Letter pubs.acs.org/ac
N‑Capping Motifs Promote Interaction of Amphipathic Helical Peptides with Hydrophobic Surfaces and Drastically Alter Hydrophobicity Values of Individual Amino Acids Vic Spicer,† Ying W. Lao,‡ Dmitry Shamshurin,† Peyman Ezzati,† John A. Wilkins,†,§ and Oleg V. Krokhin*,†,§ †
Manitoba Centre for Proteomics and Systems Biology, University of Manitoba, 799 JBRC, 715 McDermot Avenue, Winnipeg, Manitoba R3E 3P4, Canada ‡ Department of Chemistry, University of Manitoba, 360 Parker Building, Winnipeg, Manitoba R3T 2N2, Canada § Department of Internal Medicine, University of Manitoba, 799 JBRC, 715 McDermot Avenue, Winnipeg, Manitoba R3E 3P4, Canada S Supporting Information *
ABSTRACT: Capping rules, which govern interactions of helical peptides with hydrophobic surfaces, were never established before due to lack of methods for the direct measurement of polypeptide structure on the interphase boundary. We employed proteomic techniques and peptide retention modeling in reversed-phase chromatography to generate a data set sufficient for amino acid population analysis at helix ends. We found that interactions of amphipathic helical peptides with a hydrophobic C18 phase are induced by a unique motif featuring hydrophobic residues in the N1 and N2 positions adjacent to the N-cap (Asn, Asp, Ser, Thr, Gly), followed by Glu, Gln, or Asp in position N3 to complete a capping box. A favorable N-capping arrangement prior to amphipathic helix may result in the highest hydrophobicity (retention on C18 columns) of Asp/Asn (or Glu/Gln) peptide analogues among all naturally occurring amino acids when placed in N-cap or N3 position, respectively. These results contradict all previously reported hydrophobicity scales and provide new insights into our understanding of the phenomenon of hydrophobic interactions.
A
Similar to X-ray crystallography, reversed-phase chromatography (RPC) was used to determine amino acid hydrophobicity values.8 Because of the structural similarity between C18 functionality and the phospholipid bilayer, RPC is also considered an ideal tool to study peptide interactions with hydrophobic biomembranes, including the secondary structure effects.9,10 Peptides with identical amino acid composition but different sequences showed increased retention for amphipathic helical molecules.11 This was explained by the presence of a preferred binding domain such as a hydrophobic face of an amphipathic α-helix and led to the conclusion that the hydrophobic interaction of side chains dominates peptide retention in these systems. RPC amino acid hydrophobicity values were determined for both random coil conformation12 and residues situated in the hydrophobic face of an amphipathic helix.10 Tryptophan and isoleucine were found the most hydrophobic residues in these two cases, respectively. Capping
n accurate modeling of the stabilization of amphipathic helical peptides upon interaction with hydrophobic surfaces is fundamental in the development of many areas of biological/ chemical science. Specific N- and C-capping motifs stabilize helices in proteins via the saturating of unpaired NH donors and CO acceptors in the first and last turn of the α-helix, respectively.1−3 Analysis of X-ray crystallography data helped establish some of the first hydrophobicity scales4,5 and helix capping rules1−3 in proteins, as a foundation of modern structural protein chemistry. Capping motifs were found by determining empirical preferences of each amino acid at helical ends within known protein structures following the rapid growth in crystallography data in the late 1980s. Capping effects were also confirmed by extensive studies on synthetic peptides using circular dichroism measurements.6,7 Richardson and Richardson1 introduced a nomenclature to assign residues located before and after the N-cap of a helix: ...N″−N′−N-cap−N1−N2.... One of the most dominant N-capping arrangements, the “capping box”, features reciprocal hydrogen bonds between side chains and backbone groups of N-cap (Asn, Ser, Thr, Asp, Gly) and N3 (Glu, Asp) residues.2 © 2014 American Chemical Society
Received: September 5, 2014 Accepted: November 5, 2014 Published: November 5, 2014 11498
dx.doi.org/10.1021/ac503352h | Anal. Chem. 2014, 86, 11498−11502
Analytical Chemistry
Letter
identification with a ±10 ppm and ±30 mDa mass tolerance for parent and daughter ions, respectively. Tryptic peptides with confident identification scores (log(e) < −3) were chosen to populate the combined retention database. Retention times of individual peptides were converted into hydrophobicity index (HI, % acetonitrile) values by linear regression against the retention values of standard peptides spiked in each fraction.27 A version of the SSRCalc (formic acid) model14,25 without helicity corrections was used to predict HI values for identified peptides. In total, 5141 sequences with a prediction error >3% acetonitrile were interrogated using an in-house algorithm to assign the hydrophobic face of the helix, followed by manual correction of these assignments. These sequences were analyzed using Richardson and Richardson’s approach,1 and the Motif-X online tool that was designed to extract overrepresented patterns from any sequence data set.28 Normalized hydrophobicity values were calculated by a linear transformation of their respective scales5,10,12 (RPC retention values) to yield an average value equal to 0 and standard deviation equal to 1.
