Radical-Initiated Photodissociation at Tryptophan Residues within Intact Proteins in the Gas Phase

Abstract

Controlling selective fragmentation at particular residues in the gas phase could greatly improve our ability to characterize intact proteins by mass spectrometry and reveal proteoforms crucial to human biology. However, the chemical homogeneity and size of proteins frustrates characterization because fragmentation inevitably leads to thousands of product ions and highly complex spectra that resist complete interpretation. Herein, we report a method for the selective, photochemical cleavage of whole proteins in the gas phase via the photolysis of alpha-peptidyl radicals. The activation process results in predictable and selective dissociation of the peptide bond N-terminal to the initiating radical. Alpha peptidyl radicals are readily introduced in a residue specific manner via the nitrosation of tryptophan, which enables an easily triggered side chain loss. This approach affords highly controlled fragmentation of intact proteins in the gas phase, producing results analogous to those obtained by tryptic digests in solution. The smaller fragment ions produced by selective dissociation are abundant and can be further analyzed to facilitate characterization by providing more detailed information about sequence and modification sites.

Protein characterization by mass spectrometry (MS) is the core enabling technology of modern proteomics. Due to their large size and complexity, this process benefits from the selective cleavage of proteins into tractable fragments for analysis. In the bottom-up approach, this cleavage is accomplished through enzymatic protease digestion in solution prior to MS.¹ Despite the utility of this approach, proteoform-level information is inevitably degraded after digestion.² The alternative top-down MS strategy retains proteoform integrity by keeping proteins intact while introduced into the gas phase.³ Unfortunately, there are no selective options when subsequent fragmentation is employed for additional proteoform characterization, significantly complicating data collection and analysis in top-down proteomics. Fred McLafferty elegantly explained the nature of this challenge with a well-known puzzle in which one light gold coin must be identified among many.⁴ The simplest solution is clearly not to weigh each coin individually, but rather to iteratively divide the collection of coins in half, weighing each to see which fraction contains the fraudulent coin. Unfortunately, selectively cutting proteins in half has proven to be a formidable challenge due to the chemical homogeneity and robust nature of peptide backbones. Energy sufficient to break one bond is likely to cleave many others. Hence, whether collisions,^5–7 photons,^8–10 or electrons^11,12 are utilized, stochastic dissociation producing hundreds of fragments is observed.

A potential solution to this problem leverages the side chain identities, which are not chemically homogeneous, to introduce chemical functionality at specific residues. Human biology makes extensive use of this approach via posttranslational modification¹³ and more recently, chemists have made great strides in residue-specific modification strategies.^14–16 Although much of this work has focused on other applications, side chain modifications could also be used to facilitate selective fragmentation in the gas phase (Figure 1a). Along these lines, the Julian lab has previously developed modifications allowing radicals to be created at specific residues,¹⁷ but subsequent collisional activation led to significant radical migration, which ultimately sabotaged fragmentation selectivity. We have recently discovered that alpha-peptidyl radicals absorb at 266 nm, resulting in selective dissociation of the peptide bond N-terminal to the alpha radical (Figure 1b). No other appreciable bond dissociation is observed during this process, and the fragment ions so produced are abundant and amenable to further nonselective dissociation in subsequent MS³ experiments (Figure 1c).

Figure 1.

a) Enabling selective dissociation requires chemical modification of the protein to introduce a weak point. After modification, collisions or photons create alpha radicals. b) Photoactivation of the alpha radical yields extremely selective dissociation of the N-terminal peptide bond, and only that bond. The initially formed b+1 ion is unstable and spontaneously decays into an a+1 ion. c) The a/y fragments can be analyzed further by MS³ experiments to localize PTMs or confirm sequence.

The selective dissociation illustrated in Figure 1 requires the introduction of an alpha radical at a specific location. We have initially accomplished this via N-nitrosation of tryptophan (Trp) in an acidified solution of sodium nitrite,¹⁸ followed by mild collisional activation in the gas phase.¹⁹ The nitrosation chemistry is selective against most other residues except for cysteine (Cys), which must be covalently blocked if present.²⁰ The N–N bond between the Trp indole side chain and NO is thermally labile and cleaves homolytically to yield an indole radical that subsequently extrudes the entire Trp side chain (Scheme 1a).

Scheme 1.

