Reactive species driven oxidative modifications of peptides—Tracing physical plasma liquid chemistry

The 2019 Nobel Prize in Physiology and Medicine was awarded to the discovery that the degradation of the protein hypoxia-inducible factor 1α is accelerated by its hydroxylation under normal oxygen levels, while in lack of oxygen—such as stroke—the protein is not hydroxylated and accumulates, launching physiological countermeasures.^1,2 This example illustrates the important role of (oxidative) post-translational modifications (oxPTMs) in the regulation, protein functionality, and relevance for most cellular signaling pathways.^3,4 More than 200 different (ox)PTMs are known, dominated by phosphorylation, acetylation, and ubiquitinylation with a massive impact on a proteins function and lifetime.⁵ Kinase-driven protein phosphorylation⁶ activates or prevents downstream signals⁷ and was found relevant in the response of mammalian cells toward cold physical plasma treatment (CAP).⁸ 

The role of oxPTMs^9,10 on cellular signaling is in the focus of research activities, fostered by the improvement of high-resolution mass spectrometry instrumentation and workflows.^11–13 For a long time, (ox)PTMs had been considered as signs of destruction only, such as aging processes.¹⁴ A complex regulation network is currently being revealed, it has been proven that downregulation (distress) and stimulating effects (eustress) coexist,^15,16 and the cellular processes are influenced by various modifications at the same time. Apoptosis, for example, is regulated by S-nitrosylation,¹⁷ N-acetylation,¹⁸ and N-myristoylation.¹⁹ 

Controlled production of reactive oxygen and reactive nitrogen species (ROS/RNS) is also found in cold physical plasma, generating short- and long-lived reactive species, which are capable of reacting with most biomolecules and representing an emerging field of redox therapy. Due to its unique properties, physical plasma serves many applications. For long, it has been used for decontamination²⁰ or the refinement of various surfaces.²¹ Recently, physical plasma treatments have been proven to be useful in medical applications, e.g., the treatment of chronic wounds^22,23 or cancers.^24–27 However, the molecular mechanisms are not yet fully understood. It remains elusive, i.e., which species penetrate sufficiently in biological systems and which of those are (chemically) active to trigger downstream signals. While the long-lived plasma-derived species as H₂O₂, nitrite, and nitrate can be easily quantified are used as fundamental markers for CAP treatment, the short-lived species are difficult to access. However, they appear to be of significant importance.^28–35 Given their high reactivity, a penetration seems unlikely. Alternatively, it may be hypothesized that exposure of biomolecules to short-lived species yields secondary products, which have biological effects on themselves and ultimately modulate the physiological processes in a cell or tissue.

A number of reports indicate that amino acids in proteins are excellent targets for plasma-derived reactive species.^36–40 In addition, their regulation via physiological (ox)PTMs offers an excellent intersection with physical plasma. The aim of this work is, therefore, to determine the impact of plasma-derived reactive species on model peptides, focusing on the formation of artificial non-enzymatic post-translational modifications. To gain an overview of a variety of gas phase chemistries, the argon-driven kINPen was compared to the helium-driven COST-Jet⁴¹ while modulating the discharge parameters including the gas phase composition. To investigate these modifications, high-resolution mass spectrometry coupled to nanoflow liquid chromatography and subsequent bioinformatical analysis were used.

A peptide library of ten peptides with a length of ten amino acids each was designed and synthesized (ProteoGenix, Schiltigheim, France) in such a way, that the 20 major proteinogenic amino acids occur with equal incidence (Fig. 1 and Table S1 in the supplementary material). In addition, the peptide sequence was designed to locate each amino acid at the N-terminus, the C-terminus, and the center region and to have as many different neighboring amino acids as possible, exemplified for alanine (A, red) and tryptophan (W, blue) in Fig. 1.

