Nonenzymatic post-translational modifications in peptides by cold plasma-derived reactive oxygen and nitrogen species

Cold physical plasma is increasingly used for various biomedical applications. Besides decontamination and surface optimization for medical products such as implants,^1,2 acute and chronic wound management,³ and cancer treatment^4,5 are major fields of research and (pre)clinical use. Specifically, the controllable mix of short- and long-lived reactive oxygen and nitrogen species (ROS/RNS) in combination with electrical fields and UV radiation allows plasmas to interfere with cellular signaling processes on different levels.^6–8 Besides a direct interaction of ROS with cellular signal receptors, e.g., peroxiredoxins (reviewed in Ref. 9), short-lived reactive species such as singlet oxygen [O₂(¹Δ_g)], atomic oxygen O, or peroxynitrite (ONOO⁻) may introduce covalent chemical changes in a range of biomolecules, thereby influencing their structure and activity.^10–13 Assumingly, such covalent modifications contribute to the observed physiological impact of plasma specifically when the resulting structure resembles biologically active post-translational modifications (PTMs). However, mechanisms and extent remain to be elucidated.

Post-translational modifications are versatile “customizations” of proteins and peptides, controlling their activity, fate, and lifetime.^14,15 Often, they are enzymatically introduced after protein translation, and more than 200 biologically relevant types of PTMs have been identified,¹⁶ with phosphorylations, ubiquitination, and glycosylations being the most frequent.¹⁷ However, the database Unimod (http://www.unimod.org/) contains 1500 entries of protein modifications (May 2020), many of those with unknown impact in biological systems, showing the complexity of the research area. Also, nonenzymatic routes have been described and were found relevant for biological processes,^18,19 this time focusing on oxidative structures including hydroxylation, carbonylation, and nitration.²⁰ The proteome-wide identification of enzymatic and nonenzymatic PTMs by high-resolution mass spectrometry is a challenging approach receiving increasing attention.^21,22

Here, the impact of the cold plasma treatment on two model peptides, angiotensin 1–7 and bradykinin, was investigated. The widely used argon based plasma jet kINPen and the helium based plasma jet COST-Jet were tested for their ability to introduce nonenzymatic post-translational modifications and to infer on the underlying chemical reactions and reactive species, respectively. Both plasma sources vary significantly in design and reactive species output.^23–26 Bradykinin is a small pro-inflammatory peptide in mammals, which is composed of nine amino acids (Arg-Pro-Pro-Gly-Phe-Ser-Pro-Phe-Arg), and angiotensin 1–7 (Asp-Arg-Val-Tyr-Ile-His-Pro) is the amino-terminal fragment of the angiotensin I/II involved in blood pressure regulation. The amino acids present contain hydrophobic side chains (valine and isoleucine), aromatic side chains (tyrosine and phenylalanine), charged (arginine and histidine) or uncharged side chains (asparagine and serine), and some special cases (glycine and proline) and represent a suitable range of target structures. Of note, no sulfur-containing amino acid is present in both peptides since previous experiments revealed a strong oxidation of thiol groups by reactive species, precluding an unbiased view of the other amino acids’ reactivity.^27,28

Recent work by various authors rose awareness to discriminate between interface and bulk reactions in liquids that are in contact with a plasma discharge or the active effluent,^29–33 indicating that the short-lived species such as hydroxyl radicals dominate the downstream chemistry. This is addressed in a molecular dynamics simulation study on the impact of hydroxyl radicals on bradykinin and angiotensin published by Verlackt et al.,³⁴ allowing to pre-estimate target structures for the current approach and to identify relevant reactive species by comparison between theoretical and experimental results. Using the peptides as target molecules that preserve the impact of reactive species in a covalent chemical bond allows taking snapshots of the plasma chemistry at the interface and bulk simultaneously, augmenting electron paramagnetic resonance spectroscopy data present for the kINPen.^35,36

In conclusion, two different plasma sources are used to tackle the following questions: First, which PTMs are induced nonenzymatically by ROS/RNS and which amino acids are the major targets? Second, which reactive species are responsible for these modifications? And third, are the experimental results congruent with the simulation results? High-resolution mass spectrometry coupled to nanoflow liquid chromatography (nanoLC-MS/MS) along with bioinformatics tools was applied to identify and quantify the oxidative post-translational modifications resulting from the treatment by the two applied plasma sources. The results confirm the relevance of secondary effects in cold plasma effects.

