Free Radical Generation in Far-UV Synchrotron Radiation Circular Dichroism Assays—Protein and Buffer Composition Contribution

A useful tool to analyze the ligands and/or environmental contribution to protein stability is represented by the Synchrotron Radiation Circular Dichroism UV-denaturation assay that consists in the acquisition of several consecutive repeated far-UV SRCD spectra. Recently we demonstrated that the prevailing mechanism of this denaturation involves the generation of free radicals and reactive oxygen species (ROS). In this work, we analyzed the effect of buffering agents commonly used in spectroscopic measurements, including MOPS (3-(N-morpholino) propanesulfonic acid), HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid), TRIS-HCl (tris-hydroxymethil aminomethane hydrochloride), and phosphate, on the efficiency of protein denaturation caused by exposure to UV radiation. Fluorescence experiments confirmed the presence of ROS and were used to determine the rate of ROS generation. Our results indicate that the efficiency of the denaturation process is strongly influenced by the buffer composition with MOPS and HEPES acting also as scavengers and that the presence of proteins itself influenced the ROS formation rate.


Introduction
Circular dichroism (CD) spectroscopy is a valuable biophysical tool for studying the structure and conformation of biomolecules [1]. In particular, CD analysis in the far-UV region is widely used to evaluate the secondary structure of proteins/peptides in solution, as well as the influence of either ligands or environment (polarity, pH, ionic strength) on protein conformation and stability [2]. This analysis is often hampered by the buffer properties and the selection of the right buffer to use is a critical step. This is because it must preserve the activity and stability of the biological system as well as optimizing the experimental conditions by being transparent in the far-UV region.
The use of synchrotron radiation (SR) of third-generation synchrotrons as a light source has overcome the limitation of the lower photon flux of benchtop CD instruments extending the range of analysis in the vacuum-UV region. In particular, the highly collimated microbeam generated at B23 beamline of Diamond Light Source enables the measurements of CD spectra under a broad variety of conditions using: (i) capillaries with sub-milliliter volume of sample [3]; (ii) multiplate for high-throughput CD [4,5]; (iii) high pressure up to 200 MPa [6]; and (iv) CD imaging at high spatial resolution of 50 µm [7]. The high brilliance and photon flux of B23 [8] has been found to induce protein denaturation when repeated consecutive spectra were scanned. The rate of denaturation was found to be sensitive to protein folding, concentration, and environmental conditions [9], leading to the development of a novel method of protein analysis [10]. Of the different mechanisms proposed 2 of 9 to elucidate the protein UV-photo denaturation, free radicals and reactive oxygen species (ROS) are the main one [11,12]. The main targets of far-UV (180-250 nm) and near-UV (250-320 nm) light irradiation of protein are the peptide amide bond, the aromatic and cysteine residues. The excitation of these targets is followed by distinct processes including fluorescence, triplet state formation, reaction with oxygen to form radicals, or excited state photochemical or photophysical processes such as photoionization [13]. In dilute aqueous solutions, at micromolar protein concentration, UV light interacts principally with the solvent water molecules (about 55 M) that are converted into highly reactive species such as hydroxyl radical HO • , hydrated electron e − aq , and hydrogen radical H • (Scheme S1 in Supplementary Information) [14].
The protein UV-photo denaturation assay developed at B23 beamline requires a thorough investigation of the factors that can interfere in the analysis. In particular, much less is known regarding possible buffers and/or biopolymers interferences with studies on protein denaturation induced by exposure to UV radiation. Often these observations go unpublished unless they help the understanding of the systems being studied or their interferences with assays.
To overcome this lack of information, the efficiency of the SRCD UV-denaturation assay has been investigated for different proteins such as bovine fatty acid free (BSAff), pharmaceutical human serum albumin (HSA), lysozyme (HEWL), ubiquitin (Ubi), rabbit IgG (IgG) and poly-L-lysine (PLL) as representative model proteins. These proteins were dissolved in four buffer types typically used in biological and spectroscopic studies, such as MOPS, HEPES, TRIS-HCl, and phosphate (PB) ( Figure S1 in Supplementary Information). Here, the UV-denaturation assays were performed acquiring 50 consecutive SRCD spectra in the 185-260 nm range (top-up mode ring current of 300 mA), while the presence of reactive oxygen species and their rate of generation were evaluated using the ROSsensitive fluorescent probe dehydrorhodamine-123 (DHR-123) that under UV irradiation is converted to Rhodamine-123 (Rh-123) [15].
Our results indicated that the buffer composition influenced and affected the rate of both protein denaturation and free radical formation. It is important to note that the rate of DHR-123 conversion to Rh-123 increased in the presence of proteins.