Alpha Radical Generation and Theory

Due to their resonance stabilization, alpha-peptidyl radicals are thermodynamically preferred to most other protein radicals and do not readily migrate.²¹ The delocalized structure responsible for this stability (termed captodative or ambident)²² is the source of the UV photoactivity we observe. Scheme 1b (inset) shows the singly occupied molecular orbital of a model alpha-peptidyl radical on triglycine (DFT, ωB97X-D/cc-pVTZ). Time-dependent DFT calculations on this species reveal a 244 nm absorption resulting in excitation to the D₅ state (Scheme 1b) which has exothermic access to a dissociative state associated with N-terminal peptide bond cleavage as in Figure 1b. These states are not available for the closed-shell species.

To provide an example of the chemistry illustrated in Figure 1, we conducted experiments with cytochrome c (cytc), a ∼ 12 kDa heme protein comprised of 104 residues including a single Trp at position 59. Following nitrosation, electrospray ionization under mildly activating conditions produces favored loss of both NO and the Trp side chain (Scheme 1a), yielding an abundant alpha radical in the MS¹ spectrum (Figure 2a). Isolation of the alpha radical in the 17+ charge state, followed by collisional activation yields the spectrum in Figure 2b. Interestingly, backbone fragmentation occurs with significant selectivity. The alpha radical preferentially migrates to beta position of Thr58 and then cleaves the backbone by radical-directed dissociation (RDD) through previously reported mechanisms¹⁸ to yield the z₄₇, a₅₇, c₅₇, and a₅₈ ions. Additional low intensity fragments are also produced by RDD (such as z₄₂), along with numerous b/y ions generated competitively through the mobile-proton mechanism.²³

Figure 2.

a) Efficient production of alpha radicals in MS.¹ The charge states for protonated precursor ions are labeled. b) decharged²⁴ CID spectrum for the 17+ alpha radical. c) decharged spectrum for dissociation of the 17+ ion following 266 nm PD for 8 ms. The protein is cleaved into two large fragments. d) decharged MS³ CID spectrum for the internal imine product. In the sequence representations, fragments are labeled using conventional line marks, which are bolded for ions with >50% intensity and light for <10% intensity. Precursor is annotated with a blue arrow.

In contrast, excitation of the same 17+ alpha radical ion for 8 ms with 266 nm light yields the spectrum shown in Figure 2c. Despite the hundreds of bonds present in cytc, only two significant fragments are produced, which both originate from cleavage of the peptide bond N-terminal to the alpha radical. Regardless of charge state, over 95% of dissociation occurs at the predicted site (Table S1), confirming that alpha radicals are not prone to migration. We term this method for inducing selective bond cleavage radical-initiated photodissociation (RIPD).

Although the majority of the precursor ions can be converted into the fragment ions shown in Figure 2c by increasing the photoactivation time, some residual intensity is observed at the same nominal mass as the original precursor even at much longer activation times. However, if this residual peak is subjected to collisional activation, the results reveal that subtle dissociation has occurred in the form of hydrogen atom loss from the original precursor. This loss is easily overlooked because the resulting one Dalton mass shift is difficult to confidently identify for intact proteins due to their wide isotopic distributions. However, the loss is easily rationalized by an alternate beta-scission pathway that is also photoactivated (see Scheme S1a). The H-loss produces an internal imine that is susceptible to attack and dissociation by the mechanism shown in Scheme S1b, producing highly unusual (and therefore confirmatory) z₄₆-3 and c₅₈ complement ion pairs (Figure 2d). In addition, competitive mobile proton derived dissociation is observed with significant overlap relative to the b/y ions observed in Figure 2b. Thus, dissociation with varying degrees of selectivity is observed for three different pathways following creation of an alpha radical.

To further illustrate the selectivity of RIPD, we examined proteins containing two and three Trp residues (Figure 3, the double radical species is photoactivated for both proteins). To facilitate interpretation of the results, the sequence regions delineated by Trp residues are highlighted by different colors. The RIPD fragments are then marked with color-coordinated sequence regions. Importantly, selective fragmentation at Trp residues is faithfully maintained for both proteins. Indeed, both the expected terminal and internal ions (i.e., those that result from cleaving the backbone twice) are observed. Furthermore, all the complement ion pairs are observed for every dissociation point, meaning that no ions were lost to unknown dissociation pathways, and all ions are assigned. This represents an important distinction between RIPD and stochastic dissociation pathways where complement ions are frequently lost and many of the observed peaks are unassignable.

Figure 3.