effects were never studied in peptide RPC, as the use of synthetic peptides has been cost prohibitive for building the extensive data sets needed for population analysis. Another limiting factor is the inability of chromatographic measurements to locate the exact position of the helix ends upon interaction with stationary phase. Modern proteomics LC−MS methods are capable of identifying thousands of proteins in a few hours.13 The output of such analyses carries retention information for often tens of thousands peptides, which have been used to model their RPC retention.14−17 Peptide hydrophobicity in RP systems is determined by the residue count population,18 their locations relative to peptide ends,19 the peptide length,20 overall hydrophobicity,14 and secondary structure effects. Predicting the latter remains the most challenging part in peptide retention modeling. Eisenberg’s hydrophobic moment21 and the AGADIR algorithm7 have been tested as tools to predict the contribution of peptide helicity in RPC retention14,15 but with very limited success. All together these suggest that the interaction of amphipathic helical peptides with hydrophobic surfaces is governed by different rules, which to-date remain largely unknown. We recently reanalyzed22 the 1987 Houghten and DeGraw data23 on hydrophobicity measurement of 260 substitution analogues of the Ac-YPYDVPDYASLRS-Amide peptide and found significant positive deviations from calculated retention values in peptides carrying the predicted N-capping motifs. We also confirmed that the main reason for the unexpectedly high retention of tryptic peptides in proteomic RPC−MS measurements is indeed amphipathic helicity.22 These motivated the present study, which uses the retention data from large-scale proteomic analyses along with peptide retention modeling in an attempt to generate an extended collection of amphipathic peptides and establish capping rules upon their interaction with a hydrophobic surface via population analysis. Therefore, we aim to reproduce the classical work by Richardson and Richardson1 on establishing capping preferences in protein α-helices but using the output from an alternative analytical technique: peptide reversed-phase chromatography instead of X-ray crystallography.
■
RESULTS AND DISCUSSION Generating a Representative Data Set of Amphipathic Helical Peptides with Strong Affinity to the C18 Surface. Our Sequence Specific Retention Calculator (SSRCalc)14 model takes into account various parameters affecting peptide retention on C18 columns. Currently its helicity correction algorithm is driven by enumerating values assigned to potentially amphipathic motifs XXOOXX, XXOOX, XXOXX, etc., where X is hydrophobic and O is any other residue.14 For the purposes of analyzing the helical contribution, we assume that two major components (nonhelical and helical) represent RPC retention or experimental hydrophobicity index (HIExp, % acetonitrile27): HIExp = HINonHel + HIHel. Provided that we can predict the nonhelical component (SSRCalc with no helicity correction) with sufficient precision, we can express the helical component as just HIHel = HIexp − HISSRCalc NonHel. Ten 2D LC-MS “bottom-up” proteomic runs contributed to a ∼280 000 peptide collection with 0.955 R2 correlation between HIexp and HISSRCalc NonHel (Figure 1A). Peptides with large positive prediction errors are predominantly amphipathic22 (Figure 1B−D); therefore, their retention prediction error (ΔACN = HIexp − HISSRCalc NonHel) is dominated by the helical component: ΔACN ≈ HIHel. We selected 5141 sequences with ΔACN >3% acetonitrile (shown in red in Figure 1A) for further analysis. Determination of Amino Acid Preferences at Helix Ends via Population Analysis. This retention prediction based filtering does not provide the location of the capping residues within a given peptide, a critical value when analyzing amino acid preferences at the helix ends.1−3 We used the hydrophobic termini (restricted to Ala, Tyr, Met, Val, Phe, Trp, Ile, Leu) of the amphipathic stretch as reference points, as it was the only positional information relevant to helix stabilization available to us by our approach. Amphipathic helix boundaries, the first and last hydrophobic residues, were assigned for 5010 assumed helical peptide sequences (Supplementary Table 1 in the Supporting Information). This significantly exceeds the number of helixes used to derive capping preferences in proteins: 215 used by Richardson and Richardson1 and 1316 used by Aurora et al.2 Similarly, we adopted the naming convention for N-cap/C-cap,1 but referenced against the starting (S) and ending (E) positions of the hydrophobic residues: ...S″−S′− S−S1−S2− S3... and ...E2−E1−E−E′−E″.... The amino acid preference