Results and Discussion
The effect of UV irradiation on the secondary structure of protein has been evaluated collecting 50 consecutive SRCD spectra in the 185-260 nm range of protein solutions. The estimation of the secondary structure elements (α-Helix, β-Strands, Turns and Unordered) of BSAff, HSA and HEWL shows a high content in α-helix, while IgG and Ubi possess a predominant β-sheet secondary structure and the CD spectrum of PLL is consistent with a left-handed extended helix of poly-proline type II [16,17].

Effect of Light on Protein Structure
The collected SRCD spectra of BSAff in PB are reported as example in Figure 1A. A decrease in the intensity of both positive (192 nm) and negative (208 and 222 nm) bands was observed at increased number of scans indicating a progressive loss in the ordered structure from the estimation of the secondary structure content (SSE) using CONTNILL algorithm [18,19]. As shown in Figure 1B, after 50 consecutive scans, the α-helix content decreased from initially 51% to 22%, while the β-sheet content increased from 9% to 25%, accompanied by an increase in both turn (from 15 to 22%) and unordered (from 24 to 31%) elements of secondary structures content. Such behavior was observed repeating the UV-denaturation assays with BSAff dissolved in the other buffers such as HEPES, MOPS and TRIS-HCl ( Figure S2 in Supplementary Information).
To quantify the efficiency of buffers to promote BSAff denaturation, the ellipticity values at 192 nm were plotted versus the number of scans that is proportional to the exposure time to UV light. In the experiment, each scan took 3 minutes, hence the total UVdenaturation assay was completed in 2 and half hours. The data can be fitted with a firstorder exponential decay equation (insert of Figure 1A, and Figure S3 in Supplementary Information) and the decay rate (k=1/t) as well as the half-life (t1/2) values are reported in Table 1. A rapid process is characterized by higher decay rate and lower half-life values. As reported in Table 1, the higher decay rates were determined in both TRIS-HCl and PB buffer solutions (decay rate 0.039 and 0.037 min -1 , respectively), while in MOPS and HEPES the lower values were detected (decay rate 0.031 and 0.028 min -1 , respectively).
Statistical tests using two sample t-test (OriginLab) was performed, and statistically significant differences (p< 0.05) were observed (Table S1 in Supplementary Information). The low decay rate values and the increment in the half-life values determined in MOPS and HEPES buffer solutions may be explained with the capability of these molecules to interact with BSA, supporting the native structure of this protein. The interactions involved in this process essentially occur between the buffer molecules, MOPS and HEPES, and the hydration layers surrounding the peptide backbone of BSA protein [20]. Moreover, docking studies in gas phase showed that MOPS and HEPES molecules can To quantify the efficiency of buffers to promote BSAff denaturation, the ellipticity values at 192 nm were plotted versus the number of scans that is proportional to the exposure time to UV light. In the experiment, each scan took 3 minutes, hence the total UV-denaturation assay was completed in 2 and half hours. The data can be fitted with a first-order exponential decay equation (insert of Figure 1A, and Figure S3 in Supplementary Information) and the decay rate (k = 1/t) as well as the half-life (t 1/2 ) values are reported in Table 1. A rapid process is characterized by higher decay rate and lower half-life values. As reported in Table 1, the higher decay rates were determined in both TRIS-HCl and PB buffer solutions (decay rate 0.039 and 0.037 min −1 , respectively), while in MOPS and HEPES the lower values were detected (decay rate 0.031 and 0.028 min −1 , respectively).
Statistical tests using two sample t-test (OriginLab) was performed, and statistically significant differences (p < 0.05) were observed (Table S1 in Supplementary Information).
The low decay rate values and the increment in the half-life values determined in MOPS and HEPES buffer solutions may be explained with the capability of these molecules to interact with BSA, supporting the native structure of this protein. The interactions involved in this process essentially occur between the buffer molecules, MOPS and HEPES, and the hydration layers surrounding the peptide backbone of BSA protein [20]. Moreover, docking studies in gas phase showed that MOPS and HEPES molecules can bind to domains II and III ( Figure S4 in Supplementary Information) of BSA [20]. Although MOPS and HEPES are structurally related compounds, slight differences in the ability to interact and stabilize the structure of BSA have been detected [20,21], consistent with the difference in the decay rate and half-life values determined in the UV-denaturation assays (Table 1).
A similar drastic change of the CD spectral shape of the investigated proteins of Table 1, HSA (pharmaceutical and dialyzed), IgG, Ubi, PLL and HEWL has been observed in all buffers after 50 consecutive SRCD scans ( Figure S5 in Supplementary Information), with the exception of that for PLL. In this case, the CD shape remained unchanged with only a small decreased intensity of the diagnostic negative band at 197 and positive band at 215 nm. Collectively these data indicate that the exposure to UV radiation induces a loss of secondary structure more evident in highly structured proteins ( Figure S6 in Supplementary Information).
The lowest decay rate values for these proteins were determined in both MOPS and HEPES buffer solutions. These data suggest the presence of a mechanism that reduces the UV-induced protein denaturation in MOPS and HEPES buffers. This is consistent with the recent observation that ROS can efficiently react with both MOPS and HEPES molecules generating morpholino/piperazino radicals that evolve into N-oxide. The latter, subtracting free radicals to the medium, behaves as radical scavenger, hence reducing the efficiency of the overall UV-denaturation process [22,23].
The decay rate values were also correlated with the net charge of proteins. For this purpose, the charge at pH 7.0 of proteins was determined starting from their sequences by the Prot pi Protein Toll web application and plotted versus the determined decay rate values ( Figure S7 in Supplementary Information). The net charge has not been determined for IgG and PLL as the sequence and degree of polymerization, respectively, are unknown. As shown in Figure S7, the decay rate increased when increasing the net charge of protein suggesting that the UV-denaturation process is also correlated with the charge of protein that could influence the physicochemical properties of hydration water that plays different roles including the rate of transmission of ROS from the solvent phase to the protein surface.

