Integration of high resolution mass spectrometry with cryogenic ion vibrational spectroscopy

Abstract

We describe an instrumental configuration for the structural characterization of fragment ions generated by collisional dissociation of peptide ions in the typical MS² scheme widely used for peptide sequencing. Structures are determined by comparing the vibrational band patterns displayed by cryogenically cooled ions with calculated spectra for candidate structural isomers. These spectra were obtained in a linear action mode by photodissociation of weakly bound D₂ molecules. This is accomplished by interfacing a Thermo Fisher Scientific Orbitrap Velos Pro to a cryogenic, triple focusing time-of-flight photofragmentation mass spectrometer (the Yale TOF spectrometer). The interface involves replacement of the Orbitrap’s higher-energy collisional dissociation cell with a voltage-gated aperture that maintains the commercial instrument’s standard capabilities while enabling bi-directional transfer of ions between the high-resolution FT analyzer and external ion sources. The performance of this hybrid instrument is demonstrated by its application to the a₁, y₁ and z₁ fragment ions generated by CID of a prototypical dipeptide precursor, protonated L-phenylalanyl-L-tyrosine (H⁺-Phe-Tyr-OH or FY-H⁺). The structure of the unusual z₁ ion, nominally formed after NH₃ is ejected from the protonated tyrosine (y₁) product, is tentatively identified as a cyclopropane-based product.

Graphical Abstract

I. Introduction

Structural characterization of mass-selected ions with infrared (IR) vibrational spectroscopy is an increasingly valuable secondary analysis tool for mass spectrometry. [1–6] This technique is most powerful when carried out with cryogenically cooled ions, either in a mass messenger or “tagging” mode [1, 7–9] or in a tag-free approach based on two color, IR-IR or IR-ultraviolet/visible double resonance excitation.[10, 11] In that scheme, the spectra are obtained for ions frozen close to their global (or sometimes local) minima in a linear action mode such that they can be directly compared with the patterns calculated for candidate structures using increasingly sophisticated electronic structure software packages (e.g., Gaussian, GAMESS, Turbomole, Schrödinger, etc.). [1, 3–5] Ions are typically cooled with a buffer gas in a cryogenic (~4 K) radio frequency (RF) ion trap [12–21], and IR spectra are obtained by photodissociation in a tandem mass spectrometry (MS) mode. Specific implementations range from double focusing time-of-flight (TOF) [22–24] to photoexcitation in mass-selective traps followed by mass analysis. Secondary mass analysis can then be carried out with a variety of techniques such as time-of-flight, mass selection with a quadrupole mass filter, and multiplexing Fourier transform (FT) schemes based on electrostatic (Orbitrap [25, 26]) or magnetic (ion cyclotron resonance (ICR) [27]) ion traps. Many of these hybrid instruments are based on customized commercial platforms and retain their original ion processing capabilities. As such, characterization of ion structures with cold ion spectroscopy brings a powerful addition to the arsenal of secondary analysis methods already integrated into these instruments. Here we describe our integration of cryogenic vibrational spectroscopy into the Thermo Fisher Orbitrap Velos Pro (hereafter denoted “Velos”) in an arrangement where ions processed using its secondary analysis capabilities are structurally characterized by analyzing their vibrational spectra. We demonstrate the performance of this hybrid platform by applying it to several fragment ions generated by collision-induced dissociation (CID) using MS/MS (MS²) analysis of a prototypical dipeptide cation.

We focus on the structural identification of CID fragment ions because this method is widely available in commercial instruments and provides a powerful means for characterization of complex chemical samples. Particularly important applications in this regard are proteomics [28], de novo sequencing [29], and metabolite analysis [30]. Peptide analysis typically relies on previous knowledge of the fragmentation pathways to identify cleavage points. The resulting CID fragments are then catalogued by the formation of a_x, b_x and y_x ions [29] which often involve chemical rearrangements (e.g., proton migration). The utility of vibrational spectroscopy in analyzing the mechanisms that drive fragmentation has recently been demonstrated in an arrangement where infrared multiphoton dissociation (IRMPD) spectra of room temperature ions were obtained using a modified Bruker Paul trap [31]. In that case, the structures of the fragment ions generated by electron-transfer dissociation (ETD) from a five-residue polypeptide ([AAHAR+2H]²⁺) were established by comparison with calculated band patterns at the harmonic level for candidate structures.[28, 32].