Angiotensin 1–7 and bradykinin (Sigma-Aldrich-Chemie GmbH, Germany) were dissolved in 500 μl of double distilled H₂O (Millipore) with a concentration of 0.2 mg/ml. The water was thoroughly degassed by bubbling with argon gas for 30 min to reduce background reactions. For the treatments, we used the kINPen (neoplas tools GmbH, Germany)³⁷ and the COST Reference Microplasma Jet (COST-Jet is the outcome of the European COST action MP 1101 “Biomedical Applications of Atmospheric Pressure Plasma Technology”).³⁸ The kINPen consists of a grounded ring electrode and a ceramic capillary, where a centered rod electrode is located inside (Fig. 1 right). A voltage of 2–6 kV_pp is applied to the central electrode with a frequency of 1.1 MHz. Argon gas (purity 99.999%) was used as a working gas with a flow rate of 3 standard liters per minute (sLm).³⁹ For some treatments, 0.5% of the working gas was replaced with N₂, O₂, or with a mixture of both gases (15 SCCM each). The COST-Jet is a device that was developed by the European COST action MP 1101 as a reference device for plasma sources. It consists of two 1 mm thick metal plate electrodes spaced 1 mm apart in between which the plasma is ignited (Fig. 1 left). A Tektronix DPO 4104 Digital Phosphor Oscilloscope was used to measure the time resolved voltage for the plasma. For pure helium experiments, the plasma was sustained to 143 mV, for helium/oxygen to 145 mV, for helium/nitrogen to 159 mV, and for helium/oxygen/nitrogen to 165 mV. The treatments were performed at a sample distance of 4 mm and 1 sLm helium. If desired, 0.5% oxygen, nitrogen, or both gases were added (5 SCCM each). All treatments were done in 24-well plates and placed 4 mm (COST-Jet) or 9 mm (kINPen) away from the liquid surface for 30 or 60 s. After each individual treatment, samples were placed on wet ice and immediately analyzed by nanoLC-MS/MS to minimize postdischarge reactions (see Sec. II B). Three independent experiments were performed.

For further characterization of the two plasma sources, the long-lived species H₂O₂, nitrate (NO₃⁻), and nitrite (NO₂⁻) (Fig. 2) were also determined for the different compositions of gases. The quantification of nitrite and nitrate was performed with ion chromatography (ICS-5000, Thermo Fisher Scientific). For the separation, an IonPac AS23 anion exchange column (2 × 250 mm, Thermo Fisher Scientific) and an isocratic mobile phase (4.5 mM Na₂CO₃/0.8 mM NaHCO₃) with 250 μl/min were used. Hydrogen peroxide was quantified via the colorimetric reaction with xylenol orange using a commercially available assay (Pierce Quantitative Peroxide Assay Kit, Thermo Scientific) according to the manufacturer's protocol.

An UltiMate 3000 RSLCnano system equipped with an Acclaim Pepmap C18 column (150 mm × 75 μm, 2.0 μm particle size, Thermo Fisher Scientific) and corresponding precolumn and a QExactive Hybrid-Quadrupol-Orbitrap mass spectrometer from Thermo Fisher was used. The injection volume was 1 μl, equaling 200 ng of sample. A step 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 of modified peptides. All samples were injected twice.

MS acquisition was performed in a data dependent Top 10 mode (full MS/dd-MS/MS) at resolving powers of 70.000 in MS mode and 17.500 in MS/MS mode. A Nanospray Flex electrospray ion source was used in positive mode, and the spray voltage was set to 2.30 kV using stainless steel emitters. The temperature of the transfer capillary was set to 250 °C. The collision energy in the higher-energy collisional dissociation cell was fixed at 27.5 V (HCD fragmentation). The observed mass shifts allowed identification of the peptide sequence and to pinpoint modified amino acids.

