Distinguishing Isomeric Cyclobutane Thymidine Dimers by Ion Mobility and Tandem Mass Spectrometry

Abstract

Irradiation of the major conformation of duplex DNA found in cells (B form) produces cyclobutane pyrimidine dimers (CPDs) from adjacent pyrimidines in a head-to-head orientation (syn) with the C5 substituents in a cis stereochemistry. These CPDs have crucial implications in skin cancer. Irradiation of G-quadruplexes and other non-B DNA conformations in vitro produces, however, CPDs between non-adjacent pyrimidines in nearby loops with syn and head-to-tail orientations (anti) with both cis and trans stereochemistry to yield a mixture of six possible isomers of the T=T dimer. This outcome is further complicated by formation of mixtures of non-adjacent CPDs of dC=dT, dT=dC, and dC=dC, and successful analysis depends on development of specific and sensitive methods. Towards meeting this need, we investigated whether ion mobility mass spectrometry (IMMS) and MS/MS can distinguish the cis,syn and trans-anti T=T CPDs. Ion mobility can afford base-line separation and give relative mobilities that are in accord with predicted cross sections. Complementing this ability to distinguish isomers is MS/MS collisional activation where fragmentation also distinguishes the two isomers and confirms conclusions drawn from ion mobility analysis. The observations offer early support that ion mobility and MS/MS can enable the distinction of DNA photoproduct isomers.

Graphical Abstract:

We explored whether ion mobility mass spectrometry (IMMS) and MS/MS can differentiate between cis,syn and trans-anti T=T CPDs. Ion mobility analysis shows different arrival time, and MS/MS via collisional activation induces fragmentation that also differentiates the two isomers. These results are preliminary evidence that ion mobility and MS/MS can effectively distinguish DNA photoproduct isomers.

INTRODUCTION

Mass spectrometry (MS) has played an important role in the detection and quantification of DNA damage and in particular DNA photodamage.^1–4 Irradiation of B DNA, the principal conformation of genomic DNA, produces mainly cis,syn-cyclobutane pyrimidine dimers (CPDs), smaller quantities of (6–4) dipyrimidine photoproducts, and Dewar valance isomers.⁵ These photoproducts are formed between adjacent pyrimidines in the same strand at yields that depend on the dipyrimidine site (TT, TC, CT and CC) and the flanking sequence.⁶ (The photoadducts are commonly denoted as T=T, T=C, C=T and C=C, respectively. The cis designation for a CPD refers to the relative stereochemistry of the C5 substituents of the cyclobutane ring, whereas the syn designation refers to the head-to-head relative orientation of the pyrimidine rings (see Figure 1).

Figure 1.

Structures of the major thymidine CPDs. (a) Syn CPDs formed from B DNA (n = 0), and bulge loop and single strand DNA (n > 0) bulged loop DNA and (b) anti CPDs formed in a human telomeric G-quadruplex and reverse Hoogsteen DNA (shown is the trans,anti isomer). Dinucleotide and dinucleoside CPDs are released following degradation with snake venom phosphodiesterase (SVP), nuclease P1 (NP1) and calf intestinal phosphatase (CIP).

One of the principal methods for assaying these photoproducts is enzyme-coupled LC-MS/MS.^7–9 In this assay, the irradiated DNA is enzymatically degraded with DNAse I, snake venom phosphodiesterase (SVP), nuclease P1 (NP1), and alkaline phosphatase to dinucleotide and dinucleoside photoproducts that are then analyzed by LC-MS/MS. The degradation products from adjacent cis,syn, (6–4) and Dewar photoproducts have an internucleotide phosphodiester linkage and may be identified by employing a combination of chromatographic retention time, molecular mass, and fragmentation products (Figure 1a).

Although irradiation of the B form DNA only produces adjacent photoproducts, irradiation of non-B DNA can produce non-adjacent photoproducts.¹⁰ These non-adjacent dipyrimidine products can be readily distinguished from adjacent photoproducts by an enzyme-coupled assay that produces dinucleoside degradation products lacking the inter-nucleotide phosphodiester linkage (Figure 1b).¹¹ Irradiation of non-B DNA structures, such as the basket-form G-quadruplexes, under native conditions produces non-adjacent CPDs with cis,anti and trans,anti, head-to-tail stereochemistries^12,13 whereas irradiation of certain bulge loop structures produces non-adjacent cis,syn CPDs.¹⁴

In all, there are six possible regio- and stereoisomers of non-adjacent TT CPDs resulting from enzymatic degradation (Figure S1). There are also six possible isomeric non-adjacent UU CPDs arising from deamination of non-adjacent CC CPD degradation products, and eight isomers arising from deamination on non-adjacent TC and CT CPD degradation products. There is also the possibility that non-adjacent (6–4) and Dewar photoproducts may form, and they are isomers of the CPD products. As a result, the mixture of dinucleotide and dinucleoside photoproducts resulting from irradiation of DNA containing B and non-B DNA structures will be complex, occurring over a wide dynamic range and requiring analysis by conventional LC-MS/MS methods. The ability to detect and identify the much less frequent non-adjacent photoproducts is important as they serve as biomarkers for non-B DNA conformations of DNA that may play an important role in gene regulation.