Reactive Oxygen Species (ROS) Detection
Reactive oxygen species produced by UV radiation were detected by fluorescence spectroscopy using the positive fluorogenic probe DHR-123. This molecule is converted into the highly fluorescent rhodamine-123 (Rh-123) upon reaction with free radicals including the hydroxyl radical HO • and the superoxide anion radical O 2 •− (Scheme S2 in Supplementary Information). The fluorescence intensity at 524 nm of Rh-123 is proportional to the amount of oxidized DHR-123 and hence the rate of fluorescence increment is proportional to the rate of free radical generation. The irreversibility of DHR-123 oxidation has the advantage that transient events can produce permanent effects.
In these fluorescence studies, synchrotron radiation was replaced by a system consisting of four UV-C lamps emitting at 254 nm.

Starting from Buffer
Experiments irradiating the buffers alone showed that the increment in the fluorescence emission at 524 nm was proportional to the irradiation time of the solution containing the fluorogenic probe DHR-123 (the fluorescence data obtained in PB were reported in Figure 2A as an example).
A quantitative evaluation of this process can be obtained determining the oxidation rate that corresponds to the slope of the tangent to the time-course curves ( Figure 2B and Table 2). As reported in Table 2  A quantitative evaluation of this process can be obtained determining the oxidation rate that corresponds to the slope of the tangent to the time-course curves ( Figure 2B and Table 2). As reported in Table 2  To confirm that DHR-123 oxidation was mediated by free radicals, parallel experiments were performed in the presence of 0.1 mM ascorbic acid, a radical scavenger. No increase in the fluorescence emission was determined in the presence of scavenger, proving the involvement of free radicals in the probe oxidation. The time course in the presence of ascorbic acid in PB buffer is illustrated in Figure 2B. To rule out any thermal effects induced to the aqueous solution following the UV exposure as the cause of the DHR-123 conversion, fluorescence spectra were carried out in the 20-80 °C temperature range. The  To confirm that DHR-123 oxidation was mediated by free radicals, parallel experiments were performed in the presence of 0.1 mM ascorbic acid, a radical scavenger. No increase in the fluorescence emission was determined in the presence of scavenger, proving the involvement of free radicals in the probe oxidation. The time course in the presence of ascorbic acid in PB buffer is illustrated in Figure 2B. To rule out any thermal effects induced to the aqueous solution following the UV exposure as the cause of the DHR-123 conversion, fluorescence spectra were carried out in the 20-80 • C temperature range. The absence of any fluorescence emission ( Figure S8 in Supplementary Information) clearly confirmed that the increased fluorescence emission upon UV irradiation was solely due to the formation of free radicals rather than thermal effects.