Although it represents a significant advance, the IRMPD approach [33] suffers from the important complication that the resulting spectra reflect highly non-linear (> 15 photons) excitation processes and are obtained with (~300 K) ions whose internal energies are far above their global minima. Here we address this limitation by integrating a cryogenic ion vibrational spectroscopy instrument with a commercial, high-resolution tandem mass spectrometer. We demonstrate the hybrid instrument’s performance through the structural characterization of fragment ions generated by the Velos’s routine MS² operation. For this purpose, we focus on a simple dipeptide (protonated L-phenylalanyl-L-tyrosine, abbreviated hereafter as FY-H⁺) using CID in the linear quadrupole trap (LTQ) section of the instrument to obtain the fragmentation pattern as is usual in the course of sequencing by MS². The fragment ions are first analyzed using the high-resolution Orbitrap capability of the instrument, then passed on to the cryogenic ion spectrometer for further characterization by IR action spectroscopy. In this way, we effectively realize an “add-on” analysis tool that fully maintains the inherent functionality of the commercial instrument; the hybrid instrument makes possible the structural characterization of MSⁿ products or intermediates extracted directly from the fragmentation cell while leaving the essential capabilities of the commercial device intact.

II. Overview of the instrument

Fig. 1 presents a schematic diagram of the Yale hybrid instrument. The interface was achieved by replacing the Orbitrap’s C-trap exit aperture and subsequent higher-energy collisional dissociation (HCD) trap with a custom C-trap interface aperture (CTIA) and octopole ion guide to lead ions out of the commercial instrument and into the Yale triple focusing photofragmentation mass spectrometer (see photographs in Fig. S1) [11]. We implement external control of the CTIA to gate the flow of ions according to several protocols. The timing of the external components is slaved to the aperture control generated by the Velos software (see triggering scheme in Fig. S2). External control of the CTIA to a repelling mode was optimized to maintain the high resolution performance of the Orbitrap mass analyzer (see comparison in Fig. S3). In this way, routine MSⁿ measurements can be carried out on target ions with both LTQ and FT modes to allow traditional structural analysis. When one is interested in further characterizing an ion using cryogenic vibrational spectroscopy, however, an ion packet created in the LTQ part of the instrument is launched toward the C-trap with its trapping buffer gas turned off. Lowering the direct current (DC) voltage on the CTIA then allows the ion packet to pass through the C-trap to the octopole guides leading them out of the instrument and into the Yale spectrometer. In the current configuration, the drift path length between the two instruments is about 1.5 m due to the physical dimensions of the two. The transition section includes several control apertures and a gate valve [34] to enable isolation of the Orbitrap and Yale TOF vacuum envelopes, as well as a bent octopole ion guide to match the angles of the two instruments.

Figure 1.

Scheme of the hybrid instrument that integrates a high-resolution Thermo Fisher Scientific Orbitrap Velos mass spectrometer with a custom-built cryogenic, triple focusing photofragmentation mass spectrometer. Components of the commercial Velos Pro are grouped inside red dashed lines and include a commericial Thermo Fisher ESI source and LTQ ion trap, which are both part of the embedded photograph but not explicitly shown. The external ESI and linear ion guides provide an independent ion source for species not readily prepared in the commercial instrument (e.g., M⁺(H₂O)_n, etc.).

The pulse cycle of the Yale spectrometer is triggered by the Velos timing pulse (conveniently intercepted at BNC connector J6722 on the RF control board) linked to the injection of C-trap ions into the Orbitrap (see timing sequence in Fig. S2). This serves to synchronize the ion packet leaving the LTQ component of the Velos with the pulsed valve that introduces buffer gas into the cryogenically cooled Paul trap at the start of the Yale triple-TOF photofragmentation spectrometer. The (~1 ms) buffer gas pulse is introduced about 20 ms before the arrival of the LTQ ion packet. The buffer gas mixture (10% D₂ in He) is optimized to enable attachment of weakly bound adducts to the ions of interest. The vibrational predissociation spectra of these “tagged” ions are then obtained with the TOF-based, triple focusing photofragmentation mass spectrometer described previously [11]. Technical details of the interface are included in the supplementary materials (Section 1). We note that the Yale spectrometer includes its own electrospray ionization source, and the ion transfer facility of the hybrid instrument is bidirectional. Thus, it is also possible to perform high resolution (Orbitrap) mass spectrometry on the ions generated using the Yale electrospray source. A comparison of the peak shapes generated by the Orbitrap analyzer at maximum resolution (100,000 Δm/m at 200 m/z) for various configurations is presented in Fig. S3. This dual use capability is achieved by introducing the Velos ions to the Yale TOF with a DC 90° turning quadrupole (see schematic in Fig. 1). In this way, setting the turning quadrupole to transmission mode enables spectroscopic interrogation of ions from the Yale source in the mode utilized for many previous studies [1].

