Andreas
Herburger
,
Christian
van der Linde
and
Martin K.
Beyer
*
Institut für Ionenphysik und Angewandte Physik, Leopold-Franzens-Universität Innsbruck, Technikerstraße 25, 6020 Innsbruck, Austria. E-mail: martin.beyer@uibk.ac.at
First published on 15th February 2017
Protonated leucine enkephalin (YGGFL) was studied by ultraviolet photodissociation (UVPD) from 225 to 300 nm utilizing an optical parametric oscillator tunable wavelength laser system (OPO). Fragments were identified by absolute mass measurement in a 9.4 T Fourier transform ion cyclotron resonance mass spectrometer (FT-ICR MS). Bond cleavage was preferred in the vicinity of the two aromatic residues, resulting in high ion abundances for a4, a1, b3, y2 and y1 fragments. a, b and y ions dominated the mass spectrum, and full sequence coverage was achieved for those types. Photodissociation was most effective at the short wavelength end of the studied range, which is assigned to the onset of the La π–π* transition of the tyrosine chromophore, but worked well also at the Lb π–π* chromophore absorption maxima in the 35000–39000 cm−1 region. Several side-chain and internal fragments were observed. H atom loss is observed only above 41000 cm−1, consistent with the requirement of a curve crossing to a repulsive 1πσ* state. It is suggested that the photochemically generated mobile H atom plays a role in further backbone cleavages, similar to the mechanism for electron capture dissociation. The b4 fragment is most intense at the Lb chromophore absorptions, undergoing additional fragmentation at higher photon energies. The high resolution of the FT-ICR MS revealed that out of all x and z-type fragments only x3 and x4 were formed, with low intensity. Other previously reported x- and z-fragments were re-assigned to internal fragments, based on exact mass measurement.
UV and IR spectroscopy of protonated peptides14,15 are useful methods for gaining insight into electronic and geometric structure, in particular the protonation sites. The classical approach towards peptide fragmentation is collision induced dissociation (CID), where a large number of low energy collisions with a neutral gas like argon leads to fragmentation of the biomolecular ion.16,17 Infrared multiphoton dissociation (IRMPD) supplies the energy needed for fragmentation via a large number of infrared photons, typically supplied by a CO2 laser.18,19 The deposited energy per photon is low (about 0.1 eV), and the experiment has to be conducted in ultrahigh vacuum to avoid collisional deactivation. Upon ultraviolet photodissociation (UVPD), more energy can be deposited in the ion with a single photon, e.g. 6.2 eV for an ArF laser at 193 nm, which opens new fragmentation channels1 like side chain cleavages. This allows to distinguish between leucine and isoleucine.20,21 Peptides composed of naturally occurring amino acids do not contain chromophores at wavelengths longer than 310 nm.3 Peptide bonds absorb photons in the 190 nm region4 which is well suited for the application of ArF excimer lasers. The aromatic residues of tyrosine, phenylalanine and tryptophan serve as chromophores in the region of the quadrupled Nd:YAG wavelength at 266 nm.11,12,18 Due to the technical difficulties associated with the use of UV optics and laser systems, new strategies to enhance photodissociation cross sections at specific wavelengths were devised in the last decade.22–24 Wilson and Brodbelt demonstrated the sequencing of peptides at 355 nm (tripled Nd:YAG laser) by attaching a UV absorptive chromophore.25 Mass spectra were dominated by a series of either b or y ions depending on the location of the chromophore (N- or C-terminus). Another innovative derivatization method was developed by Julian and co-workers.26 They modified the free thiol form of cysteine in peptides by quinones, resulting in backbone cleavages at the modified site upon UVPD at 266 nm.
Fourier transform ion cyclotron resonance mass spectrometry27 (FT-ICR MS) combined with electrospray ionization28 (ESI) is well suited for the investigation of biomolecules and their fragments.29 Fragment ions can be assigned with high confidence due to the very high mass resolving power of the technique.
The well-studied pentapeptide leucine enkephalin10,14,15,30–32 (LeuEnk, YGGFL) contains two aromatic chromophores at the tyrosine and phenylalanine residues, which enhance photofragmentation yields. A previous study by Dugourd and co-workers was performed at three different wavelengths, 220, 260 and 280 nm,10 which did not provide a complete picture on the wavelength dependence of individual fragment intensities. The Zwier group recently applied cryogenic UV and IR-UV double resonance photofragmentation to elucidate the three-dimensional structure of protonated YGGFL,15 covering a relatively narrow region around the tyrosine absorption from 35200 to 36000 cm−1 with high resolution. In the present study an optical parametric oscillator system is utilised to record a broadband photodissociation spectrum in the 225–300 nm region, which corresponds to 33333 to 44444 cm−1. All fragments are analysed as a function of wavelength to elucidate the potential influence of the electronically excited state on the product distribution. Utilizing the high resolution of the FT-ICR mass spectrometer, the assignment of all fragments was checked, and four LID fragments and one CID fragment had to be re-assigned.
