Skip to main content
NCBI home page
ACS Omega logoLink to ACS Omega
. 2019 Aug 23;4(11):14517–14525. doi: 10.1021/acsomega.9b01749

Characterization of Ultraviolet Photoreactions in Therapeutic Peptides by Femtosecond Laser Catalysis and Mass Spectrometry

Simon K Gammelgaard †,‡, Steffen B Petersen ‡, Kim F Haselmann †, Peter Kresten Nielsen †,*
PMCID: PMC6740184  PMID: 31528806

Abstract

graphic file with name ao9b01749_0009.jpg

Peptides and proteins have diverse ultraviolet (UV) photoreaction pathways that can be activated by the energy of the UV photons absorbed. Simple light sources such as lamps are conventionally used to study these photoreactions in solution. This work provides a proof of concept that femtosecond laser technology can function as a highly potent UV source in rapidly conducting UV photostability studies of peptides. Correspondingly, sufficient quantities of photoproducts were generated in 1 min or less, allowing for identification of known and new photomodifications in the therapeutic peptides somatostatin-14 and arginine vasopressin. Identical photoproducts were also generated with a conventional continuous source. The major modifications included N-formylkynurenine, a cross-link between Trp and Phe, a Tyr product with an NH3 loss, and disruption of an unstable disulfide bond into a complex mixture of a trisulfide bond and multiple scrambled dimeric products. In conclusion, femtosecond lasers are extremely useful to drive fast UV-induced reactions for high throughput screening of photostability and modifications in amino acid polymers.

Introduction

Peptides and proteins have intrinsic chromophores that make them vulnerable to the exposure of ultraviolet (UV) light in various situations. In their in vivo environment, these endogenous biomolecules in superficial tissues such as skin or the eye lens are exposed to UV irradiation from sunlight.1 Biopharmaceuticals on the other hand can be exposed to UV light during their manufacture and administration.2 In applications where the spectroscopic properties of the amino acid residues are applied, for example, detection in chromatography, UV wavelengths are used. During these kinds of exposure, the molecules may suffer from oxidation, misfolding, and aggregation, which can be detected as abundant artifacts in analysis or, in the case of light-stressed IgG1, cause immunogenic response.3−7 Understanding the interaction between UV light and amino acid polymers can therefore help unravel the photoprocesses that occur to important proteins in vivo and increased understanding of the photochemical reactions can aid the design of more stable biopharmaceuticals. Predicting photostability and photoreactions in peptides and proteins remains challenging and requires more empirical studies in which powerful UV sources, such as femtosecond lasers, could be an advantage to use.

The amino acids with the strongest UV chromophoric side chains are Trp and Tyr. The aromatic side chains have local absorption maxima close to 280 nm resulting in this wavelength being used to illuminate peptides and proteins for detection and spectroscopic analysis. At 280 nm, the predicted molar extinction coefficient of Trp (5500 M–1 cm–1) is almost 4-fold higher than Tyr (1490 M–1 cm–1).8 Upon photoexcitation, new reaction pathways are unlocked that can lead to covalent modifications. The excitation of Trp or Tyr can be passed on to molecular oxygen in the solution to generate reactive singlet oxygen, which can modify the aromatic side chain.1,3 Among the most common photo-oxidation products derived from Trp is N-formylkynurenine (NFK). Photoexcitation can also induce photoionization of the Tyr or Trp side chain, where the excited electron leaves behind the aromatic group as a radical cation intermediate that can further react with nearby residues.9,10 The last chromophore that can absorb 280 nm UV light is the disulfide group (125 M–1 cm–1).8 The direct absorption of a UV photon can induce homolytic cleavage of the disulfide bond to form two intermediate thiyl radicals (−S•).11 The photoionization of Tyr or Trp has also been suggested to cleave the disulfide bond in a one-electron reduction from the ejected electron.12−14 These intermediates can react into thioether cross-links, free thiols, scrambled disulfides, and aromatic side chain cross-links such as di-Tyr.12,15−22

Multiple in-solution photochemistry studies of peptides and proteins used continuous wave xenon lamps as the UV light source.12,13,16,17,19,22−27 Xenon has a strong UV output and is approved by the International Conference on Harmonisation (ICH) document Q1B as a UV light source for photostability studies.28 Discontinuous light sources, such as femtosecond lasers, have been used to a lesser extent for in-solution photoreactions of peptides and proteins. Femtosecond lasers produce short but extremely powerful pulses and therefore possess the capacity to UV-irradiate many fold more powerful than the currently used lamps. Femtosecond laser technology has been used in biomolecular applications to generate stable UV-induced cross-links for immobilizing proteins onto biosensor surfaces or to trap and study interactions between peptides, proteins, and DNA–protein complexes.29−33

The aim of this work was to explore if femtosecond laser technology could be used as an efficient UV light source in photostability studies of peptides and for generating high yields of the photodegraded products for characterization of photomodifications. Femtosecond laser pulsing was compared with continuous irradiation from a xenon lamp coupled to a diffraction grating in the 280 nm photodegradation of the model tripeptide Gly–Trp–Gly (GWG). Photostability studies of the therapeutic hormone peptides somatostatin-14 (1 Trp, 1 disulfide bond) and arginine vasopressin (1 Tyr, 1 disulfide bond) were efficiently conducted using the femtosecond laser as the UV source at 280 nm. The rapidly generated major photoproducts were characterized by liquid chromatography–mass spectrometry (LC–MS), tandem MS (LC–MS/MS), and absorption spectroscopy. With shorter irradiation times using the femtosecond laser, known modifications such as NFK and thioether cross-linking were generated that were consistent with previous work using conventional sources. A known modification was revealed to exhibit new characteristics as dimeric photoproducts with complex sulfur patterns containing a combination of thioether cross-linking and disulfide scrambling. Additionally, new modifications were identified such as a Phe–Trp cross-link, a Tyr-related product involving loss of NH3, and a trisulfide bond.