The peptides were dissolved in double distilled water (Merck Millipore, Darmstadt, Germany) or phosphate buffered saline (PBS, Sigma Aldrich, Deisenhofen, Germany) at a concentration of 0.1 mg/ml. Both solvents were degassed by bubbling with argon gas for at least 30 min to reduce background reactions. The treatments with cold atmospheric pressure plasma were performed with an argon plasma jet (kINPen, neoplas tools GmbH, Greifswald, Germany) and a helium plasma jet (COST-Jet, a standard device identified by the European COST action MP 1101 “Biomedical Applications of Atmospheric Pressure Plasma Technology”). The kINPen⁴² [Fig. 2(b)] consists of a grounded ring electrode enclosing a ceramic capillary, where a powered central rod electrode is located inside (2–6 kV_pp at 1.1 MHz). A 3000 standard cubic centimeters per minute (SCCM) flow of argon (Air Liquide, 99.999%) served as the main feed gas. For some treatments, 0.5/1% of the feed gas was replaced by oxygen (15 SCCM) or by oxygen + nitrogen (15 sccm each). The COST-Jet [Fig. 2(a)] consists of two 1 mm thick metal plate electrodes leaving a 1 mm gap, where the plasma is ignited. The capacitively coupled electrodes are driven by an AC voltage at 13.56 MHz. The dissipated power was held constant at 330 mW by using a Tektronix DPO 4104 Digital Phosphor Oscilloscope in accordance with Ref. 41. The helium feed gas flow was kept at 1000 SCCM and if desired, enriched with 0.5% oxygen (5 SCCM), or 0.5% oxygen + 0.5% nitrogen (5 SCCM each).

750 μl of each peptide solution (water/PBS) was placed in 24-well plates and treated 15 or 60 s with a nozzle to a liquid distance of 4 mm (COST-Jet) or 9 mm (kINPen) for each gas mix (feed gas + O₂ or O₂/N₂). For indirect treatments, the procedure was adapted: only the solvents were plasma-treated and the peptides were added 1 min after the plasma was switched off as a bolus solution. After each individual treatment, samples were placed on ice, and immediately analyzed by nanoflow liquid chromatography–mass spectrometry (nano-LC-MS/MS), to minimize post-discharge reactions. Three independent experiments and measurements were performed. The long-lived species H₂O₂, nitrate $(NO3−)$ and nitrite $(NO2−)$ were determined according to the previously published approach.³⁸ 

For nanoflow liquid chromatography mass spectrometry, an UltiMate 3000 RSLCnano (Thermo Scientific, Dreieich Germany) system was used with an Acclaim Pepmap C18 column (150 mm × 75 μm, 2.0 μm particle size, Thermo Scientific, Dreieich Germany) with the corresponding precolumn (20 mm × 100 μm, 5.0 μm particle size). The injection volume was 1 μl, equaling 100 ng of the sample. A gradient (see Table I) consisting of water/0.1% v/v acetic acid (A) and acetonitrile/0.1% v/v acetic acid (B) was applied to achieve separation.

Mass spectrometry acquisition was carried out in a data dependent mode with a full scan range of 86.7–1300 m/z (resolution 70 000) and with the fragmentation of the top ten most abundant precursor ions (QExactive Hybrid-Quadrupol-Orbitrap, Thermo Scientific, Dreieich Germany). Precursor ions were fragmented using normalized higher energy collisional dissociation (HCD, 27.5 eV) and detected at a mass resolution of 17 500 at m/z 200. The automatic gain control (AGC) target for full MS was set to 1 × 10⁶ ions and for MS/MS to 2 × 10⁵ ions with a maximum ion injection time of 120 ms for both modes. Dynamic exclusion was set to 30 s and ions with charge state one or above eight were excluded.

To identify the peptides and their potential post-translational modifications, the MS raw data were analyzed by the software Proteome Discoverer 2.4 (Thermo Scientific, Dreieich Germany). For the detection of covalent modifications in the amino acid chain, the software Byonic (Protein Metrics), version 3.6.0, was used as a plug-in tool. In contrast to the Proteome Discoverer, this program is able to perform unlimited searches for multiple modifications in a single peptide. The precursor mass tolerance was set to 3 parts per million (ppm) and fragment mass tolerance to 10 ppm. The false discovery rate (FRD) was set to 1%. For technical limitations, the allowed maximum of modifications per peptide was set to three. The dataset was pre-filtered using different scores from Byonic.⁴³ The first filter was the Byonic score with a cut-off of 250 (range 0–1000), and the Delta Mod score was set to 5 (range from 0 to 100). Both scores result from the match of the theoretical fragment spectrum of a peptide and the experimental one. Following vendor recommendations, these cut-offs can be considered significant. To obtain semi-quantitative data, the number of peptide spectra matches (PSMs) that were detected for a specific modification were counted (PSM counting.⁴⁴)