Raw data analysis was achieved by the Proteome Discoverer 2.2 software (Thermo Fisher Scientific) using Byonic (Protein Metrics),⁴⁰ Version 3.4.1, as a plug-in to identify post-translational modifications. This tool significantly extents search options for peptide modifications. Based on the previous work,^12,41,42 an inclusion list of expected oxidative modifications was compiled (Table II). In the search, precursor mass tolerance was set to 3 ppm and fragment mass tolerance to 10 ppm. A false discovery rate of 1% was accepted. Due to calculation power restraints, maximum three PTMs were allowed per peptide. The spectra were examined manually to ensure the presence of specified modification. To obtain semiquantitative data, the number of spectra matching a detected modifications summed up (PSM counting).⁴³ 

Previous investigations on isolated amino acids determined a broad range of chemical structures to be susceptible to the impact of plasma-derived reactive species.^12,29,44 For one, it remained to be clarified whether similar covalent changes occur in a peptide chain, since the competition for targets or species may occur. For another, it should be determined whether these modifications relate to physiological PTMs since they may contribute to the observed physiological consequences of cold plasma.^8,45

The argon-driven kINPen showed a significant impact on the amino acids in both angiotensin 1–7 and bradykinin (Figs. 3 and 4). The reactivity toward the plasma-derived species differs by amino acid, its position within the peptide chain, the gas phase composition, and the plasma source (kINPen or COST-Jet). The highest numbers of post-translational modifications were observed for tyrosine (Y), isoleucine (I), and histidine (H) in the angiotensin 1–7 and phenylalanine (F) and proline (P) in bradykinin. Interestingly, the proline in angiotensin 1–7 (position 7) remains almost unmodified, while at the positions 3 and 7 of bradykinin, it is significantly attacked. In the same molecule, proline at the position 2 remains scarcely attacked by the plasma-generated species. A similar behavior was observed for phenylalanine of bradykinin at positions 5 (center) and 8 (carboxy terminal end). Assumingly, steric hindrances and local variations of the electron densities of heteroatoms or aromatic rings contribute to the observed divergence in reactivity. In Fig. 4, the most commonly observed modifications for angiotensin 1–7 (A) and bradykinin (B) are shown. For pure argon discharge, the most frequent modifications are the oxidation (Δm +15.99 Da) and nitration (Δm +44.99 Da) at the amino acids tyrosine and isoleucine of angiotensin and oxidation of phenylalanine at position 5 of bradykinin. Additionally, dioxidations (Δm +31.99 Da) were found at histidine (angiotensin) and proline (bradykinin, positions 3 and 7). A ring cleavage forming formylaspargine (Δm +4.98 Da) was found for histidine according to the literature driven by singlet oxygen.⁴⁶ Almost no modifications of the arginine moiety were observed. With respect to the positively charged guanidinium group, a significant impact of the electron-rich (and in part negatively) charged reactive oxygen species was assumed. However, in accordance with qualitative observations,¹² arginine shows a remarkably resilience to reactive species that is lifted to some extent by ROS-rich Ar/O₂ discharges.

If O₂ is added to the working gas, nitrations disappear in favor of oxygen-dominated PTMs, reflecting changes in the gas phase and liquid phase chemistry.^47,48 Oxygen enriched plasmas are known to produce O₃, ^●O, and ¹O₂,⁴⁹ species that are strong oxidants⁵⁰ with high prevalence in the gas-liquid interface.²⁹ Experimental evidence is the observation of a modification on tyrosine of Δm +47.98 Da with three oxygen atoms and the increase in Δm +31.99 Da mass shifts (e.g., isoleucine hydroperoxide and double oxidized histidine). Interestingly, it is the only condition where arginine is oxidized, implying a role for atomic and singlet oxygen.

When nitrogen was added to the working gas, the aromatic amino acids become the major target. The nitration of phenylalanine occurs only under these conditions in bradykinin, and monohydroxylation of the aromatic rings in either Phe or Tyr was found. The oxidation of proline (Δm +15.99 Da) at position 3 of bradykinin was observed only here, indicating that this modification is destroyed by harsher oxidizing conditions as in Ar/O₂.