Results and Discussion

Femtosecond Laser-Enhanced Photodegradation of a Trp-Containing Model Peptide

The performance of a femtosecond laser was compared to a xenon lamp-diffraction grating setup in a spectrofluorometer to evaluate the capabilities of the laser in accelerating photoreactions. The setups allowed comparable irradiation conditions. At 280 nm [±13 nm full width at half-maximum (fwhm)], the xenon lamp had a maximum power output of ∼1.1 mW that was used for the irradiation experiments. With the laser setup at the same wavelength maximum (±7 nm fwhm), an average output of ∼40 mW was used, which is a 36-fold increase in power compared to the xenon lamp.

Using the tripeptide GWG for the 280 nm photodegradation ensured that the only side chain damaged by the exposure to UV light was the indole group of Trp, which was tracked by measuring Trp fluorescence at 280 nm excitation. The photodegradation of GWG was accelerated with laser irradiation compared to lamp irradiation (Figure 1). The computed half-life (from a single exponential decay) was 1356 s using the lamp, while laser decreased it to 34 s. This was a 40-fold decrease in half-life that quantitatively matched the increase in power provided by the femtosecond laser.

Figure 1.

Figure 1

Open in a new tab

Photodegradation of GWG at 280 nm maximum. Y-axis is the intensity of Trp emission at 350 nm. The inset is the same plot but with irradiation time up to 90 s. Emission was measured each second during xenon lamp irradiation. Emission of the laser-irradiated samples was measured offline. Data were fitted to one phase decay. Half-life using the xenon lamp was 1356 s (22 min 36 s) and laser was 34 s.

The photoionization of Trp leading to a solvated electron can be induced by both mono- and biphotonic processes.34,35 In monophotonic reactions leading to photoproducts, linear correlations should be observed between irradiation power and photoconversion of precursor into product. Also, pulsing sources should induce the same photoconversion as a continuous source at the same average power. In the photodegradation of the model peptide GWG, the correlation between the increase in power and the decrease in half-life using the pulsing laser corresponded to a one-photon reaction.

Rapid Laser-Induced Photodegradation of Somatostatin-14 Versus Arginine Vasopressin

Somatostatin-14 (sequence: AGCKNFFWKTFTSC) is a fourteen residue peptide with one Trp residue and one disulfide bond. All degradation of the peptide was localized to the Trp residue by comparing quantification of Trp fluorescence and ultrahigh performance liquid chromatography (UHPLC) quantification of the intact molecule (Figure 2A).

Figure 2.

Figure 2

Open in a new tab

Laser-induced degradation (280 nm maximum at 40 mW) of somatostatin-14 (1 Trp, 1 disulfide) and vasopressin (1 Tyr, 1 disulfide). Quantity (y-axis) of the peptide precursor was determined by the area at 215 nm absorbance in UHPLC and by the intensity of Trp (350 nm) or Tyr (310 nm) emission. (A) Somatostatin-14 data was fitted to one-phase decay. (B) Vasopressin Tyr emission was fitted to one-phase decay and 215 nm UHPLC area was fitted to two-phase decay.

Arginine vasopressin (sequence: CYFQNCPRG) is a nonapeptide with one Tyr residue and one disulfide bond. Interestingly, the photoinstability profile of vasopressin was different compared to somatostatin-14. It required 4 min of laser irradiation of somatostatin-14 to degrade 33% precursor, while for vasopressin, it only took 30 s to degrade 34%, corresponding to an 8-fold higher photostability of somatostatin-14 versus vasopressin. This photodegradation dissimilarity is in contrast to the almost 4-fold higher 280 nm extinction coefficient of Trp versus Tyr. At 34% degradation of intact vasopressin, Tyr fluorescence degraded 3% (Figure 2B). The gap in quantification reflected that most of the photoinstability of vasopressin was not due to the Tyr residue, even though it was the strongest UV chromophore in the molecule. Interestingly, vasopressin displayed a biexponential decay with fast degradation in the first 30 s, while somatostatin-14 displayed single exponential decay. This initial faster degradation supported a concentration-dependent photoreaction that dominated the photoinstability of vasopressin.

Photoreaction and Products of Somatostatin-14

LC–MS characterization of the photoproducts of somatostatin-14 supported that all modifications were located within the Trp residue. All identified photoproducts that were generated by femtosecond laser pulsing were also generated using the lower-powered continuous xenon lamp (data not shown) and were therefore not specific to the laser irradiation.

Photoreaction of Trp to NFK in somatostatin-14 was achieved. The NFK photoproduct was identified by a mass gain of two oxygen atoms and a characteristic ∼323 nm absorption maximum (Figures 3 and S1). Also, with tandem MS fragmentation, the mass gain of two oxygen atoms was localized to the Trp residue (Figure S2). The photoproduct quantity after 1 minute was sufficient for LC–UV–MS analysis, illustrating that the laser was efficient in the production of NFK-somatostatin-14 and the remaining major photoproducts. Trp to NFK photoconversion in somatostatin-14 has previously been reported after exposing the peptide to fluorescent light for 10 days.36 Despite NFK being a very common photo-oxidation product, it was not the dominant photomodification.

Figure 3.

Figure 3

Open in a new tab

UHPLC–MS analysis of the 280 nm photoreaction of somatostatin-14 (1 Trp, 1 disulfide) using the femtosecond laser (40 mW). “M” represents the mass of the precursor. Asterisk (*) represents spectroscopic modification of the Trp residue analyzed by online UPLC absorbance spectroscopy (Figure S1).

Two other major photoproducts with isobaric somatostatin-14 masses were identified (retention time ∼10.5 min, Figure 3). Interestingly, the absorption spectra showed that the classical Trp absorption profile with a 280 nm local maximum was absent. For the isobaric photoproduct of the largest quantity, the absorption spectrum revealed two new local maxima of ∼251 and ∼284 nm (Figure S1). The loss of Trp fluorescence (Figure 2A) and alteration to two new absorption maxima could be the result of new chromophores created by photoinduced cross-linking between Trp and another aromatic residue. The tandem MS fragmentation data were consistent with the spectroscopy data. In reduced somatostatin-14, sequence coverage with b and y ions next to Trp8 was obtained, but for the reduced isobaric photoproduct, the coverage of Trp8 was worsened (Figure 4). Especially the b7 and y7 ions, which are formed from a fragmentation of the peptide bond between Phe7 and Trp8, were diminished. Usually such a decrease or disappearance of sequence coverage in MS/MS analysis derives from sequences within cyclic structures. In collision-induced fragmentation, a cyclic structure formed by a cross-link between Phe7 and Trp8 could give a fragment from the back-bone cleavage of the closest covalent bond to the cyclic structure, which is the N–Cα bond in Phe7. Such a fragment of somatostatin-14 is the c6 ion (b6 + NH3) which was strongly increased in the reduced isobaric photoproduct (Figure 4). Combining the data, it was likely that the ∼251 nm chromophore was the modified Phe7, and the ∼284 nm chromophore was the modified Trp8. Cross-linking between the two aromatic side chains permits the formation of multiple isomers and explains why two photoproducts were detected.

