Effect of Phosphorylation on the Collision Cross Sections of Peptide Ions in Ion Mobility Spectrometry

The insertion of ion mobility spectrometry (IMS) between LC and MS can improve peptide identification in both proteomics and phosphoproteomics by providing structural information that is complementary to LC and MS, because IMS separates ions on the basis of differences in their shapes and charge states. However, it is necessary to know how phosphate groups affect the peptide collision cross sections (CCS) in order to accurately predict phosphopeptide CCS values and to maximize the usefulness of IMS. In this work, we systematically characterized the CCS values of 4,433 pairs of mono-phosphopeptide and corresponding unphosphorylated peptide ions using trapped ion mobility spectrometry (TIMS). Nearly one-third of the mono-phosphopeptide ions evaluated here showed smaller CCS values than their unphosphorylated counterparts, even though phosphorylation results in a mass increase of 80 Da. Significant changes of CCS upon phosphorylation occurred mainly in structurally extended peptides with large numbers of basic groups, possibly reflecting intramolecular interactions between phosphate and basic groups.


INTRODUCTION
Protein phosphorylation is a reversible post-translational modi cation that in uences protein folding, activity, protein-protein interaction and subcellular localization. 1) It is well known to play key roles in intracellular signal transduction pathways regulating numerous cell functions. 2) Various techniques have been developed to monitor altered phosphorylation events in cells, and mass spectrometry is currently one of the most powerful techniques for proteome-wide experiments. 3,4) In LC/MS-based shotgun proteomics, MS/MS spectra are mainly used to identify peptides but other information, such as LC retention time, can also help to increase the con dence of sequence assignments. 5) With recent improvements in prediction models, [6][7][8][9][10] peptide retention time information has become an increasingly powerful aid for peptide identi cation and quantitation, especially in target mode 11,12) and data-independent acquisition mode 13) analyses. However, the models for predicting the retention times of phosphopeptides are not yet mature, in part due to the absence of a comprehensive understanding of the retention mechanisms of the phosphopeptides. Accordingly, the retention mechanism of phosphorylated peptides has been investigated by comparing pairs of phosphorylated and unphosphorylated peptides. [14][15][16][17] e insertion of ion mobility spectrometry (IMS) between LC and MS has recently attracted interest as a means of improving peptide identi cation in both proteomic and phosphoproteomic experiments. [18][19][20][21][22] IMS adds a dimension of separation and provides structural information that is complementary to LC and MS, because it separates ions on the basis of di erences in their shapes and charge states. Again, though, it is necessary to know how phosphate groups a ect the peptide collision cross sections (CCS) in order to accurately predict phosphopeptide CCS values and to maximize the usefulness of IMS. Previous IMS-MS studies [23][24][25][26] on small numbers of singly and doubly charged phosphopeptide ions have indicated that the CCS of phosphopeptides is smaller than that of unmodi ed peptides of equivalent mass. Based on these ndings, the model using the intrinsic size parameter (ISP) 27,28) for the prediction of CCS of phosphopeptides has been extended 24) ; the ISP provides the average contribution of each amino acid residue or modi cation to the CCS of a peptide, and can predict the CCS based on the amino acid composition of the peptide. However, this model does not currently take into account sequence-speci c features, such as positional information or residue combinations. Also, in this model, changes in CCS upon phosphorylation are treated as increases in CCS, 24,29) and it is not possible to explain the decrease in CCS due to phosphorylation that is seen in certain sequences. 24,30) Furthermore, the model was developed using doubly charged peptides with molecular weights ranging from 1,000 Da to 1,400 Da, and thus cannot be applied directly to larger, more highly charged ions in the 600 Da to 5,000 Da range that are commonly found in proteomics experiments. erefore, further data collection is needed to establish an improved model for CCS prediction of phosphopeptides.
Here, we systematically characterized the CCS values of phosphopeptides and their corresponding unmodi ed counterparts using trapped ion mobility spectrometry (TIMS). TIMS captures ions in an RF ion tunnel by the forces of the gas ow from the ion source and the opposing counteracting electric eld, and the trapped ions are then sequentially released according to their CCS by lowering the electric potential. 31) e CCS values of 6,544 phosphopeptide ions from HeLa tryptic digests were examined, yielding 4,433 CCS pairs of phosphopeptide and unphosphorylated peptide ions with charge states ranging from 2+ to 4+ in the mass range from 800 Da to 4,500 Da. Our results show that large changes in CCS occur mainly in structurally extended peptides with multiple basic groups.