■
EXPERIMENTAL SECTION Sample Preparation for Tryptic Digests and Synthetic Peptide Libraries. Tryptic digests of samples of different origin (flax seed protein extract, mitochondrial isolate from N. crassa, whole cell lysates of F. graminearum, E. coli, C. serevisiae, C. stercorarium, T. fusca, Thermoanaerobacter WC1, human Jurkat, and mouse L929 cells lines) were prepared using the FASP protocol.24 The synthetic peptide libraries (20-residues substitutions at position X in XIVEEIEEVIGEGER, NIVXEIEEVIGEGER) were prepared by JPT (Berlin, Germany) in Spike Tide format. LC−MS Analysis. Approximately 100 μg of the digest were used for the 2D LC−MS in each case. The separation system included a first dimension separation using RPC at pH 10 and a second dimension nanoflow RPC with formic acid as an eluent additive as described elsewhere.25 Twenty pair wise concatenated fractions were collected during first dimension separation and analyzed individually in the second dimension RP LC−MS using linear water−acetonitrile gradients (0−35% acetonitrile in 1−1.5 h each). A Triple TOF5600 mass spectrometer (ABSciex, Mississauga, ON) was used in standard data-dependent acquisition mode. Peptide Identification and Sequence Analysis. The X! Tandem search algorithm26 was used for peptide/protein 11499
dx.doi.org/10.1021/ac503352h | Anal. Chem. 2014, 86, 11498−11502
Analytical Chemistry
Letter
Type 3 (highlighted in blue) is shown for TANNNVVQVIEWVVEK (Mus musculus, ΔACN 5.2%). Preferences for Asn, Asp, Ser, Thr to maximize again at the S2 position, corresponding to cases where the hydrophobic residue in the hydrophobic face of the helix appears three positions prior to the actual N-cap residue. Similarly, the capping box is completed with Glu or Gln in the N3 (S5) position. This case illustrates the inability of RPC measurements to detect actual capping residues. The hydrophobic Ala in TANNNVVQVIEWVVEK belongs to the hydrophobic face (Figure 1D) and therefore was assigned as the first hydrophobic residue. Indeed the first hydrophobic residue in amphipathic stretch is Val, flanked by Asn and Gln as a capping box partners. We did not find any significant population trends supporting C-cap stabilization in our data set (Supplementary Table 2 in the Supporting Information). Analysis of 5010 sequences with high retention prediction errors using Motif-X28 showed significant overrepresentation of short amphipathic stretches such as (..L.E.I...) and (..L..IL...), confirming overall amphipathic character of the data set. The highest motif scores for typical N-cap residues were found corresponding to Type 2 N-caps such as (.....KDL...L.), (......SIL..L.), (......SL...L.) (Supplementary Table 3 in the Supporting Information). Hydrophobicity Scales for Amino Acids Inside a N-Cap Motif Preceding the Amphipathic Helix. The separation of synthetic peptides analogues with a 20-residue substitution at a particular position has been established as a standard procedure of determining amino acid hydrophobicity by RPC.8,10,12 Knowing N-capping preferences allows us to explore the magnitude of hydrophobicity variation observed due to N-cap stabilization. We performed RPC analysis of substitution analogues (positions 1 and 4) of the model peptide NIVEEIEEVIGEGER, which contains a favorable Type 2 Ncap box arrangement prior to a 2-turns length amphipathic helix. Strikingly, we find that Asn and Asp analogues in the N-cap position of the XIVEEIEEVIGEGER series of peptides exhibit the highest RPC retention, i.e., highest hydrophobicity among naturally occurring amino acids (Figure 2). This finding contradicts all previously reported hydrophobicity scales,4,5,10,12 where Asn and Asp are the most hydrophilic residues (apart from charged Lys, Arg, His). N-cap Ser, Thr, and Gly residues exhibit unusually high hydrophobicity values as well, as illustrated in Figure 2. Similarly we find Gln, Glu are the most hydrophobic residues in the N3 position for variations of the NIVXEIEEVIGEGER model peptide (Figure 2).
