation to the initial peptide migration direction, Fig. 1). The extracted ion signals and mass spectra of the six intentionally injected phosphopeptides from ␣-casein were shown in Fig. 4. The signals of these peptides, ␣-S1-74–94, ␣-S2-61–85

(with five phosphorylations and with four phosphorylations), ␣S2-17–36, ␣-S1-58–73, ␣-S2-141–151, were sufficiently strong to yield DDA tandem mass spectra for sequence confirmation (data not shown). MS/MS spectra were confirmed using PEAKS [46] and Mascot [47]. A database search using Mascot showed the presence of only the expected phosphopeptides. No other peptides were matched. The peak confirmation also validated the earlier hypothesis that the elution order and therefore, the peptides’ electrophoretic mobilities during selective sampling, were largely determined by their pIs. The multiply phosphorylated peptides ␣-S1-74–94 (five phosphates), ␣-S2-17–36 (four phosphates), and ␣-S2-61–85 (with either four or five phosphates) migrated faster than the doubly phosphorylated peptides (␣-S1-58–73, ␣-S2-141–151), even though the multiply phosphorylated peptides are almost double the mass of the doubly phosphorylated peptides. It is noteworthy that the elution order of increasing pI, or phosphorylation number, can be reversed simply by connecting the ESI emitter with the other end of the

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

109

Fig. 4. Extracted ion signal and mass spectra of selectively sampled and separated phosphopeptides: (A) ␣-S1-74–94 (m/z 907.6), (B) ␣-S2-61–85 (m/z 1030.0), (C) ␣-S2-61–85 (m/z 1003.3), (D) ␣-S2-17–36 (m/z 873.6), (E) ␣-S1-58–73 (m/z 964.3), and (F) ␣-S2-141–151 (m/z 706.3). The labeled masses refer to the monoiosotopic peaks. Conditions: sample injection, 10 min at −20 kV; separation, 18 min at −20 kV; otherwise same as in Fig. 3.

sampling capillary. This can be useful in shortening the analysis time when one is more interested in the highly phosphorylated species. Temporal separation of the various components was clearly observed. This demonstrates the method’s potential to reduce the sample complexity and promote the quality of MS and MS/MS data. The resolution observed in Fig. 4 was slightly lower than expected from capillary liquid chromatography or traditional CE. This is mainly because the injection volume was increased to load a greater volume of sample to enhance sensitivity. However, one can choose to focus on maximizing either resolution or sensitivity depending upon the specific sample. 4. Conclusions We demonstrated that cross-linked polyacrylamide is an optimal capillary coating for selective injection of phosphopeptides by CE. It provided a very low EOF at highly acidic conditions, and it was compatible with the use of methanol in the sampling

and running buffers. As noted in our previous work on selective sampling, the choice of sampling pH played a very important role in the selective injection of the phosphopeptides that were examined. With the significant effect of ionization suppression observed in ESI-MS, a higher sampling pH led to the detection of mainly singly phosphorylated peptides, while an acidic sampling pH was required to detect the multiply phosphorylated peptides. Selective sampling differs from conventional CE in two ways. First, it screens for a sub-set of the total sample, and reduces sample complexity of the subsequent analysis. Second, it facilitates injection of significantly large sample quantities compared with most generic separation techniques, which is important for enhancing the sensitivity of any subsequent MS analyses. Interfacing with ESI was performed off-line to the selective sampling step by infusing the capillary contents after reconnection to the ESI emitter. This has proven to be a very simple, yet effective means of coupling. Traditional on-line CE interfaces were less suitable in this case because the lack of solution flow, including EOF, is essential to the selective sampling. Given the

110

J.N.M. Ballard et al. / J. Chromatogr. A 1156 (2007) 101–110

moderate resolution associated with selective sampling, infusion is sufficiently effective in transferring the sample to the MS. The most attractive aspect of selective sampling is the sample bias towards multiply phosphorylated peptides which is complementary to that of the commonly used IMAC and TiO2 in phosphorylation analysis. Although this selection bias is based on pI and is not specific to phosphorylation, we have achieved the isolation of the multiply phosphorylated peptides for subsequent analysis. Even though this technique may be biased against selecting basic multiply phosphorylated peptides or not rejecting acidic non-phosphorylated peptides, selective sampling by CE offers an attractive alterative to IMAC and TiO2 chromatography. Acknowledgements This work is supported by the Natural Science and Engineering Research Council of Canada (NSERC), the Ontario Research and Development Challenge Fund (ORDCF), and the University of Western Ontario. The authors thank Dr. Haixia Zhang, Mr. Sean C. Bendall, Dr. J. Larry Campbell, and Ms. Paula P. Pittock for their useful discussions. References [1] R.E. Schweppe, C.E. Haydon, T.S. Lewis, K.A. Resing, N.G. Ahn, Acc. Chem. Res. 36 (2003) 453. [2] G. Manning, D.B. Whyte, R. Martinez, T. Hunter, S. Sudarsanam, Science 298 (2002) 1912. [3] M.J. Hubbard, P. Cohen, Trends Biochem. Sci. 18 (1993) 172. [4] M. Mann, S.E. Ong, M. Gronborg, H. Steen, O.N. Jensen, A. Pandey, Trends Biotechnol. 20 (2002) 261. [5] L. Andersson, J. Porath, Anal. Biochem. 154 (1986) 250. [6] M.W. Pinkse, P.M. Uitto, M.J. Hilhorst, B. Ooms, A.J. Heck, Anal. Chem. 76 (2004) 3935. [7] M.P. Washburn, D. Wolters, J.R. Yates 3rd, Nat. Biotechnol. 19 (2001) 242. [8] H. Ohguro, K. Palczewski, FEBS Lett. 368 (1995) 452. [9] P. Kapkov´a, E. Lattov´a, H. Perreault, Anal. Chem. 78 (2006) 7027. [10] P.O. Vacratsis, B.S. Phinney, D.A. Gage, K.A. Gallo, Biochemistry 41 (2002) 5613. [11] H. Zhang, C. Zhang, G.A. Lajoie, K.K.-C. Yeung, Anal. Chem. 77 (2005) 6078. [12] K. Kakehi, M. Kinoshita, D. Kawakami, J. Tanaka, K. Sei, K. Endo, Y. Oda, M. Iwaki, T. Masuko, Anal. Chem. 73 (2001) 2640. [13] H. Zhang, G.K. Hunter, H.A. Goldberg, G.A. Lajoie, K.K.-C. Yeung, Anal. Chim. Acta 581 (2007) 268.