In total, 103 497 PSMs were analyzed with R (v4.0.2)⁴⁵ using RStudio (v1.3.1093) and the Tidyverse (v1.3.0)⁴⁶ package. The observation data were normalized on the total number of spectra matches for each peptide to account for variances in the ionization efficiency after background subtraction, yielding the occurrence of de novo modifications in percent. Linear models [later referred to as n-way analysis of variance (ANOVA)] were used to analyze the effects of discharge parameters on peptide modification. This statistical method was chosen due to its simplicity, the existence of negative values in the dependent variable (percentage of modifications), and the assumption that in general modification percentages in similar treated samples show a Gaussian distribution. The percentage of one particular PTM was modeled as a result of the categorical effects: (1) amino acid, (2) peptide, (3) solvent, (4) plasma, (5) indirect or direct treatment, and (6) treatment time. To address the issue, that the amino acid sites of the modification (not only the type of modification) might be dependent on the effects (2) to (6) as well, models were additionally fitted considering those interactions. Fitted models with and without interactions were compared for each modification and the models with interaction were significantly more fitting (F-Test; p-value <2.2 × 10⁻¹⁶). This indicates an important impact of the “biochemical” background for the occurrence of modifications.

The dominant targets within the investigated peptide library were the sulfur-containing amino acids cysteine (Cys) and methionine (Met) and the aromatic amino acids tryptophan, tyrosine, and phenylalanine (see Fig. 3), corroborating previous studies using cold physical plasma (CAP).^36–39 The thiol group of cysteine and the thioether group of methionine are excellent targets for oxidation, carrying predominantly one oxygen (Met) or even three (Cys), in this case forming the strongly acidic sulfonic acid moiety. This oxidation is irreversible; prohibiting the formation of sulfide bridges and consequently interferes with protein folding and signaling processes. Alongside, modifications containing only one (cysteine sulfenic acid) or two oxygen atoms (cysteine sulfinic acid) are possible and have been found using cysteine as a model compound during CAP treatment.^28,30 These oxidation states are still reversible in physiologic conditions, e.g., via sulfiredoxin activity, but under CAP impact quickly convert into the stable sulfonic acid form. Further oxidation yielding sulfate and a residual alanine has been found for CAP-driven cysteine oxidation; however, a conversion from cysteine to alanine was not observed in the studied peptide library. Methionine monoxidation is prevalent in vivo and is a marker for oxidative stress.⁴⁷ It occurs during sample preparation by oxidation via triplet oxygen as an artifact, indicating the high reaction probability at the thioether. Methionine oxidation is linked to conformational changes of proteins, modulating their functionality.⁴⁸ 

From the group of aromatic amino acids, tryptophan (Trp) is the major target. The possible chemical reactions are far more complex than for Cys or Met, with special regards to the heterocyclic nature of one of the two annealed rings.⁴⁹ The position of the nitrogen within the rings increases the electron density in both rings, facilitating electrophile substitution reactions introducing oxygen, nitrogen, or chloride atoms that subsequently allow elimination or oxidation reactions enriching the observable product portfolio. In addition, the N-containing 5-ring substructure is prone to ring cleavages via a variety of reaction pathways (see below). Due to its bulkiness and relative hydrophobicity, Trp is typically found in the inner and/or hydrophopic parts of a protein, impeding access by reactive species from the outside. The impact on protein function is hard to predict due to the diversity of the products.

Next to Trp, tyrosine (Tyr), and phenylalanine (Phe) as hydroxybenzene resp. benzene derivatives of α-amino propionic acid belong to the most relevant targets of CAP-derived reactive species. However, due to the higher electron density in the aromatic ring of Tyr resulting from the hydroxyl group, this amino acid is far more often modified than Phe. Again, electrophile substitution reactions dominate, and downstream reactions yield a variety of products. While one or two additional hydroxyl groups and the corresponding oxidation products (quinones, semi-quinones) dominate, an interesting role can be attributed to nitrogen-derived modifications, especially nitration. Nitrated tyrosine residues are reported as markers for oxidative stress. The presence of the electron-pulling nitro group in the aromatic system decreases the pKa value of the amino acids hydroxyl group, subsequently increasing the probability for its phosphorylation by regulatory proteins (tyrosine kinases).^50,51 Therefore, a significant impact on protein functionality can be assumed.