Materials
UltraPure ™ Tris Bu er was purchased from ermo Fisher Scienti c (Waltham, MA, USA). Sequencing grade modi ed trypsin was purchased from Promega (Madison, WI, USA). Water was puri ed by a Millipore Milli-Q system (Bedford, MA, USA). Porous titanium dioxide beads (TitansphereTiO, 10 µm) and Empore disks (SDB-XC and C8) were obtained from GL Sciences (Tokyo, Japan). All other chemicals and reagents were purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan), unless otherwise speci ed.

Cell culture and protein digestion
HeLa cells were cultured to 80% con uence in DMEM containing 10% FBS in 10 cm diameter dishes. Cells were washed twice with ice-cold PBS, collected using a cell scraper, and pelleted by centrifugation. Hela cell lysates were digested by means of the phase-transfer surfactant (PTS)-aided trypsin digestion protocol, as described previously. 32) A er digestion, the sample was desalted using SDB-XC StageTips. 33)

Phosphopeptide enrichment
Metal oxide chromatographic (MOC) tips were prepared as described previously. 34) Brie y, C8 StageTips packed with TiO 2 beads (0.5 mg/tip) were equilibrated with 80% ACN with 0.1% tri uoroacetic acid (TFA) and 300 mg/mL lactic acid as a selectivity enhancer (solution A). e samples were diluted with an equal amount of solution A and loaded onto the MOC tips. A er washes with solution A and 80% ACN with 0.1% TFA, phosphopeptides were eluted with 0.5% piperidine. e eluate was acidi ed with 10% TFA and desalted using StageTips.

Dephosphorylation of phosphopeptides
One-third of the phosphopeptide sample was dried and dissolved in 25 µL of 100 mM Tris-HCl bu er (pH 9.0). Alkaline phosphatase (from calf intestine; 5 units) was added, and the solution was incubated for 3 h at 37°C. A er the reaction, the bu er was acidi ed by adding 10% TFA 10 µL. e samples were desalted using StageTips.
e TIMS section was operated with a 200 ms ramp time and a scan range of 0.6-1.5 Vs cm −2 . One cycle was composed of 1 MS scan followed by 10 PASEF MS/MS scans. MS and MS/MS spectra were recorded from m/z 100 to 1,700. Singly charged ions were excluded from the precursor ions based on their m/z and 1/K 0 values. e quadrupole isolation width was set to 2 Da. e TIMS elution voltage was calibrated linearly to obtain the reciprocal of reduced ion mobility (1/K 0 ) using three selected ions (m/z 622, 922, and 1,222) of the ESI-L Tuning Mix (Agilent, Santa Clara, CA, USA).

Database searching and data processing
Peptides and proteins were identi ed through automated database searching using MaxQuant 35,36) (version 1.6.14.0) in the TIMS-DDA mode against the human database from UniprotKB/Swiss-Prot release 2017/04 with a strict Trypsin/P speci city allowing for up to 2 missed cleavages. Carbamidomethyl (C) was set as a xed modi cation. Oxidation (M), Acetyl (Protein N-term) and Phospho (STY) were allowed as variable modi cations. e resulting evidence.txt le was used for the analysis. Collision cross sections were calculated from the reciprocal of reduced ion mobility (1/K 0 ) using MaxQuant.

Data availability
e MS raw data and analysis les have been deposited with the ProteomeXchange Consortium (http:// proteomecentral.proteomexchange.org) via the jPOST partner repository 37,38) (https://jpostdb.org) with the data set identi er PXD019746. To view all data, MaxQuant so ware should be downloaded from the website (https://www. maxquant.org/).