16 ms RIPD spectrum for the 15+ double alpha-radical precursor of a) ProtW2 (122 residues) and b) ProtW3 (126 residues). RIPD fragments are labeled by their N- and C-terminal residue numbers (both numbered from the N-terminal side). RIPD fragments are also color-coded into the relevant sequence regions delineated by Trp residues. Line marks specify the type of ion and cleavage site. # secondary ammonia loss, ← secondary RIPD of a+1 ion, downward arrow, precursor ion.

It is noticeable that some peaks in RIPD spectra are present as doublets. Importantly, these ions all correspond to secondary losses of small molecules from the terminal ends created by RIPD backbone dissociation. The mechanism of RIPD produces a radical a+1 ion and a terminal imine y-2 ion. The terminal imine group is susceptible to attack by lysine (Scheme S1c). Attack of terminal imines ultimately results in loss of ammonia. For peptides where lysine (or another suitable side chain) is in close proximity to the imine, loss of ammonia from the y-2 fragment is often observed and can be abundant. For the a+1 radical, certain C-terminal residues will favor an additional small molecule loss following backbone dissociation. For example, in cytc, the C-terminal residue is threonine, which favors loss of water following formation of the a-ion (Scheme S2a). Although these losses lead to additional dispersion of signal, they also provide information about sequence or fragment-ion type that can increase the confidence of assignments.

The selectivity afforded by RIPD also facilitates subsequent analysis in MS³ experiments because the RIPD fragment ions are produced in high abundance. MS³ dissociation of RIPD fragments has several advantages over direct dissociation of the precursor, which we demonstrate with experiments on chemically acetylated cytc. This procedure yields a highly complex mixture of one to four acetylations distributed throughout the sequence, presumably favoring modification at lysine residues. In other words, there are likely dozens (if not hundreds) of proteoforms present in the resulting mixture, a tremendous challenge for characterization by any method. Focusing on the singly acetylated peak, we started with the intact protein and used traditional methods such as electron-transfer dissociation (ETD, Figure 4a bottom) and higher-energy collisional dissociation (HCD, Figure 4a top) to attempt to localize modifications (to simplify data interpretation we assume that Lys modification is dominant). For traditional top-down, two regions that likely contain an acetylated Lys residue are identified from the combination of both ETD and HCD data as indicated by the green boxes in Figure 4b. Although it is certain that additional acetylated Lys residues exist between these regions, it is not possible to localize additional sites because any mass shifted fragments observed at more interior sequence locations can also be attributed to the same Lys residues already indicated by the green boxes.

Figure 4.

a) Top-down ETD (upper) and HCD (lower) spectra for 17+ singly acetylated cytc. b) Combined sequence coverage and acetylated regions identified from intact cytc. c) RIPD spectrum for the +17 ion, followed by MS³ ETD (upper) and HCD (lower) on the singly acetylated a₅₈+1-H₂O (tan) and y₄₆-2 (blue) ions. d) Combined sequence coverage and acetylation site identification, including two additional sites. Bold sequence ticks indicate acetylated fragments. Green boxes indicate localization of the acetylation site.

In contrast, in Figure 4c, RIPD is again used to selectively cleave the singly acetylated protein into two pieces. The results for ETD and HCD on each fragment are shown in the spectra below the corresponding RIPD fragment. The combined HCD/ETD sequence coverage and acetylation localization are shown in Figure 4d, with shading to distinguish RIPD fragments. Importantly, the initial selective dissociation creates new N/C termini, which allows the localization of two additional acetylation sites as indicated by the interior green boxes. N/C-terminal acetylation sites are also identifiable with this approach and are highly consistent with the sites identified by the traditional top-down approach. It is important to emphasize that the two additional interior acetylation sites cannot be identified utilizing the terminal ions generated by any traditional dissociation method, making them only accessible with RIPD. Furthermore, if we consider the cumulative sequence coverage between the two methods, they are comparable with a slight increase observed for the RIPD MS³ data (72% vs 78%). Importantly, because the RIPD fragments are significantly smaller than the original protein, the stochastic dissociation data in Figure 4c is less complex and easier to assign than the data in Figure 4a, which may account for the gain in sequence coverage.

In summary, we present a method for selectively fragmenting intact proteins in the gas phase. The dissociation is highly predictable, yields both complement ion pairs, and is amenable to subsequent MSⁿ activation steps. Importantly, RIPD creates new termini, which facilitates deeper identification of PTMs and proteoforms. This tool provides great promise for expanding the reach of top-down proteomic.

Supplementary Material

ASSOCIATED CONTENT

Supporting Information

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacs.5c07784.

Additional experimental details, materials, and methods, including dissociation mechanisms and RIPD yields. (PDF)