The other 15 amino acids are modified by CAP-derived species to a (far) lesser extent. However, some of the observed modifications are indicative of the presence of specific reactive species and may serve as a diagnostic marker (see below). Ahead of all, lysine (Lys) with its ε-positioned primary amino group is sensitive toward hypochlorite, a species that is generated from atomic oxygen when conditions allow. The products comprise simple chlorination, and addition–elimination reactions ultimately yielding multiple bonds in the Lys molecule.

Somewhat unexpectedly, most of the aliphatic amino acids like valine (Val), leucine (Leu), and isoleucine (Iso) were often targeted, indicating a significant role of primary or secondary generated OH radicals. The products were hydroxylated molecules that in the physiologic context may be relevant by increasing the hydrophilicity of the respective amino acid yielding a modulation of protein conformation and interaction with binding partners. The polar amino acids aspartic acid (Asp), asparagine (Asn), glutamic acid (Glu), glutamine (Gln), arginine (Arg), and serine (Ser) rarely bore modifications. Since these amino acids carry electron-rich structures such as hydroxyl and carboxyl groups or a guanidinium moiety, a significant reactivity was expected.⁵² However, most bonds in the respective moieties are resonance stabilized, reducing their reactivity. In addition, they can reform without a permanent change in the chemical composition exemplified in the following reaction (1):

The formation of an alkoxy radical via H-atom abstraction, shown in (1) for a serine moiety, is a typical reaction of an OH radical.⁵³ The radical may return to ground state by reaction with water. Alkoxy radicals are not very reactive, yet dimerization forming a peroxide or hydroperoxide can occur.⁵⁴ Since this structure is not stable, the reverse reaction yields to the initial hydroxyl group. Alternatively, atomic oxygen, O(³P), can replace the OH radical, as shown in (2),⁵⁵ 

When applying the reaction (1) to an aliphatic amino acid, a two-step reaction yields hydroxylated amino acids. The resulting product 3-hydroxy valine (3b) is stable and yields a permanent change that is annotated as hydroxylation (+15.99 Da),

The annular amino acids histidine (His) and proline (Pro) remain predominantly unchanged (Fig. 3), except when Pro has an exposed amino group in position one in the peptide (peptide 5, NH₂–PQRWSFNMT–COOH). His has a special reactivity toward singlet oxygen, yielding ring-open products (see below).

In all, 17 different modifications were observed in the investigated peptide library. Assumingly, all of these derived from the attack of reactive species contained a differing number of oxygen atoms. In some cases, oxygen is eliminated during formation or downstream reactions. Of the detected modifications, 13 contained oxygen atoms, up to three (Fig. 4 and Table II). Two modifications (dehydrogenation/double dehydrogenation) were solely characterized by a loss of atoms, forming multiple bonds. In one modification (chlorination), chlorine apparently replaces one hydrogen atom in the initial molecule. Nitrogen, as the second most relevant contributor to the CAP-derived reactive species, contributed to five modifications. However, the abundance of such modifications is significantly lower than that for the oxygen counterparts. Besides a lower prevalence of reactive nitrogen species in the investigated discharges, subsequent elimination reactions further reduce their presence as indicated by the deamination of free amino groups that are observed under some conditions. Some of these modifications are created by enzymatic processes during physiological post-translational processing and via redox-controlled signaling events,^3,18,56 pointing at the potential to interfere with protein and ultimately cell functionality via CAP-derived reactive species. The major modifications base on the introduction of gas phase derived hydroxyl radicals or atomic oxygen as shown for the aliphatic amino acids [reaction (3)].^30,32 The hydroxylation observed in the aliphatic amino acids, according to (3a) and (3b), may be driven by hydroxyl radicals. However, due to the slow reaction of atomic oxygen with water (2.2 × 10⁻²³ m³ s⁻¹), yielding to hydroxyl radicals [reaction (2)],⁵⁷ their role may be limited while the direct attack of atomic oxygen at the peptide amino acid moieties is in favor (1.1 × 10⁻¹⁷ m³ s⁻¹).³² This is in contrast to experiments using Fenton’s reaction to generate hydroxyl radicals in liquids, indicating that this model suffers limitations when used to mimic plasma-derived reactive species.⁵⁸ A significant role of singlet oxygen is indicated by the fact that by far the insertion of two oxygen atoms, yielding a dihydroxylation or peroxidation, prevails over monoxidation (hydroxylation) or triple hydroxylation. Since the peroxides are unstable during sample preparation and analysis, two hydroxyl groups are the oxidative modification observed.⁵⁹ 