III. Cryogenic vibrational spectra of MS² fragments and comparison with harmonic calculations of candidate structures

FY-H⁺ was isolated and fragmented in the LTQ section of the Orbitrap Velos Pro with normalized collision energy (NCE) of 29 and 10 ms activation time. The resulting CID fragment ions were transferred into the cryogenic Paul trap for cooling and D₂ tagging. The ions were then mass separated using the Yale TOF photofragmentation spectrometer. Additional sample preparation details are included in the SI.

A typical FY-H⁺ fragmentation mass spectrum is shown in Fig. 2 alongside molecular composition assignments of the fragmentation products derived from common fragmentation pathways (a₁, y₁, a₂) [28]. After the masses of the parent and fragment ions were confirmed using the high resolution capability of the Orbitrap, they were sent on to the Yale TOF spectrometer for structural analysis with cryogenic vibrational spectroscopy. Scans of the vibrational spectra were acquired in two different spectral regions (900–2000 cm⁻¹ and 2800 – 3800 cm⁻¹), with each scan taking 15 minutes. Overall signal to noise was improved by averaging over many individual scans. Depending on the ion abundance, the acquisition times for the results presented in Fig. 3 ranged from 2 hours for the most abundant ion (FY-H⁺) to 8 hours for the least abundant ion (the z₁ fragment). The structures of most of the fragments are anticipated based on precedent: the a₁ ion likely occurs via formation of the iminium ion, the a₂ ion can be generated by the loss of neutral formic acid, and the y₁ fragment presumably forms after cleavage of the amide bond.

Figure 2.

CID mass spectrum of the FY-H⁺ parent ion (329 m/z, black) displaying four major fragmentation pathways: a₁ (120 m/z, green), y₁ (182 m/z, purple), a₂ (283 m/z, brown), and a weak z₁ product (165 m/z, blue). Molecular compositions were generated using the ThermoQual software. See Table S1 for simulated masses and deviations.

Figure 3.

Photofragmentation spectra of the D₂-tagged FY-H⁺ parent (A), a₁ (D) and y₁ (G) fragments. Calculated spectra, derived from electronic structure calculations of the fragments’ structural minima, are inverted underneath the corresponding spectra in B, E, and H demonstrating good agreement between experiment and theory. Chemical structures for each species are shown in C, F, and I. Labeled peaks in A, D, and G can be found in Table S2. Labels for assignment are colored black for CC and CH modes, blue for N modes, and red for O modes.

The D₂-tagged spectra of the FY-H⁺, a₁ and y₁ ions are shown in Fig. 3. Each displays a series of sharp peaks throughout the range 900–4000 cm⁻¹. Density functional theory (DFT) calculations (B3LYP/6–311G++(d,p)) using Gaussian 09 [35] converged on structures that are consistent with those expected for the FY-H⁺ and a₁ ions, while the y₁ spectrum is very similar to that reported in detail for protonated tyrosine [36]. The multiplet structure in the y₁ spectrum (Y₃, Y₄ and Y₅) indicates that this ion occurs in several different conformers upon cooling that have been spectroscopically isolated using IR/UV double resonance methods.[36] The FY-H⁺ and a₁ structures at the right of Fig. 3 were the lowest in energy among other locally stable minima. The identification of the y₁ fragment ion as protonated tyrosine through its characteristic cryogenic vibrational spectrum highlights one of the powerful features of this method. Specifically, this is possible because the spectrum is obtained using linear action and applies to the vibrationally cold ions. As such, these spectra are directly comparable with those reported earlier even though the ions may have been prepared under very different conditions. Consequently, a library of such spectra [37, 38] would create a more accurate fingerprint of the molecular structures because they are not subject to variations that are intrinsic to the MS² approach. These variations include different protocols for pressure, collision gas, collision energy, as well as whether dissociation is carried out under single or multiple collision regimes [39, 40].