Figure 1. Selection of amphipathic helical peptides with strong RPC retention for population analysis. (A) Ten 2D LC−MS proteomics runs were used to generate a ∼280 000 peptide retention data set. In total, 5141 peptides (highlighted in red) with retention prediction error >3% acetonitrile were selected for the analysis of amino acid preferences at helix ends; (B−D) axial helical projections with an assigned hydrophobic face (highlighted in red) for selected peptides, with the Ncapping motifs underlined (see Table 1 for details).
values in Table 1 are a measure of their abundance near the start of the amphipathic helix within the 5010 selected peptides, relative to their abundance in the general (∼280 000) population of all peptides in the data set. We observed that the distribution of preferences in Table 1 encodes the superposition of three possible locations for strong N-cap residues relative to first hydrophobic residue in the helical stretch. We denote these as Types 1, 2, and 3. Type 1 (highlighted in green in Table 1) is identical to the classical N-cap arrangement found in proteins: Asn, Asp, Ser, Thr, Gly (all typical N-cap residues1,2) preferences maximize at position S4′. Taking N-cap as a reference point, preferences for hydrophobic residues maximize at position N4 (S), Pro in N1 (S‴), and Glu in N3 (S′) to complete the capping box. A typical example of this arrangement is shown in Figure 1B for the peptide GEINPTEVVSALIDR from C. stercorarium with ΔACN of 4.8% acetonitrile. Type 2 (highlighted in pink) shows the highest N-cap preferences of the three types. The N-cap (S′) position preceding the first hydrophobic residue S has values dominated by Asn, Asp, Ser, Thr, Gly. Amino acids Glu, Gln are favorable for position N3 (S2), completing the capping box. There are no Pro residues in the N1 (S) position, as its being assigned for hydrophobic residues. Amphipathic peptides propagate further: S3 and S4 (N4 and N5) are preferred for hydrophobic residues completing the first turn of the amphipathic helix. The peptide ESIEEAVSEVVNALK from E. coli (ΔACN 10.7% acetonitrile) represents this type N-cap motif (Figure 1C).
■
CONCLUSIONS We demonstrate for the first time that chromatographic measurements can be used to establish capping motifs, which promote interaction of amphipathic helices with hydrophobic surfaces via hydrogen bonding. The major divergence between N-cap motifs in peptides exhibiting unexpectedly strong interactions with the C18 phase and proteins or peptides in solution is the location of the first hydrophobic residue. For peptides interacting with the C18 phase that is at N1/N2 adjacent to the N-cap, whereas for proteins or peptides in solution it is at N4. The retention of peptides with strong Ncapping prior to their amphipathic helices cannot be predicted accurately using the current paradigms of hydrophobicity and retention modeling. However, this approach has permitted the extraction of a large pool of peptides with N-cap stabilization from an enormous collection of random tryptic peptides, sufficient in size for sequence population analysis. These findings 11500
dx.doi.org/10.1021/ac503352h | Anal. Chem. 2014, 86, 11498−11502
Analytical Chemistry
Letter
Table 1. N-Terminal Amino Acid Preferencesa in Helices with Strong RPC Interaction
a
The normalized frequency (f) is calculated as f = (fraction of i in position j of 5010 helical sequences)/(fraction of i in 280 000 peptide data set). Strong N-cap box partners for reciprocal hydrogen bonding. cCarboxamidomethyl-Cys residue. dN-cap assignments according to Richardson and Richardson1 nomenclature. eNumber of occurrences of strong N-cap residues in S4′ (Type1), S′ (Type 2) and S2 (Type 3) positions. Note the overlap between Types 1 and 2, 455 instances; Types 2 and 3, 1398 instances; Types 1 and 3, 366 instances. b
Figure 2. Comparison of previously reported hydrophobicity scales and those observed via amino acid substitutions within the N-capping box of model peptides. (A) Kyte−Doolittle scale5 based on a combination of accessible surface area measurements and water−vapor partitioning values; (B) Kovacs et al.’s scale12 for peptides in a random coil conformation; (C) Sereda et al.’s scale10 for substitution within the Leu face of an amphipathic helix; (D) substitution at N-cap position in XIVEEIEEVIGEGER; (E) substitution at N3 position in NIVXEIEEVIGEGER (parts B−E determined by RPC).
■
open up avenues to computational modeling of peptide interactions with hydrophobic surfaces, further improvement in peptide retention prediction in RPC, and adding new dimensions to our understanding of hydrophobic interactions in biological systems.
■
AUTHOR INFORMATION
Corresponding Author
*Fax: (204) 480 1362. E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ASSOCIATED CONTENT
S Supporting Information *
ACKNOWLEDGMENTS We thank Drs. D. Court, K. Coombs, D. Levin, R. Sparling, B. Hasinoff (all University of Manitoba), S. Fong (Virginia
Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. 11501
dx.doi.org/10.1021/ac503352h | Anal. Chem. 2014, 86, 11498−11502
Analytical Chemistry
Letter
Commonwealth University), and C. Rampitch (Agriculture and Agri-Food Canada) for providing cell cultures and useful discussions. This study was supported by the Natural Sciences and Engineering Research Council of Canada Discovery Grant (O.V.K.) and Genome Canada (MGCB2 project, O.V.K, J.A.W.).