Starting from Protein Solution
In the presence of proteins, the oxidation rate values were greater in all buffers as reported in Table 2. This increment could be due to electron and energy transfer reactions between excited proteins and molecular oxygen that produces more reactive oxygen species [11,13]. This is conceivable as proteins can directly absorb UV light either by sidechains chromophores such as Trp, Tyr, Phe and the disulfide bond or by the peptide bonds. Upon absorption of UV radiation, residues are converted to their excited singlet states that readily are converted to the triplet states with much longer lifetimes and able to undergo electron transfer to a nearby group and hydrogen abstraction [11,13]. Photolysis studies also revealed that the excited states can eject an electron to the solvent, yielding solvated electrons, e − aq , that react with O 2 generating O 2 •− radical [11,13].
The higher oxidation rate observed in the investigated buffers with pharmaceutical HSA could also be due to the acetyl-tryptophan added as stabilizer in the pharmaceutical HSA formulation. To confirm this hypothesis, parallel experiments were performed on PB buffer solutions containing either dialyzed pharmaceutical HSA or acetyltryptophanamide (Ac-Trp-NH 2 ). The oxidation rates of the dialyzed pharmaceutical HSA (13.8 a.u./min) were about 9-fold lower than the non-dialyzed (130.6 a.u./min) (Figure 3). On the other hand, the oxidation rate of Ac-Trp-NH 2 was almost identical to that of pharmaceutical HSA (129.0 a.u./min and 130.6 a.u./min, respectively). These data clearly confirmed the involvement of the aromatic Trp residue in the ROS production, indicating this molecule as a strong pro-oxidant ( Figure S9 in Supplementary Information).
cies [11,13]. This is conceivable as proteins can directly absorb UV light either by sidechains chromophores such as Trp, Tyr, Phe and the disulfide bond or by the peptide bonds. Upon absorption of UV radiation, residues are converted to their excited singlet states that readily are converted to the triplet states with much longer lifetimes and able to undergo electron transfer to a nearby group and hydrogen abstraction [11,13]. Photolysis studies also revealed that the excited states can eject an electron to the solvent, yielding solvated electrons, eaq, that react with O2 generating O2 •− radical [11,13].
The higher oxidation rate observed in the investigated buffers with pharmaceutical HSA could also be due to the acetyl-tryptophan added as stabilizer in the pharmaceutical HSA formulation. To confirm this hypothesis, parallel experiments were performed on PB buffer solutions containing either dialyzed pharmaceutical HSA or acetyltryptophanamide (Ac-Trp-NH2). The oxidation rates of the dialyzed pharmaceutical HSA (13.8 a.u./min) were about 9-fold lower than the non-dialyzed (130.6 a.u./min) (Figure 3). On the other hand, the oxidation rate of Ac-Trp-NH2 was almost identical to that of pharmaceutical HSA (129.0 a.u./min and 130.6 a.u./min, respectively). These data clearly confirmed the involvement of the aromatic Trp residue in the ROS production, indicating this molecule as a strong pro-oxidant ( Figure S9 in Supplementary Information). Apart from HSA, ubiquitin and rabbit IgG were the proteins that affected more the oxidation rate of DHR-123. On the contrary, PLL was the least active. This is not a surprise as the dominant conformation is an extended left-handed helix not stabilized by intramolecular hydrogen bonds between NH and CO groups but rather with water molecules. BSAff and HEWL had very similar effects on the oxidation rate.
Analyzing each protein as a function of buffer it can be observed that the higher oxidation rate was observed in PB buffer for HEWL, IgG and Ubi respectively, or in TRIS buffer for BSAff, while the lower values were found in MOPS and in HEPES buffers (Figure 4 and Table 2). The rate of protein UV-photo denaturation as a function of protein Apart from HSA, ubiquitin and rabbit IgG were the proteins that affected more the oxidation rate of DHR-123. On the contrary, PLL was the least active. This is not a surprise as the dominant conformation is an extended left-handed helix not stabilized by intramolecular hydrogen bonds between NH and CO groups but rather with water molecules. BSAff and HEWL had very similar effects on the oxidation rate.
Analyzing each protein as a function of buffer it can be observed that the higher oxidation rate was observed in PB buffer for HEWL, IgG and Ubi respectively, or in TRIS buffer for BSAff, while the lower values were found in MOPS and in HEPES buffers ( Figure 4 and Table 2). The rate of protein UV-photo denaturation as a function of protein formulation revealed the scavenger property of both MOPS and HEPES for the ROS formation.