The predicted (scaled, see SI for scaling protocols) harmonic spectra for these structures are presented as the inverted traces in Fig. 3. The vast majority of the observed transitions are indeed recovered at the simple harmonic level with remarkably similar intensity profiles. This agreement provides detailed structural identifications of the fragments that are indeed consistent with accepted fragmentation patterns. For example, protonated tyrosine (the y₁ fragment) is nominally formed by cleavage of the amide bond (purple arrow in Fig. 3c), and thus retains the OH stretches (Y₁ and Y₂) that are present in the FY-H⁺ spectrum (FY₁ and FY₂). Similarly, both FY-H⁺ and y1 spectra contain features arising from their other shared structural motifs, such as the acid C-O stretches (FY₇ and Y₆), O-H bends (FY_12&13, Y_11&12), and tyrosine C-OH stretch (FY₁₁, Y₁₀). Furthermore, comparison of the observed and calculated spectra of FY-H⁺ allows us to establish that protonation occurs at the N terminus. The NH groups of this – NH₃⁺ moiety form linkages with the amide C=O group and the π clouds of the tyrosine and phenylalanine side chains. The NH stretches account for bands (FY₄, FY₅, Y₄, Y₅), which occur ~80 cm⁻¹ below those found in an isolated –NH₃⁺ motif [23]. On the other hand, the a1 fragment [41], is generated by break up at the green arrow in Fig. 3c (with proton migration), leading to a new strong band (A₄) arising from the C=N stretch in the iminium functionality with the higher energy doublet (A₁ and A₂) derived from the symmetric and asymmetric stretches of the NH₂ group. Note that these bands are also red-shifted (~40 cm⁻¹ and 100 cm⁻¹ for the asymmetric and symmetric stretches, respectively) relative to their calculated positions in the higher energy rotamer where the phenyl group is displaced away from the NH₂ motif (Figs. S4–S6).

The weak z₁ fragment is interesting in that it corresponds to the mass expected for heterolytic cleavage at the strong C_α-N bond (blue arrow in Fig. 3c), which is an unusual pathway for CID of protonated peptides [42]. This yields a closed shell carbocation with several chemically distinct rearrangement pathways, for which there is no strong precedent upon which to formulate a structural hypothesis. It is clear from the D₂-tagged spectrum (Fig. 4a), however, that the observed fragment features two free OH groups (Z₁ and Z₂) and retains the acid C=O (Z₃). We therefore carried out an extensive search for possible structures that retain these functionalities, starting from different initial conformations and applying geometric constraints to either promote or prevent chemical rearrangements such as ring formation. Details about the constraints are included in the SI (section 2). Three low lying structural isomers were identified (hereafter denoted isomers I, II and III) that are indicated in the inserts in Fig. 4 with rotating pdf images in Figs. S9–S11. These isomers differ in the structure of the hydrocarbon backbone, with the lowest energy isomer (I) based on a hydroxytropylium scaffold, while the two higher energy isomers feature a 1,4-cyclohexadiene hydrocarbon core. The strong Z₄ feature is not present in the predicted spectrum of I, but is recovered in the harmonic prediction for the two higher energy isomers as due primarily to the CO displacement in the C=OH⁺ functionality. Finally, the telltale (Z₉ and Z₁₀) bands, which dominate the low energy region of the observed spectrum, are only present in the highest energy isomer (III). In fact, the overall agreement with the observed spectrum is compelling for the assignment of the observed fragment to this interesting spiro-bicyclic scaffold with a cyclopropane motif. Further investigation of this species is clearly warranted to verify this hypothesis, and if confirmed to explore the pathways for isomer interconversion. This observation highlights how cryogenic spectroscopy can reveal unexpected chemical rearrangements in minor species present in CID analysis but not generally utilized for sequencing.

Figure 4.

Photofragmentation spectrum of the z₁ fragment (A) and calculated spectra corresponding to candidate structures involving formation of a cyclopropane-containing spiro compound (B), proton migration (C), and formation of a hydroxytropylium cation (D). Calculated spectra are derived from DFT calculations of the species’ structural minima. Labeled peaks in A can be found in Table S3.

IV. Conclusion and outlook

We described the design, operation and demonstration of a hybrid instrument that combines a commercial high resolution mass spectrometer with a custom-built, cryogenic ion photofragmentation infrared spectrometer. This scheme adds bond-specific structural information to the MSⁿ scheme widely used for compound identification. The vibrational predissociation spectra of the ~20 K fragment ions formed by CID in the LTQ trap of a commercial Thermo Fisher Scientific Orbitrap Velos Pro mass spectrometer were obtained by photodissociation of weakly bound D₂ adducts. The resulting linear action spectra reveal well resolved band patterns that yield structural identifications of the products by comparison with harmonic calculations for candidates. The hybrid instrument also enables high resolution mass analysis of ions from an external ion source using the Orbitrap analyzer capability of the Velos platform. The latter capability provides a powerful way to monitor photofragments generated in the vibrational spectroscopy part of the instrument, a next generation adaptation that is currently being pursued at Yale.