Figure 4.

Figure 4

Open in a new tab

MS characterization of somatostatin-14 and the major photoproduct in their reduced form. (A) Higher energy collisional dissociation (HCD) MS/MS spectra at 26% collision energy of the 2+ charged clusters. Arrows indicate either a decrease or increase of intensity of the specific fragment compared to the precursor. (B) Sequence of somatostatin-14 and summary of the identified fragments.

The photoproducts containing the Phe7–Trp8 cross-link were challenging to identify because of the identical mass to the native peptide. The total mass of the two side chains could be preserved through an atomic rearrangement induced by photonic excitation of Trp8 followed by photoionization. Thus, the indole group of Trp8 transfers an electron and proton to the phenyl group of Phe7 to reduce a double bond. The neutral radicals of indolyl and phenyl form the observed cross-link.

The spatial proximity between Phe and Trp in somatostatin-14 enables cross-linking upon UV light exposure and may also occur in other peptides or proteins with similar positions of these residues. By using the efficient UV laser source for a limited irradiation time, LC–UV–MS analysis was capable of identifying the Phe–Trp photo-cross-link, which is reported for the first time here.

Overview of the Photoreaction and Products of Arginine Vasopressin

Analyzing the photoreaction of vasopressin with LC–MS revealed why the photodegradation of Tyr did not match the total laser-induced degradation of the peptide (Figure 2B). A complex mixture of diverse photoproducts was observed where the original disulfide group was modified by the UV irradiation (Figure 5). The sulfur photoproducts could be categorized into either monomeric or dimeric species. The monomeric species were all modified in the number of sulfur atoms in the original disulfide bridge. The identifications were a monosulfide (M – S, RSR), a trisulfide (M + S, RSSSR), a trisulfide trioxide (RS(SO3)SR), and a photoproduct with a mass loss corresponding to an NHS moiety (M – NHS). The identified dimeric species had the mass of two oxidized vasopressin molecules (2M), the mass of two oxidized molecules with loss of two hydrogen atoms (2M – 2H) or the mass of two molecules with loss of an NH3S moiety (2M – NH3S). The identified photoproducts were also generated using the lower-powered continuous xenon lamp (data not shown) and were therefore not specific to the femtosecond laser pulsing.

Figure 5.

Figure 5

Open in a new tab

UHPLC–MS analysis of the 280 nm photoreaction of vasopressin (1 Tyr, 1 disulfide) using femtosecond laser (40 mW). “M” represents the mass of the precursor. Asterisk (*) represents spectroscopic modification of the Tyr residue analyzed by online UPLC absorbance spectroscopy (Figure S3).

The Tyr residues of the sulfur photoproducts were intact in the ultraperformance liquid chromatography (UPLC) absorption analysis. In one major photoproduct, M – NH3 (retention time 49.5 min, Figure 5), a modulated Tyr absorption profile with a new maximum of ∼294 nm was identified (Figure S3). The mass loss, modification of the aromatic side chain, and the proximity of Tyr to the N-terminus, all together resembled a previous UV photostudy of GWG, where the same mass loss was observed.37

LC–MS analysis of the 30 s sample (Figure 5) revealed two 2M photoproducts that dominated the photodegradation reflecting a concentration-dependent photoreaction (Figure 2B). At the same time, the remaining photoproducts, such as the trisulfide, 2M – 2H species, and M – NH3, had a linear correlation between irradiation time and quantity.

UV Light-Induced Trisulfide Modification

Monosulfide (thioether) and trisulfide vasopressin were the initial photoproducts in the chromatogram (Figure 5). Generation of a monosulfide from a disulfide bond upon UV irradiation has previously been reported for the vasopressin homologue oxytocin.15 However, there is no detailed information that UV light can introduce trisulfides. High-resolution MS allowed us to distinguish mass shifts corresponding to two oxygen atoms (31.98 Da) or to one sulfur atom (31.97 Da). The sulfur contents in mono-, di-, and trisulfide vasopressin were also confirmed by collision-induced fragmentation. In the fragmentation of a disulfide group, the sulfur bridge can be cleaved at either the C–S or S–S bonds to yield a sulfur triplet pattern.38 In the vasopressin disulfide group, distributed cleavage of the three bonds in the (Cys1)CH2–S–S–CH2(Cys6) bridge gave a sulfur triplet pattern of the y4 ion corresponding to zero, one, or two sulfur atoms in the y4 ion (Figure 6). The identification of a y4 + S (y4 + 31.97 Da) ion confirmed that two sulfur atoms were present in the disulfide bridge. In the monosulfide group, the (Cys1)CH2–S–CH2(Cys6) bridge was supported by the complete disappearance of the y4 + S ion. In the trisulfide group, the (Cys1)CH2–S–S–S–CH2(Cys6) bridge was supported by identification of y4 + S and y4 + S2 ions.

Figure 6.

Figure 6

Open in a new tab

Characterization by MS of the sulfur structure in disulfide vasopressin and its two photoproducts monosulfide- and trisulfide vasopressin. (A) HCD MS/MS spectra at 40% collision energy of the 2+ charged clusters. (B) Sequence of vasopressin and description of the observed sulfur variants in its y4 fragment.