In addition, ring cleavage reactions in histidine and tryptophan (see below) are indicative for 2 + 2 or 4 + 2 cycloadditon reactions known for singlet oxygen. In the presence of chloride ions, CAP-derived atomic oxygen leads to the formation of hypochlorite (OCl⁻).^31,33 The OCl⁻ ion is unstable; it acts as a chlorination agent and an oxidant, yielding a number of prominent modifications, above all chlorination, carbonylation, and dehydrogenation. The molecule is reactive and reacts both with primary amines, yielding N-chloro- or N,N-dichloro compounds (chloramines)⁶⁰ and with aromatic systems,⁶¹ yielding mono-, di-, or trichloro derivatives (electrophilic substitution). Accordingly, lysine and tyrosine were the main targets observed (Fig. 5 and Fig. S2 in the supplementary material). Surprisingly, no significant chlorination could be found at phenylalanine; although its aromatic ring is similar to Tyr, it lacks the electron-pushing hydroxyl group. Further modifications related to the presence of atomic oxygen are dehydrogenations and the formation of oxo groups (carbonylation). In both cases, addition–elimination reaction occurs. The peptide backbone (amide bonds) was insensitive toward hypochlorite and no fragments have been detected. Since chloride ions are ubiquitous in biological systems, hypochlorite and its downstream products

While it was already shown that the feed gas composition of a given plasma source influences the presence of modifications for some amino acids bound in a peptide,^37–39 the current approach using a peptide library should provide a more general insight. Despite a general concordance in the main target amino acids and the type of introduced modifications (Figs. 3 and 4), explicit differences can be found for a number of specific gas plasma conditions [Figs. 6(a) and 6(b)]. The most obvious difference between the two plasma sources is the introduction of chlorination, carbonylation, and additional double bonds (one/two dehydrogenations) that were observed for the COST-Jet (He/O₂) mainly and are related to the prominent generation of atomic oxygen by this device. For example, the amino acid lysine that was not attacked by kINPen was chlorinated extensively. However, when the COST-jet is driven in the He-only mode, the reactivity is almost completely lost. Instead, kINPen treatment yielded trioxidation (e.g., three hydroxyl groups) and the cleavage of heterocyclic rings that are related to the reactivity of singlet oxygen. In contrast to the COST-Jet, the kINPen also introduced nitrogen-containing modifications, especially nitration, in part combined with oxidation. Both plasma sources did not differ in the total number of the most common modification, the simple hydroxylation (+16 Da). Some amino acids were modified by the COST-Jet modified to a larger extent, e.g., serine, methionine, and cysteine, probably due to the atomic oxygen density of this device. In contrast, phenylalanine was a better target for the kINPen-derived species. Most aliphatic amino acids (e.g., valine and isoleucine) are modified to a similar extent by both plasma sources, and no clear correlation with the working gas composition was found.

The observations reflect the differences of the working gases argon and helium regarding the formation and lifetime of higher energy states and the different electrode configuration/driving power, yielding differences in reactive species formation and dynamics, which is further modulated by the addition of molecular gases.^63,64 For the helium-only COST-Jet, an almost complete absence of any modifications was noticed, indicating the inability of He higher energy states to reach the surface of the target and a very limited impact of ambient air species due to a laminar gas flow leading to a minimum of reactive species.