Chemicals were purchased from Sigma Aldrich. Leucine enkephalin was dissolved in 1:1 water/methanol at a concentration of 100 μMol l−1. 0.5% acetic acid was added to the mixture to improve ionization efficiency. The pressure in the hexapole collision cell was kept at around 4 × 10−6 mbar. For LID ions where preselected with the quadrupole, followed by broadband chirp and resonant single frequency excitation of unwanted ions in the ICR cell. In case of collision induced dissociation (CID) only the most abundant isotope of LeuEnk was guided to the collision cell, where they were activated in energetic collisions with argon.
The pressure in the ICR cell was kept in the lower 10−9 mbar region. A full cycle, ion accumulation, laser irradiation and detection, took 1–3 s depending on the applied storage parameters and laser irradiation times. 10 experiment cycles were averaged to provide reliable signal for fragment ions with small intensities.
In the absence of other loss channels, like ion molecule reactions, the photodissociation cross section σ for single photon activation is given by eqn (1)33
(1) |
(2) |
(3) |
(4) |
At the position of the ICR cell, 3.3 m from the exit of the OPO system, the diameter of the laser beam is 11 mm. At thermal cyclotron radii in the range of 100 μm, complete overlap of the laser beam with the ion cloud stored in the ICR cell is in principle possible. However, laser beam misalignment, space charge effects due to a too large number of ions in the cell or unintentional excitation of the selected ion can lead to incomplete overlap. Because of the large number of different fragments, high signal intensity had to be used in the present experiment. As a consequence, 15% of the ions were missed by the laser beam. This was accounted for in the analysis.
Fig. 1 Mass spectrum of leucine enkephalin upon LID at 44444 cm−1. Full sequence coverage for a, b, and y fragments has been achieved. Other intense fragments are discussed in the text. |
Overall, a, b, and y fragments are most abundant, Fig. 2. This is in agreement with other experiments that showed protonation at the amide nitrogen to be favorable for decomposition. Charge directed dissociation then leads to the formation of a, b or y ions.1 Additional fragments are formed by photochemical pathways.24 All those fragments (a1–a4, b1–b4, y1–y4) have been identified with a mass error smaller than 0.4 mDa. c1, c2, c3, x3 and x4 are also formed, with very low intensities for the latter two, and a relatively high mass error for the c3 fragment of 4 mDa. c4, x1, x2 and z-fragments are not detected. There are strong peaks at the nominal mass of x1, x2, z1 and z2 fragments, which are, however, about 0.023 Da off the exact sequence ion mass. A possible explanation would be the exchange of NH2 against O (Δm = 0.0238 Da). Mass errors would be much smaller (<3 mDa), but there are two reasons that such an exchange is not very probable. First, the z1 fragment does not contain any nitrogen, so the required exchange is impossible. Second, a substitution of NH2 for x and z-ions would require the exchange of a backbone nitrogen atom, since x, y, and z-fragments do not contain an amine group. Even though there is no known rule that suppresses the formation of these fragment types, and they have been measured frequently upon LID at 157 nm for other peptides,3 the difference between expected and measured mass is too big. Thus the fragments observed at the nominal mass of x1, x2, z1 and z2 must be reassigned. We succeeded to assign these to internal fragments, former x1 to YGGF int3 + H, former z1 to GGF int4 + H, former x2 to YGGFL int1, and former z2 to YGGF int1, Scheme 2 and Table S2 (ESI†).
Some sequence fragments exhibit notably high signals, Fig. 2. Backbone fragmentation next to Tyr and Phe, the amino acids with aromatic side chains, is a well-known fragmentation pattern.34 Dugourd and co-workers photolyzed the aromatic side chain of phenylalanine and isolated the resulting m/z 465 ion for renewed LID at 220 nm. Without the chromophore, no enhanced fragmentation at the a4 position was observed.10 Thus the high intensities of the a4 and a1 fragments are explained by direct bond cleavage due to the chromophores. Enhanced cleavage near the aromatic residues was also reported upon LID at 266 nm.13 While in LID, the a4 fragment is formed directly, in CID the a4 fragment is mainly formed by consecutive fragmentation of the b4 ion.30 Since in LID, fragmentation occurs mainly around the two chromophores, it is not surprising that fragment ions near the z4 bond do not form frequently (x4, b1, y4, c1, a2, x3). The other sequence ions observed do not show any unexpected behaviour.