The laser results efficiently demonstrated that a trisulfide bridge can be formed in a disulfide bond-containing peptide due to UV light exposure. This could have implications, for example, in the field of recombinant monoclonal antibodies where UV light exposure occurs in manufacturing (e.g., during UV detection in chromatography) and because antibodies have been reported to harbor trisulfides.39,40

UV Light-Induced Disulfide Bond Scrambling

The dimeric species carried scrambled intermolecular disulfide bonds because the species were depleted under reducing conditions (Figure 7A). In the MS/MS analysis of the dimeric products, fragmentation at the intermolecular disulfide bonds was observed which resulted in monomeric fragments of varying sulfur contents, revealing the number of intermolecular disulfide bonds (Figure S4). The intact mass of 2M observed for two photoproducts supported that all cysteine residues participated in two intermolecular disulfide bonds in different patterns, while monomeric fragments with two additional sulfur atoms confirmed the two intermolecular disulfide bonds (Figure S4, first spectrum). With four Cys residues in a homodimer, the disulfide bond pattern can arrange the monomers in parallel or antiparallel, which is why two 2M species were observed. The intact mass of 2M – 2H supported that one intermolecular disulfide bond was present and that the monomers (M – H) each had one cross-linking thioether that each contributed to the loss of one hydrogen atom. The monomeric fragment with only one additional sulfur atom was present, confirming the one intermolecular disulfide bond (Figure S4, second spectrum). The 2M – NH3S photoproducts also contained one intermolecular disulfide bond. Additionally, cleavage of the bond gave rise to a monomeric M – NHS fragment as well as a fragmentation pattern similar to 2M – 2H (Figure S4, third spectrum). It is therefore likely that the total loss of NH3S was a combination of two modifications that together scrambled to dimeric species.

Figure 7.

Figure 7

Open in a new tab

UHPLC–MS analysis of the vasopressin sample that was laser-irradiated (280 nm, 40 mW, 5 min) followed by (A) reduction with TCEP or (B) reduction with TCEP and alkylation with IAA. “M” represents the mass of the precursor. Alkylated sulfur atoms are annotated with asterisk (*). (B) Cys6 dithiohemiacetal structural isomer is designated by (1) and the Cys1 dithiohemiacetal structural isomer by (2).

Surprisingly, disulfide scrambling was the dominating photodamage in vasopressin, found as intermolecular bridges causing the dimeric structures in the abundant 2M and 2M – 2H species. The formation of scrambled disulfide bonds by exposure to UV light could be a secondary reaction following the initial photoconversion of the native disulfide bond into thiols that subsequently reoxidize into non-native disulfide bonds. The UV photoreaction of vasopressin illustrated that the photoconversion of disulfide bonds to free thiols can be the primary photoreaction in a peptide or protein. Disulfide scrambling induced by UV light exposure has also been observed in a monoclonal antibody and human growth hormone.18,19 However, it has not previously been reported that photoproducts with a thioether can participate in scrambling due to the initial presence of free thiols from the other sulfur atom in the original disulfide group.

Irreversible Photomodification of the Disulfide Bond: the Dithiohemiacetal Isomers

Further characterization of 2M – 2H unraveled in more detail the complex sulfur structure. In short, the data supported that the vasopressin disulfide group (R1CH2–SS–R2) could photoreact into an irreversible dithiohemiacetal (R1CH(−SH)–S–R2), as previously reported for the vasopressin homologue oxytocin.15 The dithiohemiacetal group in a peptide is a β-carbon from one of the two cysteine side chains bound to a primary sulfur (thiol, +H) and secondary sulfur (thioether, −H) as a chiral center. If the photogenerated thiol in the dithiohemiacetal reoxidizes to a non-native disulfide bond with another dithiohemiacetal, the resulting mass is 2M – 2H.

While a reduction of the 2M species would result in vasopressin with two thiols (M + 2H), a reduction of the 2M – 2H species resulted in monomers with one thiol and the mass of M. A large population with the mass of M was therefore stable under the reducing conditions (retention time 13 min, Figure 7A). Alkylation of the reduced laser-irradiated sample confirmed that only one thiol was present in the irreversible monomeric photoproducts (Figure 7B). The alkylation of the thiol revealed that the chiral β-carbon in dithiohemiacetal was distributed between the side chains of Cys1 and Cys6 to two structural isomers. The first dithiohemiacetal species to elute in the chromatography (retention time 11 min, Figure 7B) was identified to contain the chiral β-carbon in the Cys6 side chain while the last to elute (retention time 14 min) contained the chiral β-carbon in Cys1. Characterization of the sulfur patterns that identified and described the dithiohemiacetal modification and their isomers was possible because the thioether between the β-carbons was cleaved to yield a2 and y4 ions that correspondingly covered the Cys1 and Cys6 residues. Thus, Cys1-dithiohemiacetal vasopressin fragmented into the alkylated a2 ion and conventional y4 ion, while Cys6-diothiohemiacetal oppositely fragmented into the conventional a2 ion and alkylated y4 ion (Figure S5). Because the dithiohemiacetals had a chiral β-carbon, both R/S-enantiomers could be produced from each of the two structural isomers to yield a total of four isomers. This was confirmed by the additional two isomers found in the second peak in the chromatogram (retention time 11.5 min, Figures 7B or S5), which contained a partial co-elution of the fragment pairs from Cys1- or Cys6-dithiohemiacetal. Four dithiohemiacetal isomers oxidizing into intermolecular disulfide bonds also explained why there were ten 2M – 2H species under nonreducing conditions (Figure 8A, second panel). The two Cys sites photomodified to dithiohemiacetal groups in the dimeric isomers were identified by tracking the conventional a2 and y4 ions (which only include unmodified Cys residues) in the fragmentation of all ten isomers (Figure 8A, third and fourth panels). The distribution of the two dithiohemiacetal groups reoxidized to a scrambled disulfide bond enables three possible structural isomers of dimeric dithiohemiacetal-vasopressin (Figure 8B). For the dithiohemiacetal-vasopressin structural heterodimer (Cys1 + Cys6 dithiohemiacetal), all four possible stereoisomers (2n) from the two chiral centers (n = 2) were confirmed in the MS/MS characterization by the simultaneous presence of a2 and y4 ions. For each of the two dithiohemiacetal structural homodimers (2 × Cys1- or 2 × Cys6 dithiohemiacetal), all three possible stereoisomers (2n – 1) were observed by the exclusive presence of the a2 or y4 ion. The loss of the fourth stereoisomer in the homodimers is because the R,S configuration in the homodimers is a meso structure that because of symmetry constitutes one isomer.

Figure 8.