kINPen in the Ar-only mode retained a high basal generation of ROS and RNS due to the turbulent effluent and the subsequent interaction with the ambient air, and the higher mobility of argon higher energy states. The device showed an interesting modification pattern regarding carbonylation, ring cleavage reactions, and nitration. Carbonylation, which is dominant for the COST-jet and induced by atomic oxygen, was detected for Ar/O₂ treatment and decreases again with N₂-admixture. This indicates that kINPen forms atomic oxygen to a limited extent, but the observed chemistry is dominated by singlet oxygen³⁵ and N_xO_y molecules.⁶⁵ Second, ring cleavage reactions at tryptophan or histidine, which are maximal in argon only conditions and decrease with each additional gas, are exclusively formed after kINPen treatments. The reaction is attributed to singlet oxygen and the potential mechanism for the reaction with tryptophan is shown in Fig. 7.^49,66 Initially, singlet oxygen attacks the carbon atom and forms tryptophan hydroperoxide, which converts to the tryptophan dioxetane. Alternatively, a 2 + 2 cycloaddition yields directly the dioxetan ring. Subsequent rearrangement forms the N-formylkynurenine. In the last step, the final product 2-hydroxykynurenine is formed under the attack of a hydroxyl ion. It remains elusive why the addition of oxygen does not favor an increase of Trp cleavage, since the generation of singlet oxygen increases in this condition.³⁵ It may be speculated that either changes in reactive species dynamics at the boundary layer and the liquid bulk reduced the fraction of the chemically active singlet oxygen (e.g., ozone formation) or subsequent secondary reactions after the initial cleavage of the Trp moiety led to non-recognizable modifications. Third, the introduction of nitro groups into aromatic amino acids (nitration, nitration + oxidation), which is exclusive for the kINPen, appears for Ar or Ar/O₂/N₂ admixtures but not in Ar/O₂, supporting the above conclusion. Nitration is mainly found in peptides 1 (NH₂–AWQDHGLYSK–COOH) and 6 (NH₂–IGYKALEVCH–COOH) at tyrosine (Fig. 11). An assumed mechanism for the nitration of the peptide is shown in Fig. 8.

Assumingly, peroxynitrite is the main responsible reactive species. Its generation by plasmas is well-recognized^68,69 and production at the interface from nitric oxide and superoxide anion radicals can be assumed.⁷⁰ If only oxygen is available in the discharge, $NO⋅$ can no longer be generated in sufficient quantity and thus ONOO⁻ generation ceases, yielding a reduced number of nitrations [Fig. 6(b)]. Additionally, secondary reactions remove the bulky nitro groups in favor of hydroxyl groups or quinones.⁷¹ In Fig. 9, the amino acids Cys, Lys, Pro, Ser, Trp, and Tyr are compared, displaying strong differences in type and amount of modifications in dependence on the used plasma source and between the different amino acids. This is specifically obvious for amino acids that can be modified by chlorination (Tyr, Lys), or singlet oxygen (Trp).

It remained questionable, which contribution from long-lived reactive species can be expected for the above-described modifications. Hydrogen peroxide, nitrite, nitrate, or ozone are formed to a varying degree during the treatment.³⁸ The medium-lived OCl⁻ chemistry preserves the short-lived atomic oxygen and requires to be considered likewise. Therefore, the peptides were exposed to CAP in two orthogonal approaches: the peptides were present either during the discharge was ignited (direct treatment) or were added to the liquid after the discharge had been switched off (indirect treatment). A significant difference between direct and indirect treatment regimens was observed for the two plasma sources [Figs. 10(a) and 10(b)], predominantly inflicted by the hypochlorite chemistry [Figs. 10(c) and 10(d)]. Amino acids sensitive to OCl-driven chlorination (Tyr, Lys) or carbonylation (Ser, Trp) were major targets of the indirect COST-Jet treatment. The co-occurrence of the two modifications strongly suggests that the hypochlorite ion is an active agent for both cases. In contrast, kINPen favored modifications generated during direct treatments, driven by the chemical impact of hydroxyl radicals, atomic oxygen, or singlet oxygen (Cys, Trp, Tyr). Therefore, nitration and the ring cleavages that occur only in direct treatments emphasize the role of the short-lived singlet oxygen and NO₂ radicals.⁷² 

Perpetuating investigations on the role of the environment during the impact of plasma-derived ROS, the modification of peptides dissolved in water or a buffered, chloride ions containing system (PBS) was compared. In addition to the discussed hypochlorite chemistry, PBS favored reactions that benefit from higher pH such as the oxidation of thiol groups and aminations. Direct oxidations by gas phase species (atomic oxygen, singlet oxygen) decrease in PBS since the abundant chloride ions depress species density at the boundary layer significantly. Therefore, the number of ring cleavages at Trp is reduced in PBS. Reactions, which accelerate at low pH, e.g., nitrations, were found in water predominantly.