For the kINPen argon/O₂/N₂ plasma, nitrogen-containing modifications were frequent, with nitrosylation of phenylalanine as a marker for the presence of nitric oxide standing out in bradykinin. In the angiotensin 1–7 peptide, nitrotyrosine is formed via reaction with peroxynitrite ONOO⁻,^51,52 which yields at the gas-liquid interface from the reaction of superoxide (O₂^●−) and nitric oxide (^●NO),⁵³ and it is already shown that this molecule also occurs in physical plasma.⁵⁴ This nitrotyrosine is not formed when oxygen or nitrogen alone is used as an admixture. This supports the notion that nitric oxide is formed at the interphase by the reaction of nitrogen dioxide radicals and atomic oxygen and that it serves as a reactant for the formation of peroxynitrite,⁵⁵ and such requires both ROS and RNS.

For COST-Jet, similar amino acid modifications were observed (Fig. 3). Their overall number is lower than for the kINPen. This is particularly evident for the 60 s helium plasma treatment of angiotensin [Fig. 3(a)]. While the same three amino acids, tyrosine, isoleucine, and histidine, are modified, its total extent is only 25%–30% of the kINPen treatments. The main modifications of tyrosine are quinone formation (Δm +29.97 Da) and monoxidation (Δm +15.99 Da). The structure is shown in Fig. 5(a). In bradykinin, phenylalanine at position 5 is the major target [Figs. 3(c) and 3(d)]. With the pure helium discharge, hydroxylations (+O, Δm +15.99 Da) dominate by far (Fig. 7). Two proline residues at positions 3 and 7 are modified to carry two or three oxygen atoms (Δm +31.99 or Δm +47.98 Da). The oxygen admixture to the feed gas led in a massive increase of tyrosine oxidation (one to three oxygen atoms added) in angiotensin. This result mirrors observations using phenol as a target molecule³² assumingly due to atomic oxygen and/or singlet oxygen that are produced in high levels by COST-Jet in these conditions.²⁶ Of note, a modification of the arginine residue at position 1 occurs in the same kind and amount as with the kINPen. Phenylalanine was oxidized to a 1,2- or 2,5-quinone [Δm +29.97 Da, Fig. 5(b)].

By adding nitrogen to COST-Jet, the modification of phenylalanine is further increased, yielding, especially, trihydroxylation (+3O, Δm +47.98 Da). However, under these conditions, RNS were expected, yet no nitration or nitrosylation was observed for any COST-Jet conditions. This is in accordance to the literature, with oxygen species dominating in the discharge. If both molecular gases were added, tyrosine at position 4 of angiotensin 1–7 and phenylalanine at position 5 of bradykinin remain the main targets for the ROS/RNS in the plasma [Figs. 3(b) and 3(d)].

In general, the treatment of a peptide by COST-Jet led to a significantly lower total number of PTMs than the kINPen (Fig. 3). This is due to a lower gas flow rate and the significantly lower specific density of helium. The combination of heavy argon and a higher gas flow rate create an entirely different gas dynamics and mechanical impact of the kINPen,⁵⁶ yielding a richer interface chemistry. Both plasma sources yielded different modification patterns at the two model peptides (Figs. 6 and 7) reflecting their different gas-phase and liquid-phase chemistries that were observed for cysteine and glutathione model systems.^27,28