Figure 8

Open in a new tab

(A) UHPLC–MS and −MS/MS analysis of the disulfide-scrambled dithiohemiacetal (2M – 2H) species from the laser-irradiated (280 nm, 40 mW, 5 min) sample of vasopressin. Cys annotated with asterisk (*) contains a dithiohemiacetal isomer. For the 2M – 2H species, the extracted ion chromatogram (XIC) m/z range was 722.95–722.97, which covered the most abundant isotopic peak in the 3+ charged cluster. The XICs of the MS/MS fragmentations of the 3+ charged clusters covered the monoisotopic peaks of the 1+ charged fragments: a2 (239.08–239.09), y4 (431.21–431.23). The MS/MS fragmentations were at 30% collision energy. (B) Deduced structural isomers of dimeric dithiohemiacetal-vasopressin (2M – 2H) photoproducts. Asterisk (*) here designates the chiral carbon atoms in the dithiohemiacetal structural isomers.

The characterization and identification of the dithiohemiacetal photomodification was not straightforward. One reason was that the thiol reoxidized into the more complex 2M – 2H species, second was that the reduced monomeric form of the dithiohemiacetal group has identical mass to its precursor disulfide group, and the third was that the thioether linkage between the original Cys1 and Cys6 disrupts MS/MS fragmentation. The formation of the dithiohemiacetal group upon photoexcitation of a disulfide bond has previously been identified in a 12 residue model peptide, IgG1 antibody, and human growth hormone.18,41,42 This diversity in the molecular formats emphasizes that the photoconversion of the disulfide group to a dithiohemiacetal is not a molecule-specific photomodification but relies on fundamental reaction mechanisms induced by UV light. The photoreaction was proposed to be initiated by a direct photoexcitation of the disulfide group which dissociates to thiyl radicals (−S•) that cause the atomic rearrangements. The details of the mechanisms are described elsewhere.15,41 Our LC–MS analysis showed that the dithiohemiacetal is a dominant photomodification of the disulfide group consistent with previous results.15,18,41 There is no prior information of UV light-induced dithiohemiacetal isomers. The isomers may be challenging to separate as was the case for reduced and alkylated dithiohemiacetal-vasopressin. However, in one UV photochemistry study, an additional photoproduct was found with an isobaric mass to the identified dithiohemiacetal product.41 The authors reported that the MS/MS data were too poor for detailed characterization. It is likely that the uncharacterized photoproduct was an isomeric variant of dithiohemiacetal. Dithiohemiacetals can potentially be harmful photomodifications regarding the function of the peptide or protein. The modification is irreversible due to the thioether cross-link and the free thiol can introduce nonnative disulfide bonds as have been seen for the 2M – 2H species of vasopressin.

Conclusions

Our evaluation of the femtosecond laser delivered promising results and demonstrated that the technology can function as a highly advantageous UV light source in driving fast photoreactions of amino acid polymers. Using a continuous wave source, it was shown that by moving to femtosecond laser technology, the photochemistry was accelerated by a factor of 40, roughly proportional to the measured power ratio between the two UV light sources. With the increased acceleration, photostability studies can be conducted within few minutes depending on the molecule and the wanted extent of degradation. Analyzing the same samples, sufficient quantities were rapidly generated of a diverse set of known and novel photomodifications without any additional sample treatment before the chromatographic UV detection and characterization by MS. Two recent UV photoreaction studies of monoclonal antibodies used, for the LC–MS analysis, irradiation times between 15 and 60 min at the wavelengths of 254 and 305 nm maximum, or 24 h following the ICH irradiation procedure.19,43 A recent photodegradation study of Trp- and disulfide-containing model peptides at 280 nm maximum reported irradiation times of 2–3 h for about 30% degradation of Trp fluorescence.27 Here, the 30% laser degradation of Trp fluorescence of somatostatin-14 was achieved after roughly 3.5 min, corresponding to a 34 to 51-fold shorter irradiation time. In a previous photodegradation of somatostatin-14, it was reported that LC–MS analysis after 20 min irradiation at 305 nm maximum did not detect any significant difference.36 In comparison, using the femtosecond laser, it is probable that an irradiation time of 1 min or less would be adequate for stability screening and analysis of the major photoproducts of peptides and proteins. The sample volume can be decreased to further optimize the irradiation times.

Using the femtosecond laser, the presented data unexpectedly illustrated that there is not necessarily a clear correlation between chromophoric strength and photoreactivity into covalent products. Even though somatostatin-14 has a higher predicted extinction coefficient, it was still a more UV photostable peptide than vasopressin. Also, the remarkable difference in sulfur photoreactivity between the intramolecular disulfide bonds in somatostatin-14 and vasopressin interestingly exemplified that the photoreaction pathways of the same chromophore can be highly molecule-specific. In somatostatin-14, there was no photoreactivity of its disulfide bond, while for vasopressin, its disulfide bond was the most unstable 280 nm chromophore in both peptides. This was interesting because the 280 nm extinction coefficient of the disulfide bond is expected to be 12-fold lower than Tyr and 44-fold lower than Trp. Overall, the photostability of an amino acid polymer should not exclusively be predicted by the chromophores in the sequence.

All of the identified photoproducts of somatostatin-14 and vasopressin that were generated by femtosecond laser pulsing were also generated using the lower-powered continuous xenon lamp. This supported that the diverse photoreactions in these peptides were monophotonic processes. Promisingly, without introducing artifact photoproducts, the more powerful pulse sources could therefore function as viable UV light sources and replace continuous sources for higher experimental throughput in photoreaction studies of therapeutic peptides and proteins.

Experimental Section

Sample Preparation

GWG, somatostatin-14, and arginine vasopressin (vasopressin) were synthesized in-house. Peptides were solubilized in 20 mM sodium phosphate buffer (pH 7.4) and diluted to relevant concentrations: GWG (40 μM), somatostatin-14 (500 μM), and vasopressin (500 μM).

Sample Irradiation with UV Light

Samples were exposed to continuous wave irradiation or femtosecond pulsing at 280 nm maximum. Power of the irradiation beams were measured with a power meter (StarLite Meter, Ophir, Israel). The spectra of the beams were collected with a spectrometer (AvaSpec-HS-TEC, Avantes, Netherlands) for the determination of the wavelength spread at fwhm. Samples of 3 mL were irradiated in a 1 cm path length in fluorescence macrocuvettes (101-QS, Hellma Analytics, Germany) at 21 °C while magnetically stirred. Aliquots were stored at −18 °C.