[14] E.A. Doherty, R.J. Meagher, M.N. Albarghouthi, A.E. Barron, Electrophoresis 24 (2003) 34. [15] Y.-F. Huang, C.-C. Huang, C.-C. Hu, H.-T. Chang, Electrophoresis 27 (2006) 3503. [16] S. Hjert´en, J. Chromatogr. 347 (1985) 191. [17] M. Gilges, M.H. Kleemiss, G. Schomburg, Anal. Chem. 66 (1994) 2038. [18] K. Srinivasan, G. Pohl, N. Avdalovic, Anal. Chem. 69 (1997) 2798. [19] G.J.M. Bruin, J.P. Chang, R.H. Kuhlman, K. Zegers, J.C. Kraak, H. Poppe, J. Chromatogr. 471 (1989) 429. [20] J. Preisler, E.S. Yeung, Anal. Chem. 68 (1996) 2885. [21] H. Lindner, W. Helliger, B. Sarg, C. Meraner, Electrophoresis 16 (1995) 604. [22] J.F. Kelly, S.J. Locke, L. Ramaley, P. Thibault, J. Chromatogr. A 720 (1996) 409. [23] R.W. Chiu, J.C. Jimenez, C.A. Monnig, Anal. Chim. Acta 307 (1995) 193. [24] F.B. Erim, A. Cifuentes, H. Poppe, J.C. Kraak, J. Chromatogr. A 708 (1995) 356. [25] Y.J. Yao, S.F.Y. Li, J. Chromatogr. A 663 (1994) 97. [26] D. Corradini, G. Cannarsa, E. Fabbri, C. Corradini, J. Chromatogr. A 709 (1995) 127. [27] J. Varghese, R.B. Cole, J. Chromatogr. A 652 (1993) 369. [28] J.M. Cunliffe, N.E. Baryla, C.A. Lucy, Anal. Chem. 74 (2002) 776. [29] K.K.-C. Yeung, C.A. Lucy, Anal. Chem. 69 (1997) 3435. [30] H. Zhang, K.K.-C. Yeung, Anal. Chem. 76 (2004) 6814. [31] D.C. Simpson, R.D. Smith, Electrophoresis 26 (2005) 1291. [32] A. Puerta, J. Axen, L. Soderberg, J. Bergquist, J. Chromatogr. B 838 (2006) 113. [33] J.R. Catai, J.S. Torano, G.J. de Jong, G.W. Somsen, Electrophoresis 27 (2006) 2091. [34] B. Yu, H.L. Cong, H.W. Liu, Y.Z. Li, F. Liu, J. Sep. Sci. 28 (2005) 2390. [35] R.D. Smith, C.J. Barinaga, H.R. Udseth, Anal. Chem. 60 (1988) 1948. [36] B.A. Williams, G. Vigh, Anal. Chem. 68 (1996) 1174. [37] L. Gao, S. Liu, Anal. Chem. 76 (2004) 7179. [38] S. Liu, C. Zhang, J.L. Campbell, H. Zhang, K.K.-C. Yeung, V.K. Han, G.A. Lajoie, Rapid Commun. Mass Spectrom. 19 (2005) 2747. [39] J. Horvath, V. Dolnik, Electrophoresis 22 (2001) 644. [40] C. Schwer, E. Kenndler, Anal. Chem. 63 (1991) 1801. [41] N.A. Assuncao, L.A.G. Deziderio, L.G. Paulino, K.O. Lupetti, E. Carrilho, Electrophoresis 26 (2005) 3292. [42] F. Steiner, M. Hassel, J. Chromatogr. A 1068 (2005) 131. [43] Y. Yang, R.I. Boysen, M.T. Hearn, J. Chromatogr. A 1043 (2004) 91. [44] E. Gasteiger, A. Gattiker, C. Hoogland, I. Ivanyi, R.D. Appel, A. Bairoch, Nucleic Acids Res. 31 (2003) 3784. [45] P. Cao, J.T. Stults, Rapid Commun. Mass Spectrom. 14 (2000) 1600. [46] B. Ma, K. Zhang, C. Hendrie, C. Liang, M. Li, A. Doherty-Kirby, G. Lajoie, Rapid Commun. Mass Spectrom. 17 (2003) 2337. [47] D.N. Perkins, D.J. Pappin, D.M. Creasy, J.S. Cottrell, Electrophoresis 20 (1999) 3551.