Several non-sequence fragments are observed, Scheme 2 and Table S3 (ESI†). The ions at a mass to charge ratio of m/z 91 and 107 correspond to the phenylalanine and tyrosine aromatic side chain fragments, F asc and Y asc, respectively. Ions resulting from loss of the aromatic side chain at m/z 465 (MH+-F asc), 449 (MH+-Y asc) and 448 (MH+-Y asc – H) are also observed. The peaks at m/z 120 and 136 represent the phenylalanine (F im) and tyrosine iminium ions (Y im), respectively. In this case Y im and a1 are identical, since tyrosine is located at the N-terminus of the peptide. Several internal fragments are detected: the ions at m/z 147, 162, 177, 205, 219, 234, 262, 290 and 305 formed at the F chromophore are identified as GF int1, GF int2, GF int3, GGF int1, GGF int2, GGF int3, YGGF int1, YGGF int2, and YGGFL int1, respectively. In addition, sequence fragments with loss of water or ammonia are found, at m/z 375 and at m/z 380. The four remaining heavy fragments are: loss of a hydrogen atom MH+-H, loss of formic acid MH+-HCOOH, loss of NH3 and HCOOH MH+-(NH3 + HCOOH) and loss of water MH+-H2O. The latter three are intense for CID and will be discussed later. All non-sequence fragments are identified with a mass error smaller than 1 mDa. A complete list of fragment ions is provided as ESI.†
Fig. 3 Kinetics of different fragment types: (a) sum of all fragments, (b) a- and b-type fragments, (c) aromatic side chain fragments, (d) most abundant internal fragments. |
Fig. 3b–d show the kinetics of a- and b-type fragments, aromatic side chain fragments, and the most abundant internal fragments, respectively. Most fragment ions already form upon the first laser pulse, which seems to be at odds with the induction period of 5 pulses described above. However, dissociation of the excited ion competes with infrared emission, so that the initially thermalized ions are difficult to dissociate. The internal energy increases with each absorbed photon, increasing the chance for dissociation.
Notable exceptions are the phenylalanine and tyrosine aromatic side chain ions F asc and Y asc, respectively, which are detected for the first time at 5 pulses with a steeply increasing intensity, Fig. 3c. This is consistent with their formation as fragments of the intense a4 and a1 ions, respectively. One might expect that also internal fragments requiring two bond cleavages are formed with a delay, but there is no evidence for such a behaviour in the data. However, one photon seems to be sufficient to photolyze the C–C bond in the intact MH+ precursor ion, since MH+-F asc and MH+-Y asc are observed already at the first laser pulse.
The phenylalanine iminium ion F im dominates the mass spectrum at 40 pulses due to higher order fragmentation, Fig. 3d. This is plausible concerning the preferred fragmentation at the aromatic side chain residues.10
Fig. 3b shows the kinetics of a- and b-type ions relative to the precursor ion. a2, a3 and b1 appear relatively late due to their low intensity, but the overall parallel behaviour to the more abundant sequence ions indicates that they are also formed as primary ions. The small a1 fragment, identical to F im, is abundant since it is located at the N-terminus of the peptide and contains the tyrosine chromophore. The intensities of b2, b3 and a4 decline slowly after 10 laser pulses, indicating further fragmentation. This effect is most pronounced for b3.
Fig. 4 Total photodissociation cross section of protonated leucine enkephalin calculated assuming sequential absorption of 1, 2 or 3 photons. |
The open source software Fityk36 was used to fit the data with Gaussian peaks. Regardless of the number of photons assumed to be absorbed, three local photodissociation maxima are identified for the broad Lb transition at 35452 cm−1, 36696 cm−1 and 38355 cm−1. These may origin from different conformations of the protonated peptide. The maximum of the La transition lies outside the OPO range, and was fixed to 52000 cm−1 to ensure a stable fit. It is very probable that also in this broad peak, different conformers contribute, but there is not sufficient information in the data to extract the relevant peak positions.
Other experiments showed strong fragmentation upon LID at 51813 cm−1 using an ArF excimer laser at 193 nm8,9 and 63694 cm−1 with an F2 excimer at 157 nm5–7 due to the high UV cross section of peptide bonds.4 The Zwier group performed UV and IR spectroscopy on cold protonated YGGFL.15 For the tyrosine chromophore, they found an absorption maximum around 35700 cm−1, which matches our experiments. The literature UV/VIS spectrum of tyrosine in solution37 shows a fast decline of absorption cross sections below 35600 cm−1, similar to our data. The Zwier group did not record a photofragmentation spectrum for the phenylalanine chromophore. UV laser-induced-fluorescence spectroscopy of bare phenylalanine reveals absorption bands in the 37500–37650 cm−1 region.38 This absorption should be present in our data. Due to the compared to tyrosine relatively small absorption cross section of phenylalanine, however, the F-chromophore is probably only a minor contribution. Oh et al. reported in a series of papers12,13,22 that for chromophore-containing peptides, photodissociation is significantly enhanced at 266 nm, i.e. 37600 cm−1, the wavelength of a frequency quadrupled Nd:YAG laser, consistent with our spectrum. No fragmentation is observed below 33500 cm−1.