The role of the chemical environment of a given amino acid in the peptide chain determined by its position and neighboring amino acids is considered. To exemplify the situation, the nitration of tyrosine is considered (+44.98 Da; + N + 2O–H). While mostly observed for the kINPen (Ar, Ar/N₂/O₂), its appearance fluctuated for the different tyrosine residues present in the library (see Fig. S4 in the supplementary material). Clearly, the sequence and the spatial structure of the peptides are relevant: in peptide 3 (NH₂–CHAGRYFVPW–COOH) and peptide 8 (NH₂–RYVFDEASIL–COOH), phenylalanine (Phe) is in close neighborhood to the modified tyrosine. The weakly reactive Phe has a large aromatic ring, interfering either by reducing access to the tyrosine moiety by steric hindrance, or by reducing the reactivity of the tyrosine’s aromatic ring via hydrogen bonds or charge-transfer complexes.⁷³ In contrast, the tyrosine moieties of peptide 1 (NH₂–AWQDHGLYSK–COOH) and 6 (NH₂–IGYKALEVCH–COOH) do not have amino acids with bulkier or aromatic side chains nearby and can, therefore, be nitrated unhindered (Fig. 11). A similar observation has been made for complete proteins.⁷⁴ A second example is the chlorination (+33.96 Da, +Cl–H) (Fig. S2 in the supplementary material). Comparing all peptides in the library, the tyrosine in peptide 6 showed the highest sensitivity to chlorination. A possible explanation is that tyrosine is the exclusive target for oxidative chlorination in this peptide. The only other aromatic structure in the peptide is the imidazole ring of histidine, but this is located far from the targeted tyrosine and not prone for oxidative chlorination. Besides the ε-amino groups of lysine within the peptide chain, the amino groups at the N-terminus of all peptides were chlorinated in appropriate conditions. The α-amino groups incorporated in the peptide chain as part of the peptide bond are not attacked. Cysteine is among the most frequently modified amino acids in the set and shows distinct differences in type and frequency of modifications (Fig. S5 in the supplementary material). The amino acid occurs at different positions in the peptides 2 (NH₂–MFCEPITRNV–COOH), 3 (NH₂–CHAGRYFVPW–COOH), 6 (NH₂–IGYKALEVCH–COOH), 7 (NH₂–TKHNQCPWMG–COOH), and 9 (NH₂–NLMPCTQHAT–COOH). It is worth mentioning that the cysteine in peptides 3 and 6 both have histidine as a neighboring amino acid and in peptides 7 and 9, proline is next to the cysteine. However, the modifications at cysteine in peptides 3 and 6 differ significantly.

While mono-/di-/trioxidation was found on the cysteine in peptide 6 in most treatment conditions, this is not the case for the cysteine in peptide 3. Here, cysteine is located at the N-terminus, while it is close to the C-terminal in peptide 6. Potentially, the local pH, which modulates the thiol group reactivity drastically, differs. The pKa value of the thiol group may range between pKa = 3.5 and 10, spanning several log-steps.⁷⁵ A high degree of dissociation aligns with a reaction probability. Yet, the isoelectric points (pI) of peptides 3 (pI = 8.23) and 6 (pI = 6.74) do not suggest a higher probability of cysteine oxidation in peptide 6. However, peptide 3 has a C-terminally located tryptophan, which is an excellent scavenger for the plasma-generated species. It was shown that tryptophan reacts with a rate constant of 3.2 × 10⁷ M⁻¹ s⁻¹⁷⁶  with singlet oxygen (¹O₂), which is the fastest compared to other amino acids. In addition, since Trp is hydrophobic and the N-terminus protonated, the peptide adopts a protecting inward position of cysteine at the highly reactive interface toward the liquid bulk. Positively charged amino acids stabilize the resulting thiolate anion and contribute to the lower pKa, in parallel increasing the thiol group reactivity of the cysteine.⁷⁷ A similar observation was made for the cysteine residues the peptides 7 and 9. Redox-signaling proteins, e.g., peroxiredoxins, exploit this diversity in reactivity.⁵⁶ In addition, there is increasing evidence that oxidative modifications are in crosstalk with other regulatory post-translational modifications, controlling cells performance.⁵¹ 