For continuous irradiation, the excitation setup in a spectrofluorometer (RTC 2000, Photon Technology International, Canada) was used. In the setup, the UV light source was a 75 W xenon arc lamp coupled to a diffraction grating for wavelength selection. When excitation slits were set to 6 mm for the highest 280 nm output (±13 nm fwhm), power was ∼1.1 mW at the entrance port to the sample chamber. The steady-state fluorescence mode was used to conduct the continuous irradiation.

For the femtosecond laser irradiation, a Tsunami XP (Spectra-Physics, CA, USA) laser was pumped by a Millennia eV laser (Spectra-Physics, CA, USA) to generate 840 nm pulses of ∼100 femtosecond duration at a repetition rate of 80 MHz. The 840 nm output was frequency-tripled to 280 nm by a UHG module (Spectra-Physics, CA, USA). The pulse beam at 280 nm maximum (±7 nm fwhm) was tuned to a stable average output of ∼40 mW at the sample holder.

Quantification of Photodegradation by Fluorescence Spectroscopy

The spectrofluorometer used for the continuous wave irradiation was also used for quantifying the degraded intensities of Trp or Tyr fluorescence. Trp or Tyr emission intensities were collected at 280 nm excitation. During the xenon lamp irradiation of GWG (using the steady-state fluorescence mode), emission was also collected each second. For the laser-irradiated samples, emission spectra were collected with the emission scan mode at 1 nm resolution.

Sample Treatment after UV Irradiation

Reduced samples were prepared by incubation with 5 mM tris(2-carboxyethyl)phosphine (TCEP) for 30 min at room temperature. Reduced and alkylated samples were prepared by incubation with 5 mM TCEP and 5 mM iodoacetamide (IAA) for 30 min at room temperature.

LC–UV–MS Analysis of the UV-Irradiated Samples

An Orbitrap Fusion Lumos mass spectrometer (Thermo Fisher Scientific, CA, USA) was coupled to a Vanquish Horizon UHPLC system with a VF-D40 detector (Thermo Fisher Scientific, CA, USA). A Synapt G2-Si (Waters, UK) was coupled to an Acquity UPLC Classic system with the Acquity Photodiode Array (PDA) detector (Waters, UK). Reverse-phase chromatography was performed with the Acquity UPLC CSH C18 column, 1.0 × 150 mm, 1.7 μm (Waters, UK) using solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile) as the mobile phase. The remaining procedures and settings were standard.

Software and Data Analysis

Xcalibur Qual Browser 3.0.63 (Thermo Fisher Scientific, CA, USA), MassLynx 4.1 (Waters, UK), and GPMAW 9.51 (Lighthouse Data, Denmark) were used to analyze the LC–UV–MS data. GraphPad Prism 7.04 was used for graphs and fitting. Accelrys Draw 4.1 SP1 was used to draw molecular structures.

Acknowledgments

This work was funded by the Novo Nordisk STAR Fellowship Programme and Innovation Fund Denmark. We thank Heidi Jensen (Novo Nordisk A/S) for the technical support and Jesper Lau (Novo Nordisk A/S) for the synthesized peptides.

Glossary

Abbreviations

UV

ultraviolet

NFK

N-formylkynurenine

ICH

International Conference on Harmonisation

GWG

Gly–Trp–Gly

LC

liquid chromatography

MS

mass spectrometry

fwhm

full width at half-maximum

UHPLC

ultrahigh performance liquid chromatography

UPLC

ultra performance liquid chromatography

TCEP

tris(2-carboxyethyl)phosphine

IAA

iodoacetamide

PDA

photodiode array

HCD

higher energy collisional dissociation

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsomega.9b01749.

  • UPLC absorption spectra from PDA detection of photoproducts with modified Trp or Tyr; MS/MS spectrum of NFK-somatostatin-14; Representative MS/MS spectra of the three types of dimeric photoproducts of vasopressin; XICs by either full MS or MS/MS scans of reduced and alkylated dithiohemiacetal-vasopressin (PDF)

The authors declare no competing financial interest.

Supplementary Material

ao9b01749_si_001.pdf (611.3KB, pdf)