RESULTS & DISCUSSION
First, a sample containing thousands of pairs of phosphorylated and unphosphorylated peptides was fractionated using the C4 column to reduce the sample complexity. 17) A er fractionation, each fraction was analyzed by LC/ TIMS/MS/MS to obtain paired CCS values. To simplify the analysis, peptides with multiple CCS values were aggregated to the CCS with the highest peak intensity, and peptides containing methionine oxidation and N-terminal acetylation were removed. Finally, CCS values of 6,544 phosphopeptide ions and 22,666 unphosphorylated peptide ions, including more than 4,000 paired CCS values, were obtained from the HeLa tryptic digest (Table 1). Figure 1A shows the CCS versus m/z plots obtained for unphosphorylated peptide ions. In this plot, the doubly, triply and quadruply charged populations are clearly separated. ere is a strong correlation between mass and CCS within each charge state, and the triply charged species are clearly split into two subpopulations: extended and compact forms.
ese observations agree with previous IMS-MS ndings. 18,[39][40][41] Figure 1B shows CCS versus m/z plots of phosphorylated peptide ions. e trend for phosphorylated peptide ions was similar to that for unphosphorylated peptide ions. For further investigation, phosphorylated and unphosphorylated peptide ions were classi ed according to their charge state and overlaid on the CCS versus mass plot ( Figs. 2A-C). e di erence in the average CCS values of phosphorylated and unphosphorylated peptides in each molecular weight bin, plotted in Fig. 2D, shows that phosphorylated peptides are generally smaller than unmodi ed peptides of the same mass, which is consistent with previous IMS-MS studies. [23][24][25] While previous studies have focused primarily on singly and doubly charged peptide ions, we have con rmed the same trend for triply and quadruply charged peptides.
To further evaluate the e ect of phosphorylation of peptides on CCS, pairs of mono-phosphorylated and unphosphorylated peptides with identical sequences were compared. Considering the correlation between CCS and mass, the mono-phosphorylated peptide should have a larger CCS than the unmodi ed sequence, since phosphorylation results in an 80 Da mass increase. In fact, the median CCS of the mono-phosphopeptides was 4.0 Å 2 larger than that of the unphosphorylated peptides (Fig. 3A). However, 32.0% (1419/4433) of the mono-phosphorylated peptide ions had lower CCS values than the corresponding unphosphorylated form, suggesting that a signi cant number of peptides underwent compaction of their conformation upon phosphorylation (Fig. 3A). e content of the CCScompressed phosphopeptides (∆CCS <0) was much higher than that reported for 66 doubly charged phosphopeptides, which was 19.7% (13/66). 24) Indeed, for doubly charged mono-phosphopeptides in our data, the content was 23.4% (642/2743), as shown in Fig. 3B. In the case of triply and quadruply charged species, the contents of the phosphopeptides with ∆CCS <0 were 46.2% (688/1488) and 44.1% (89/202), respectively (Figs. 3C, D). ese results can be partly explained by the molecular weight distribution of the  peptides in each charge state, since the more highly charged peptides have higher molecular weights and the ∆CCS values were weakly negatively correlated with the molecular weight of the peptides (Fig. 4). For larger peptides, the mass shi of 80 Da was small compared to the total mass of the peptide, and thus the e ect of the mass increase on CCS was expected to be relatively small. However, for some of the larger peptides, a signi cant change in CCS was observed, which could be attributed to the rearrangement of peptide conformation upon phosphorylation. On the other hand, relatively small doubly charged peptides, such as those examined in the previous study, did not show such a substantial change, and mainly showed a positive shi in CCS due to the positive mass shi . However, even within the speci c MW range of 1,500 Da to 2,000 Da, 26.8% (306/1140) and 48.0% (231/481) of doubly and triply charged mono-phosphorylated peptide ions, respectively, showed CCS compaction upon phosphorylation.
ese results suggest that ions with higher charge   states show more pronounced conformational compaction upon phosphorylation. is may be attributed to the wider distribution of CCS for the triply charged ions than for the doubly charged ions, as has been indicated in previous reports. 39,42) e structural diversity of the triply charged ions may allow for a more dynamic change of structure upon phosphorylation, but further studies are needed to test this idea.
To assess what kinds of peptide features are associated with the CCS change upon phosphorylation, we examined the relationship between m/z, CCS and ∆CCS values (Fig.  5). e colors of the markers indicate the ∆CCS values; the most marked change in CCS appears to be associated with the extended and compact subpopulations of triply charged ions. Although less pronounced, similar trends were observed for doubly and quadruply charged ions. For the compact subpopulation, the CCS increase up to 30 Å could be explained by the increase in mass without the transition to the extended form, whereas the CCS reduction upon phosphorylation would be due to the transition from extended to compact form.
For further investigation, we focused on peptides in the extended subpopulation of triply charged species. e extended forms were visually classi ed based on their distributional shapes ( Figure S1). In addition, to assess the phosphorylation position dependence of CCS changes, only phosphopeptides with con dently localized phosphosite 43) (phosphosite localization probability >0.75) were considered, leaving 277 pairs of phosphopeptides and unphosphorylated peptides. ese pairs were used to examine the relationship between ∆CCS and the relative position of the phosphorylated amino acid residue within the peptide. It appeared that phosphorylation in the terminal region of the peptide is more likely to cause conformational compression, although the tendency is not large (Fig. 6). Although there is insu cient molecular structure information to interpret the CCS data properly, the observations might be explained in terms of intramolecular interactions involving the phosphate groups. Phosphorylated residues are thought to be involved in multiple intramolecular interactions of peptide ions, and previous work 30,44) on several doubly charged phosphopeptide ions has suggested that intramolecular salt bridges or ionic hydrogen bonds 45) are formed between phosphate and basic groups in speci c sequences.
It is possible that the phosphate group and its distal basic group may interact to compress the peptide structure, as shown in previous studies. 30,44) Another possibility is that the phosphate group may interact with its proximal basic group to disrupt the balance of positive charges in the peptide sequence, which would lead to changes in the peptide structure. 39,46,47) In other words, terminal phosphorylation may alter the charge balance within the peptide, leading to structural change.
To further evaluate these possibilities, we investigated the e ects of basic amino acids on ∆CCS. In the dataset of 277 triply charged pairs, we found that the number of basic groups (Lys, His, Arg, and N term) of the peptide was weakly negatively correlated to ∆CCS (Fig. 7A).
is suggests that basic groups are involved in conformational compression upon phosphorylation. Considering that salt bridges or ionic hydrogen bonds require additional basic groups that do not contribute to the overall charge state of the peptide, peptides with more basic groups are more likely to have these intramolecular interactions. erefore, we further focused on 127 of 277 triply charged peptide pairs with more basic groups than their charge state. ese peptides would have basic groups that do not contribute to the total number of charges. Considering that the phosphate group in the terminal region of the peptide always has a proximal basic group derived from an N-terminal amine group or a C-terminal Lys or Arg, we examined the relationship between the ∆CCS and the distance between the phosphosite and its closest basic group, but no speci c correlation was observed (Fig. 7B). Next, we considered that the distance between other basic groups and the phosphorylation site may be important for the structural change of the peptide upon phosphorylation, and indeed, we found that the relative positional distance between the phosphosite and its second nearest basic group was weakly negatively correlated with ∆CCS (Fig. 7C). ese results can be explained by either of the above-mentioned mechanisms. Phosphorylation negates the positive charge of the closest basic group, and  because the next closest basic group (or positive charge) is relatively far from the closest basic group, this disrupts the balance of charge in the peptide and causes a change in structure. Alternatively, an intramolecular interaction between the phospho site and the basic group distal to it may result in compression of the peptide ion structure. In any case, the results obtained in this study suggest that the interaction between phosphate and basic groups contributes signi cantly to the structural compression of phosphopeptides.