Nitration and nitrosylations were found for conditions rich in RNS deposition as reported for kINPen (Ar, Ar/N₂/O₂).^55,57,58 When atomic oxygen was a major species (COST-Jet), quinone formation at tyrosine or histidine was observed. As well, the dioxidation of tyrosine does occur predominantly for variants rich in atomic oxygen [Figs. 4(a), 5(a), and 6]^32,59 but not in the singlet oxygen rich kINPen that favored the formylasparagine modification (+2O − H − C − H) at histidine [ninefold higher than COST-Jet, Fig. 6(a)]. Other modifications attributed to hydroxyl radicals were found in both plasma sources in the same extent as the monoxidation of histidine or the didehydration of proline (Δm −2.02 Da). A strong variation in the modification rate of the same amino acids at a different position (=a different chemical environment) was observed at the bradykinin’s phenylalanine residue at position 5. For COST-Jet, this amino acid was oxidized particularly strong (Figs. 3 and 7). When treated with He/N₂, 5 F modification by the COST-Jet exceeded that of the kINPen by a factor of 5. In contrast, the phenylalanine residue at position 8 remained without detected modifications. Bradykinin’s proline (3P) is the main target for the kINPen derived reactive species with dioxidation as the most common modification. The same amino acid loses two hydrogen atoms (didehydration) by the COST-Jet treatment, assumingly via an addition-subtraction reaction mechanism driven by hydroxyl radicals. These observations fit to previous results,²⁹ showing that the kINPen liquid chemistry is influenced by singlet oxygen O₂(¹Δ_g), leading to the dioxidation of thiols. On the other hand, the role of atomic oxygen in case of COST-Jet³⁸ leads to hydroxyl radical driven oxidations such as hydroxylations (+O) or H abstractions (−2H). Besides the short-lived species, both plasma sources deposit long-lived species in liquids (see Fig. 2). Hydrogen peroxide is a major product in kINPen and COST-Jet for argon or helium only discharges while nitrite and nitrate dominate conditions with N₂ and N₂/O₂ admixtures. Despite the significant concentrations (up to 150 μM), only a very limited impact on the peptides could be attributed to their presence. In indirect treatments (peptides were added after the plasma was switched off), only a negligible number of PTMs was observed. This is in accordance with previous results using cysteine as a tracer.^27,42,44

Verlackt et al. published in 2017 a study investigating the oxidation of bradykinin and angiotensin 1–7 by OH radicals³⁴ using a combined approach of reactive and nonreactive long time scale molecular dynamics simulation. In angiotensin, the four amino acids, asparagine, tyrosine, histidine, and proline, were identified as major targets; for bradykinin, the corresponding targets were arginine, serine, and proline. Predominantly, an H-atom abstraction was predicted as an initial step. Only for bradykinin, a minor 3% of the reactions were OH additions at phenylalanine 8, yielding predominantly tyrosine or positional isomer thereof. The current experimental data, comparing kINPen and COST-Jet, revealed both similarities and differences with the model (Fig. 8). For angiotensin 1–7, a good agreement between expected and observed chemical reactivity toward kINPen (Ar)- and COST-Jet (He)-derived reactive species was found for the amino acids valine, tyrosine, and histidine. The aromatic amino acids tyrosine and histidine are electron-rich and intermediates stabilize via electron delocalization, accordingly the model and the experiment coaligned on their intense modification. In the simulation, valine and isoleucine are inert to the OH attack in the tested time frame (10 ps), while experimentally, valine and, especially, isoleucine were oxidized by both plasma sources. This is in agreement with Takai et al. who reported the formation of hydroxylated dicarbonic acids.¹² The longer side chain of isoleucine protruding from the peptide backbone and the presence of an additional secondary C-atom discriminate it from the less modified valine. Presumably, time scales >10 ps or reactive species not considered in the simulation are relevant in the attack. Atomic oxygen, the major species produced by COST-Jet in the He/O₂ regime, did not yield to a stronger isoleucine oxidation than singlet oxygen rich kINPen. A discord between the model and the experiment was also observed for aspartic acid: in contrast to predictions, no modifications were detected, which is again in agreement with Takai et al. It may be concluded that the assumed abstraction of an H-atom at the side chain carbonic acid moiety is not relevant for subsequent covalent modifications since the resulting radical is resonance stabilized at low energy. Additionally, with a pKa of 3.9, the β-COOH moiety is quite acidic and deprotonates in near neutral solutions. The resulting electron-rich COO⁻ moiety deters electron-rich and negatively charged reactive species and thereby protecting the carbon backbone from modification. A similar observation was made for Pro at carboxy terminus position 7 in angiotensin. With a pKa value of 1.99, the COOH moiety deprotonates under the experimental conditions, deterring the attack of electron-rich reactive species. In contrast, for bradykinin, where proline’s carboxyl group is masked as an amide, modifications (dioxidations/ring cleavages) were frequent. Finally, arginine (Arg) was expected to be a major target, but neither source led to observed modifications, in good agreement with Takei et al. In the model, bradykinin’s phenylalanine at position 5 was expected to be inert, yet the experimental observations showed that the amino acid is indeed a good target due to its large and resonance stabilized aromatic moiety. The discrepancy with simulation can be explained by technical restraints in the model and, additionally, by the relevant reactive species (assumingly atomic oxygen) that the model did not consider. Regarding the comparison of the experimental data obtained for the oxygen admixture with the models’ prediction, a gradual improvement could be observed. Specifically, for COST-Jet He/O₂ and peptide bradykinin, a good correlation between the prediction and the observation—with the exception of phenylalanine—can be stated. Under this condition, the COST-Jet is known to produce reactive oxygen species, with atomic oxygen as major species. In contrast, reactive nitrogen species have low abundancy. Such model, taking OH radicals into account that easily yield from the reaction of atomic oxygen with water molecules, has a higher prediction power for this source. The kINPen, with its significant production of reactive nitrogen species, highlights the limits of the simulation.