Using statistical tools, the relationship between the plasma source, the mode of treatment, the treatment time, and the solvent system was analyzed [Figs. 12(a) and 12(b)]. The best correlation between the targeted amino acids or observed modifications and the conditions under test was found for the mode of treatment (direct vs indirect). Corroborating, the solvent system (water or PBS) had a strong impact on the target. Both indicators point toward short-lived reactive species including the atomic oxygen-derived hypochlorite as major ROS/RNS. Of note, the treatment time exerted a very limited impact compared to the other parameters. This indicates to some degree the technical limitations of the adopted approach of PSM counting to quantify the observed modifications. It even more emphasizes the relevance of interface reactions, where the full amount of gas phase reactive species is present at any given time point during the treatment process. Whether a depletion of the boundary layer by fast gas phase species–target molecule reactions might be of relevance, leading to an increasing inefficacy of the treatment with longer treatment time, might be debatable.

A boundary layer depletion was observed for hypochlorite formation, and cysteine or glutathione thiol group oxidation.^28,33,39,62 Experiments segregating the target peptides and the boundary layer yielded to a different pattern of modifications, indicating again that interface reactions are dominant for all conditions where either atomic oxygen is irrelevant or chloride ions are not present. For future studies, it seems to be worth extending the investigated solvent systems, since the dominant role of chloride ions depends according to the literature on their concentration, modulating the hypochlorite formation. In addition, the presence of scavengers common to complex biological systems influences the modification pattern. Pilot experiments indicated a fundamental impact of free histidine on the modification pattern. In the human skin, urocanic acid, a catabolite of histidine, may serve the same function.⁷⁸ 

This work focused on the detection of covalent modifications introduced by reactive oxygen and nitrogen species from cold physical plasmas in a model peptide library. Major targets were the amino acids cysteine, methionine, tyrosine, and tryptophan bearing additional oxygen, nitrogen, or chlorine atoms forming additional hydroxyl, oxo, nitro, or chloro groups. The type of the modification and their extent depend on the CAP source, treatment conditions, the environment of the peptide or protein, and the local chemical environment of each amino acid. Dominant short-lived reactive species were atomic oxygen, singlet oxygen, and to a lesser extent hydroxyl radicals. The formation of hypochlorite and peroxynitrite showed a significant contribution to peptide modification. Long-lived species like hydrogen peroxide, nitrite, or nitrate were irrelevant. The cleavage of histidine and tryptophan pointed on singlet oxygen, a major product of the Ar-driven kINPen. Atomic oxygen, a major product of He/O₂ discharges (COST-Jet), left a significant mark via chlorination when favored by conditions.

Most of the newly introduced groups are identical or equivalent to the enzymatically introduced post-translational modifications (oxPTMs) in physiologic protein processing or cell signaling. They are stable enough to withstand sample preparation and mass spectrometry analysis, suggesting that such de novo introduced peptide or protein modifications occur in vitro and in vivo after a CAP treatment and may subsequently modulate local signaling or protein functionality. Reports on the impact of myeloperoxidase-derived products on the fibrous matrix protein fibronectin emphasize this notion,^61,79 underlining the unique potential of CAP to interfere with biological processes via the modification of biomolecules.

In the supplementary material, a complete overview of all oxidation (Fig. S1) and chlorination events (Fig. S2) for all peptides in the library is shown. The modifications are normalized and divided between the different peptides, gas compositions, and treatment conditions (Fig. S1) or the solvent system (Fig. S2). Figure S3 shows the impact of the different treatment times (15 and 60 s) for nine different amino acids (S3A) and ten different modifications (S3B). Figure S4 gives a similar overview, like Fig. S1, for all modifications identified at cysteine in the five peptides that contained this amino acid. Finally, Table S1 displays the peptide sequence (single letter amino acid code) for the whole library together with the corresponding molar mass and the calculated isoelectric point.

Funding from the German Federal Ministry of Education and Research (Grant No. 03Z22DN11 to S.B. and 03Z22DN12 to K.W.) supported this work. The authors thank Niklas Lengner for laboratory assistance.

The data that support the findings of this study are available from the corresponding author upon reasonable request.