CONCLUSION
In this study, the CCS values of phosphopeptides and the corresponding unmodi ed counterparts were systematically pro led using TIMS. In agreement with previous reports, phosphorylation mainly results in compaction of peptide conformations. We also found that the CCS values of triply and quadruply charged ions are more strongly a ected by phosphorylation, which could be explained at least in part by conformational changes of these larger peptides. e most prominent CCS changes were seen for triply charged peptides with extended structures and a larger number of basic groups, presumably re ecting intramolecular interactions between the phosphate group and the basic groups. We believe these ndings will improve the CCS prediction of phosphopeptides.
Abbreviations LC, liquid chromatography; IMS, ion mobility spectrometry; MS, mass spectrometry; MS/MS, tandem mass spectrometry; TIMS, trapped ion mobility spectrometry; Q/TOF, quadrupole time-of-ight; DDA, data-dependent acquisition; PASEF, parallel accumulation-serial fragmentation; K 0 , reduced ion mobility; ISP, intrinsic size parameter; MOC, metal oxide chromatography; TFA, tri uoroacetic acid; ACN, acetonitrile; Lys, lysine; His, histidine; Arg, arginine distance between their phosphosite (pSite) and its nearest basic group. e relative positional distance was calculated by dividing the positional distance between the phosphosites and the basic groups by the length of the peptide sequence. e line represents a linear regression, and the shading indicates the 95% con dence interval.