In conclusion, the observed divergence between the model and the experiment is multifaceted: beyond OH radicals, further species such as atomic oxygen, singlet oxygen, and the triad nitric oxide/superoxide/peroxynitrite contribute to liquid chemistry.^{35,44,55,60–62} Technical constraints of simulation studies limit the ability to address, e.g., local electron density changes or the degree of solvation. Experimental limitations include that natural occurring PTMs were focused, limiting the detection of unexpected modifications. However, the significant overlap with the plasma treatment of isolated amino acids suggests their extent to be minor.

Despite the known efficacy of cold plasma in biological systems in vitro and in vivo, a gap between the ns/µs time scales of the gaseous plasma chemistry and the short lifetime of the reactive species involved compared to the substantial longer time scales of cellular processes (seconds to hours) must be acknowledged. In addition to direct effects of long-lived species (e.g., hydrogen peroxide), secondary processes resulting from the modifications of biomolecules by the short-lived species may modulate cell response and signaling processes. It was observed that the cold plasma treatment interferes with the protein structure and subsequently its function.^11,63–67 More recently, oxidative modifications to proteins have been connected to changes in the immune response, possibly attenuating cancer cell immunoediting. The current results reveal parts of the underlying molecular basis showing that cold plasma can introduce nonenzymatically post-translational modifications in a number of amino acids and that most of these PTMs belong to the natural subset that can be interpreted by the cellular machinery. Heavy oxidation directs a protein toward protein ubiqutinylation and subsequent degradation, or activates chaperons-like heat shock proteins to regain its normal function. Along with the significant oxidation of cysteine, seminal processes like the unfolded protein response that control cell fate and function can be triggered.⁶⁸ Less extensive modifications like nitrations or nitrosylations at tyrosine can be involved in kinase signaling cascades, influencing phosphorylation events as has been observed in the nitric oxide research community for the human aldolase A.⁶⁹ In this light, a strong link between the short-lived species in the discharge and the cells prolonged response can be found in the covalent modifications of proteins and peptides.

Utilizing the model peptides bradykinin and angiotensin 1–7, nine different post-translational modifications (PTMs) were observed after cold plasma treatment (kINPen and COST-Jet). Major targets were the amino acids tyrosine, tryptophan, histidine, and phenylalanine as well as the aliphatic amino acids proline and isoleucine. In contrast, the amino acids valine, arginine, and aspartic acid remained mostly unchanged. The observed PTM patterns indicate the activity of the short-lived species hydroxyl radicals, atomic and singlet oxygen, and peroxynitrite. Frequent observed PTMs were nitration, oxidation (plus one/two oxygen atoms), and ring cleavage (e.g., histidine). Experimental data only in part overlap with a previously published molecular dynamics simulation, indicating that hydroxyl radicals are not the only—or major—oxidant. The chemical environment and steric restriction within the peptide chain significantly influenced the amino acids’ reactivity, a fact requesting consideration when analyzing complex 3D structures like proteins. Since the observed PTMs belong to naturally occurring modifications, it can be assumed that CPPs derive their impact on biological systems in part via protein modification. The presented data pave the way for further studies to evaluate their impact on cellular signaling pathways in inflammatory processes in cancer and wound biology in full.

Funding from the German Federal Ministry of Education and Research [Grant Nos. 03Z22DN11 (S.B.) and 03Z22DN12 (K.W.)] supported this work.