The measured data points in part deviate significantly from the fit with Gaussian lines. This may be due to unresolved structure in our broad-band experiment with ions thermalized at room temperature. Close to the range limit of the OPO system, however, pulse-to-pulse variations of the laser power increase significantly, which may also contribute to the scattering of the data points around 44000 cm−1, and sequential multi-photon absorption increases this problem.
Fig. 5 shows the branching ratio of characteristic fragments, averaged over three data points in narrow wavelength ranges at strong absorptions. The data for all fragments is available as ESI.† Some fragments are preferred at short wavelengths, like H atom loss (-H), YGGF int1, or YGGFL int1, but also sequence ions like b1. The sequence ions a4, b4 and aromatic side chain loss MH+-Yasc dominate at long wavelengths, but also internal fragments like YGGF int2 may reach their highest branching ratio at longer wavelengths. Others, like F im or y2, have a nearly constant branching ratio throughout the photodissociation spectrum. This indicates that genuine photochemistry with bond cleavages or rearrangements in electronically excited states does play a role in LID of peptides at the investigated frequencies. The clearest case of direct photochemical bond cleavage is MH+-H, which is only detected in the La transition above 41000 cm−1. As pointed out by Dugourd and co-workers,10 it is most likely formed via internal conversion through a conical intersection between a 1ππ* state and a repulsive 1πσ* state of the tyrosine aromatic side chain, as proposed by Sobolewski et al.39 In close analogy to the mechanism of electron capture dissociation,40 the nascent H radical is mobilized and may induce bond cleavages along the peptide backbone, which would seamlessly explain the observation of the modified internal fragments YGGF int3 + H and GGF int4 + H.
However, the fact that at least two, and possibly three photons must be sequentially absorbed before dissociation occurs clearly indicates that rapid internal conversion occurs, in agreement with the well-known photochemical properties of aromatic biomolecules.39 This makes statistical fragmentation from a vibrationally excited molecule in its electronic ground state not merely a plausible scenario, but rather the preferred dissociation pathway. A clear indication for this behaviour is the branching ratio of the b4 fragment, which is significantly enhanced at longer wavelengths, while the a4/b4 ratio, Fig. S4, increases with decreasing wavelengths. This is consistent with the higher photon energy causing additional fragmentation of b4, particularly above 40000 cm−1.
Fig. 7 shows relative intensities of backbone fragments. Compared to LID there are fewer sequence ions, since less energy is deposited per collision event than by photon absorption, and some fragmentation pathways which are possible upon LID will not open. a4 and b4 make up more than 90% of the fragment intensities.
Fig. 7 Relative intensities of backbone fragments compared to the precursor ions for CID of protonated leuEnk, at 13.5 V. For a qualitative comparison of ion signals with LID (Fig. 2) the experiment was performed at the CID voltage, where about the same amount of precursor ions remained. |
Fig. 8 Collision induced dissociation of protonated leucine enkephalin with argon for CID voltages from 0 to 20 V. (a) Most abundant fragments >397 Da, (b) most abundant fragments <397 Da. |
Fig. S5b (ESI†) provides a closer look on a- and b-type ions and Fig. S5c (ESI†) on internal fragments. a1, a2 and a3 only form for CID voltages above 10 V, b3 and b2 form a little earlier, starting around 8 V. There are fewer internal fragments in CID than in LID. These form at voltages above 7 V. Since one additional bond has to be ruptured, a higher energy transfer into the precursor is required. The two iminium ions F im and Y im start at 10 V and behave almost identically.
CID of the protonated leucine enkephalin by RF excitation in a hexapole collision cell leads to higher mass fragment ions than LID. The a4 and b4 fragments are most abundant and full sequence coverage is obtained for a and y ions. The loss of water, HCOOH or HCOOH together with NH3 are clearly visible. The difference between CID and LID lies in the different internal energy content of the molecular ion. In CID, multiple collisions add small amounts of energy, thus slowly approaching energy thresholds for dissociation. This favours dissociation pathways with low-lying tight transition states. In LID in the studied wavelength region, 4.3–5.5 eV are added upon absorption of a single photon. The overall larger energy content of UV excited molecular ions favours dissociation via high-lying lose transition states.
Footnote |
† Electronic supplementary information (ESI) available: Fig. S1–S6 and Tables S1–S6 as described in the text. See DOI: 10.1039/c6cp08436b |
This journal is © the Owner Societies 2017 |