References

  1. Pattison D. I.; Rahmanto A. S.; Davies M. J. Photo-oxidation of proteins. Photochem. Photobiol. Sci. 2012, 11, 38–53. 10.1039/c1pp05164d. [DOI] [PubMed] [Google Scholar]
  2. Kerwin B. A.; Remmele R. L. Jr. Protect from light: photodegradation and protein biologics. J. Pharm. Sci. 2007, 96, 1468–1479. 10.1002/jps.20815. [DOI] [PubMed] [Google Scholar]
  3. Davies M. J. Singlet oxygen-mediated damage to proteins and its consequences. Biochem. Biophys. Res. Commun. 2003, 305, 761–770. 10.1016/s0006-291x(03)00817-9. [DOI] [PubMed] [Google Scholar]
  4. Roy S.; Mason B. D.; Schöneich C. S.; Carpenter J. F.; Boone T. C.; Kerwin B. A. Light-induced aggregation of type I soluble tumor necrosis factor receptor. J. Pharm. Sci. 2009, 98, 3182–3199. 10.1002/jps.21750. [DOI] [PubMed] [Google Scholar]
  5. Xie J.; Qin M.; Cao Y.; Wang W. Mechanistic insight of photo-induced aggregation of chicken egg white lysozyme: the interplay between hydrophobic interactions and formation of intermolecular disulfide bonds. Proteins 2011, 79, 2505–2516. 10.1002/prot.23074. [DOI] [PubMed] [Google Scholar]
  6. Schweikart F.; Hulthe G. HPLC-UV-MS Analysis: A Source for Severe Oxidation Artifacts. Anal. Chem. 2019, 91, 1748–1751. 10.1021/acs.analchem.8b05845. [DOI] [PubMed] [Google Scholar]
  7. Bessa J.; Boeckle S.; Beck H.; Buckel T.; Schlicht S.; Ebeling M.; Kiialainen A.; Koulov A.; Boll B.; Weiser T.; Singer T.; Rolink A. G.; Iglesias A. The immunogenicity of antibody aggregates in a novel transgenic mouse model. Pharm. Res. 2015, 32, 2344–2359. 10.1007/s11095-015-1627-0. [DOI] [PubMed] [Google Scholar]
  8. Pace C. N.; Vajdos F.; Fee L.; Grimsley G.; Gray T. How to measure and predict the molar absorption coefficient of a protein. Protein Sci. 1995, 4, 2411–2423. 10.1002/pro.5560041120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bent D. V.; Hayon E. Excited state chemistry of aromatic amino acids and related peptides. I. Tyrosine. J. Am. Chem. Soc. 1975, 97, 2599–2606. 10.1021/ja00843a002. [DOI] [PubMed] [Google Scholar]
  10. Bent D. V.; Hayon E. Excited state chemistry of aromatic amino acids and related peptides. III. Tryptophan. J. Am. Chem. Soc. 1975, 97, 2612–2619. 10.1021/ja00843a004. [DOI] [PubMed] [Google Scholar]
  11. Soorkia S.; Dehon C.; Kumar S. S.; Pedrazzani M.; Frantzen E.; Lucas B.; Barat M.; Fayeton J. A.; Jouvet C. UV Photofragmentation Dynamics of Protonated Cystine: Disulfide Bond Rupture. J. Phys. Chem. Lett. 2014, 5, 1110–1116. 10.1021/jz500158j. [DOI] [PubMed] [Google Scholar]
  12. Prompers J. J.; Hilbers C. W.; Pepermans H. A. M. Tryptophan mediated photoreduction of disulfide bond causes unusual fluorescence behaviour of Fusarium solani pisi cutinase. FEBS Lett. 1999, 456, 409–416. 10.1016/s0014-5793(99)00990-4. [DOI] [PubMed] [Google Scholar]
  13. Neves-Petersen M. T.; Gryczynski Z.; Lakowicz J.; Fojan P.; Pedersen S.; Petersen E.; Bjørn Petersen S. High probability of disrupting a disulphide bridge mediated by an endogenous excited tryptophan residue. Protein. Sci. 2002, 11, 588–600. 10.1110/ps.06002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Neves-Petersen M. T.; Klitgaard S.; Pascher T.; Skovsen E.; Polivka T.; Yartsev A.; Sundström V.; Petersen S. B. Flash photolysis of cutinase: identification and decay kinetics of transient intermediates formed upon UV excitation of aromatic residues. Biophys. J. 2009, 97, 211–226. 10.1016/j.bpj.2009.01.065. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Mozziconacci O.; Schöneich C. Photodegradation of oxytocin and thermal stability of photoproducts. J. Pharm. Sci. 2012, 101, 3331–3346. 10.1002/jps.23204. [DOI] [PubMed] [Google Scholar]
  16. Vanhooren A.; Devreese B.; Vanhee K.; Van Beeumen J.; Hanssens I. Photoexcitation of Tryptophan Groups Induces Reduction of Two Disulfide Bonds in Goat α-Lactalbumin†. Biochemistry 2002, 41, 11035–11043. 10.1021/bi0258851. [DOI] [PubMed] [Google Scholar]
  17. Permyakov E. A.; Permyakov S. E.; Deikus G. Y.; Morozova-Roche L. A.; Grishchenko V. M.; Kalinichenko L. P.; Uversky V. N. Ultraviolet illumination-induced reduction of α-lactalbumin disulfide bridges. Proteins 2003, 51, 498–503. 10.1002/prot.10371. [DOI] [PubMed] [Google Scholar]
  18. Steinmann D.; Mozziconacci O.; Bommana R.; Stobaugh J. F.; Wang Y. J.; Schöneich C. Photodegradation Pathways of Protein Disulfides: Human Growth Hormone. Pharm. Res. 2017, 34, 2756–2778. 10.1007/s11095-017-2256-6. [DOI] [PubMed] [Google Scholar]
  19. Wecksler A. T.; Yin J.; Lee Tao P.; Kabakoff B.; Sreedhara A.; Deperalta G. Photodisruption of the Structurally Conserved Cys-Cys-Trp Triads Leads to Reduction-Resistant Scrambled Intrachain Disulfides in an IgG1 Monoclonal Antibody. Mol. Pharm. 2018, 15, 1598–1606. 10.1021/acs.molpharmaceut.7b01128. [DOI] [PubMed] [Google Scholar]
  20. Lehrer S. S.; Fasman G. D. Ultraviolet Irradiation Effects in Poly-L-tyrosine and Model Compounds. Identification of Bityrosine as a Photoproduct*. Biochemistry 1967, 6, 757–767. 10.1021/bi00855a017. [DOI] [PubMed] [Google Scholar]
  21. Malencik D. A.; Anderson S. R. Dityrosine formation in calmodulin. Biochemistry 1987, 26, 695–704. 10.1021/bi00377a006. [DOI] [PubMed] [Google Scholar]
  22. Correia M.; Neves-Petersen M. T.; Jeppesen P. B.; Gregersen S.; Petersen S. B. UV-light exposure of insulin: pharmaceutical implications upon covalent insulin dityrosine dimerization and disulphide bond photolysis. PLoS One 2012, 7, e50733 10.1371/journal.pone.0050733. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Weisenborn P. C. M.; Meder H.; Egmond M. R.; Visser T. J. W. G.; van Hoek A. Photophysics of the single tryptophan residue in Fusarium solani Cutinase: evidence for the occurrence of conformational substates with unusual fluorescence behaviour. Biophys. Chem. 1996, 58, 281–288. 10.1016/0301-4622(95)00079-8. [DOI] [PubMed] [Google Scholar]