■
REFERENCES
(1) Richardson, J. S.; Richardson, D. C. Science 1988, 240, 1648−1652. (2) Aurora, R.; Rose, G. D. Protein Sci. 1998, 7, 21−38. (3) Aurora, R.; Srinivasan, R.; Rose, G. D. Science 1994, 264, 1126− 1130. (4) Janin, J. Nature 1979, 277, 491−492. (5) Kyte, J.; Doolittle, R. F. J. Mol. Biol. 1982, 157, 105−132. (6) Doig, A. J.; Baldwin, R. L. Protein Sci. 1995, 4, 1325−1336. (7) Munoz, V.; Serrano, L. J. Mol. Biol. 1995, 245, 275−296. (8) Mant, C. T.; Kovacs, J. M.; Kim, H. M.; Pollock, D. D.; Hodges, R. S. Biopolymers 2009, 92, 573−595. (9) Blondelle, S. E.; Ostresh, J. M.; Houghten, R. A.; Perez-Paya, E. Biophys. J. 1995, 68, 351−359. (10) Sereda, T. J.; Mant, C. T.; Sonnichsen, F. D.; Hodges, R. S. J. Chromatogr., A 1994, 676, 139−153. (11) Ostresh, J. M., Buttner, K.; Houghten, R. A. In HPLC of Peptides and Proteins: Separation, Analysis and Conformation; Mant, C., Hodges, R. S., Eds.; CRC Press: Boca Raton, FL, 1991; pp 633−642. (12) Kovacs, J. M.; Mant, C. T.; Hodges, R. S. Biopolymers 2006, 84, 283−297. (13) Mann, M.; Kulak, N. A.; Nagaraj, N.; Cox, J. Mol. Cell 2013, 49, 583−590. (14) Krokhin, O. V. Anal. Chem. 2006, 78, 7785−7795. (15) Petritis, K.; Kangas, L. J.; Yan, B.; Monroe, M. E.; Strittmatter, E. F.; Qian, W. J.; Adkins, J. N.; Moore, R. J.; Xu, Y.; Lipton, M. S.; Camp, D. G., 2nd; Smith, R. D. Anal. Chem. 2006, 78, 5026−5039. (16) Moruz, L.; Tomazela, D.; Kall, L. J. Proteome Res. 2010, 9, 5209− 5216. (17) Gilar, M.; Xie, H.; Jaworski, A. Anal. Chem. 2010, 82, 265−275. (18) Meek, J. L. Proc. Natl. Acad. Sci. U.S.A. 1980, 77 (3), 1632−1636. (19) Krokhin, O. V.; Craig, R.; Spicer, V.; Ens, W.; Standing, K. G.; Beavis, R. C.; Wilkins, J. A. Mol. Cell. Proteomics 2004, 3, 908−919. (20) Mant, C. T.; Burke, T. W.; Black, J. A.; Hodges, R. S. J. Chromatogr. 1988, 458, 193−205. (21) Eisenberg, D.; Weiss, R. M.; Terwilliger, T. C. Nature 1982, 299, 371−374. (22) Reimer, J.; Spicer, V.; Krokhin, O. V. J. Chromatogr., A 2012, 1256, 160−168. (23) Houghten, R. A.; DeGraw, S. T. J. Chromatogr. 1987, 386, 223− 228. (24) Wisniewski, J. R.; Zougman, A.; Nagaraj, N.; Mann, M. Nat. Methods 2009, 6, 359−362. (25) Dwivedi, R. C.; Spicer, V.; Harder, M.; Antonovici, M.; Ens, W.; Standing, K. G.; Wilkins, J. A.; Krokhin, O. V. Anal. Chem. 2008, 80, 7036−7042. (26) Craig, R.; Beavis, R. C. Bioinformatics 2004, 20, 1466−1467. (27) Krokhin, O. V.; Spicer, V. Anal. Chem. 2009, 81, 9522−9530. (28) Chou, M. F.; Schwartz, D. Using the scan-x Web Site to Predict Protein Post-Translational Modifications. In Current Protocols in Bioinformatics; John Wiley & Sons, Inc.: New York, 2011, Vol. 36, Unit 13.16, 13.16.1−13.16.8.
11502
dx.doi.org/10.1021/ac503352h | Anal. Chem. 2014, 86, 11498−11502