  24. Vanhooren A.; De Vriendt K.; Devreese B.; Chedad A.; Sterling A.; Van Dael H.; Van Beeumen J.; Hanssens I. Selectivity of Tryptophan Residues in Mediating Photolysis of Disulfide Bridges in Goat α-Lactalbumin†. Biochemistry 2006, 45, 2085–2093. 10.1021/bi0517638. [DOI] [PubMed] [Google Scholar]
  25. Correia M.; Snabe T.; Thiagarajan V.; Petersen S. B.; Campos S. R. R.; Baptista A. M.; Neves-Petersen M. T. Photonic activation of plasminogen induced by low dose UVB. PLoS One 2015, 10, e0116737 10.1371/journal.pone.0144794. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Abaskharon R. M.; Gai F. Direct measurement of the tryptophan-mediated photocleavage kinetics of a protein disulfide bond. Phys. Chem. Chem. Phys. 2016, 18, 9602–9607. 10.1039/c6cp00865h. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Nagy T. M.; Knapp K.; Illyes E.; Timari I.; Schlosser G.; Csik G.; Borics A.; Majer Z.; Kover K. E. Photochemical and Structural Studies on Cyclic Peptide Models. Molecules 2018, 23, 2196. 10.3390/molecules23092196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. ICH Harmonised Tripartite Guideline . Stability Testing: Photostability Testing of New Drug Substances and Products Q1B. https://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q1B/Step4/Q1B_Guideline.pdf (accessed Aug 2019).
  29. Neves-Petersen M. T.; Snabe T.; Klitgaard S.; Duroux M.; Petersen S. B. Photonic activation of disulfide bridges achieves oriented protein immobilization on biosensor surfaces. Protein Sci. 2006, 15, 343–351. 10.1110/ps.051885306. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Gonçalves O.; Snider S.; Zadoyan R.; Nguyen Q. T.; Vorum H.; Petersen S. B.; Neves-Petersen M. T.. Novel microfabrication stage allowing for one-photon and multi-photon light assisted molecular immobilization and for multi-photon microscope. SPIE BiOS; SPIE, The International Society for Optical Engineering: Bellingham, WA, 2017; p 100790F.
  31. Lejnine S.; Durfee G.; Murnane M.; Kapteyn H. C.; Makarov V. L.; Langmore J. P. Crosslinking of proteins to DNA in human nuclei using a 60 femtosecond 266 nm laser. Nucleic Acids Res. 1999, 27, 3676–3684. 10.1093/nar/27.18.3676. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Leo G.; Altucci C.; Bourgoin-Voillard S.; Gravagnuolo A. M.; Esposito R.; Marino G.; Costello C. E.; Velotta R.; Birolo L. Ultraviolet laser-induced cross-linking in peptides. Rapid Commun. Mass Spectrom. 2013, 27, 1660–1668. 10.1002/rcm.6610. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Itri F.; Monti D. M.; Della Ventura B.; Vinciguerra R.; Chino M.; Gesuele F.; Lombardi A.; Velotta R.; Altucci C.; Birolo L.; Piccoli R.; Arciello A. Femtosecond UV-laser pulses to unveil protein-protein interactions in living cells. Cell. Mol. Life Sci. 2016, 73, 637–648. 10.1007/s00018-015-2015-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Stevenson K. L.; Papadantonakis G. A.; LeBreton P. R. Nanosecond UV laser photoionization of aqueous tryptophan: temperature dependence of quantum yield, mechanism, and kinetics of hydrated electron decay. J. Photochem. Photobiol., A 2000, 133, 159–167. 10.1016/s1010-6030(00)00235-5. [DOI] [Google Scholar]
  35. Tsentalovich Y. P.; Snytnikova O. A.; Sagdeev R. Z. Properties of excited states of aqueous tryptophan. J. Photochem. Photobiol., A 2003, 162, 371–379. 10.1016/s1010-6030(03)00376-9. [DOI] [Google Scholar]
  36. Mozziconacci O.; Schöneich C. Effect of conformation on the photodegradation of Trp- and cystine-containing cyclic peptides: octreotide and somatostatin. Mol. Pharm. 2014, 11, 3537–3546. 10.1021/mp5003174. [DOI] [PubMed] [Google Scholar]
  37. Dillon J. THE ANAEROBIC PHOTOLYSIS OF TRYPTOPHAN CONTAINING PEPTIDES. Photochem. Photobiol. 1980, 32, 37–39. 10.1111/j.1751-1097.1980.tb03983.x. [DOI] [Google Scholar]
  38. Janecki D. J.; Nemeth J. F. Application of MALDI TOF/TOF mass spectrometry and collision-induced dissociation for the identification of disulfide-bonded peptides. J. Mass Spectrom. 2011, 46, 677–688. 10.1002/jms.1938. [DOI] [PubMed] [Google Scholar]
  39. Pristatsky P.; Cohen S. L.; Krantz D.; Acevedo J.; Ionescu R.; Vlasak J. Evidence for trisulfide bonds in a recombinant variant of a human IgG2 monoclonal antibody. Anal. Chem. 2009, 81, 6148–6155. 10.1021/ac9006254. [DOI] [PubMed] [Google Scholar]
  40. Gu S.; Wen D.; Weinreb P. H.; Sun Y.; Zhang L.; Foley S. F.; Kshirsagar R.; Evans D.; Mi S.; Meier W.; Pepinsky R. B. Characterization of trisulfide modification in antibodies. Anal. Biochem. 2010, 400, 89–98. 10.1016/j.ab.2010.01.019. [DOI] [PubMed] [Google Scholar]
  41. Mozziconacci O.; Kerwin B. A.; Schöneich C. Photolysis of an intrachain peptide disulfide bond: primary and secondary processes, formation of H2S, and hydrogen transfer reactions. J. Phys. Chem. B 2010, 114, 3668–3688. 10.1021/jp910789x. [DOI] [PubMed] [Google Scholar]
  42. Mozziconacci O.; Kerwin B. A.; Schöneich C. Exposure of a monoclonal antibody, IgG1, to UV-light leads to protein dithiohemiacetal and thioether cross-links: a role for thiyl radicals?. Chem. Res. Toxicol. 2010, 23, 1310–1312. 10.1021/tx100193b. [DOI] [PubMed] [Google Scholar]
  43. Kang H.; Tolbert T. J.; Schoneich C. Photo-induced tyrosine side chain fragmentation in IgG4-Fc: mechanisms and solvent isotope effects. Mol. Pharm. 2018, 16, 258–272. 10.1021/acs.molpharmaceut.8b00979. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ao9b01749_si_001.pdf (611.3KB, pdf)

Articles from ACS Omega are provided here courtesy of American Chemical Society

RESOURCES