Tandem MS has become a widely utilized method for characterizing nucleic acids.¹⁵ The most commonly used ion activation technique for the analysis of oligonucleotides has been collision induced dissociation (CID). Additionally, tandem MS has been instrumental in characterizing the isomeric structures of thymidine photoproducts.⁹ These past studies reveal discernible differences in fragmentation patterns among various types of photoproducts. MS/MS, however, is likely insufficient for specificity, and additional structural information will be needed for complex mixture analysis. Ion mobility spectrometry-mass spectrometry (IMS-MS) is emerging as a valuable tool for analyzing isomeric small molecules, leveraging their distinct ion mobilities that are a measure of their shape and size.^16–19 This approach has had success when combined with MS/MS to investigate isomeric nucleosides and nucleotides.^20–23

Herein we describe an early example of the ability of IMS-MS to differentiate structural isomers of non-adjacent CPDs. We first chose to study cis,syn and trans,anti thymidine CPDs because they were the major products and could be isolated in good purity whereas the other isomers formed in lower yields and were more difficult to obtain in pure form. We also complemented IM with isomer-specific fragmentations in the gas phase of two thymidine dimers, for which we investigate mechanisms and energetics for those processes. IMS-MS provides additional structural information through ion mobility separation. Integrating MS/MS approaches may overcome the limitations of each method individually, offering a comprehensive strategy for the characterization of non-adjacent CPDs in DNA photodamage studies. The ability to distinguish these stereoisomers bodes well not only for future studies of photodamage but also for efforts to capture in living cells the presence of non-B DNA structures, such as G-quadruplexes. The ultimate goal is to develop a multi-attribute monitoring approach that allow rapid and reliable identification and differentiation all of isomeric DNA photoproducts. This capability is crucial for comprehensive analysis, especially when dealing with complex mixtures of DNA photoproducts.

RESULTS AND DISCUSSION

Ion Mobility MS of Non-Adjacent Sodiated Thymidine Dimers:

We carried out ion mobility MS experiments on the two major non-adjacent CPDs of thymidine by using a Waters Synapt G2 traveling wave ion mobility spectrometry (TWIMS).²⁴ TWIMS uses a series of radio frequency (RF) voltage pulses applied to a stacked ring ion guide (SRIG), creating a travelling wave that ions “surf” along, driving ions forward and transporting smaller sized ions more rapidly than larger ions. Isomeric photoproducts may have significantly different ion mobilities in the TWIMS and, therefore, be separated. Comparing with reversed-phase chromatography, IMS offers rapid separation of isomers in milliseconds, providing a more rapid means of differentiation. This not only simplifies the analytical setup but also minimizes solvent-related issues such as column conditioning and carryover, enhancing reproducibility and reliability. Although HPLC can certainly separate these isomers, the level of some isomeric photoproducts may below the detection limit of traditional HPLC methods. The amounts of analyte required for IM is relatively low.

We produced non-adjacent thymidine CPDs by triplet-sensitized UVB irradiation of thymidine in aqueous acetone and separated them by off-line reversed-phase HPLC (Figure S1b). The two major photoproducts eluting at 10.0 min and 29.2 min were chosen for IM-MS. These isomers were chosen because they form in the highest yield and could be obtained in relatively good purity for the initial study. Moreover, while inspection of the isomers also suggested that of the many pairs of isomers these two were the most likely to be separable by IM. We confirmed their identities to be the cis,syn and trans,anti T=T CPDs, respectively, by NMR spectroscopy and by comparison with published data (see Figure S2).²⁵ We predict that the cis,syn CPD has a more compact structure because the 2’-deoxyribose rings are on the same side compared to the trans,anti regioisomer, which is a more extended structure, because the 2’-deoxyribose rings are extended in opposite directions. Furthermore, the pyrimidine rings of the cis isomer are less exposed than those of the trans isomer, explaining the shorter retention time of the cis than the trans isomer on reversed-phase chromatography.

To achieve appropriate ion mobility resolution without largely decreasing the ion abundance, we optimized the traveling wave velocity (Figure S3), wave height (Figure S4), and buffer gas velocity (Figure S5), as discussed in Supporting Information.²⁶ These conditions, with some tuning, permit differentiation by IMS-MS of the isomers of each individual photoproduct and of a mixture of the two CPDs (Figure 2).

Figure 2.

Mass spectra and ion mobility spectra for thymidine dimers. (a) Mass spectrum of the component eluting at 10 min by HPLC and corresponding to T[c,s]T. (b) Mass spectrum of the component eluting at 29 min by HPLC, corresponding to T[t,a]T, (c) mass spectrum of the mixture. The ion mobility spectra for sodiated (d) T[c,s]T, (e) T[t,a]T and (f) their mixture. The asterisk denotes an impurity from the HPLC system.

Under optimized settings, the arrival time of cis,syn T=T CPD (T[c,s]T) was 7.13 ms (Figure 2d), compared to 7.69 ms for trans,anti T=T CPD (T[t,a]T) (Figure 2e), indicating that the cis,syn isomer has a more compact structure. To determine whether the current ion mobility resolving power was sufficient to differentiate the two isomers in mixture, we submitted a mixture of the two isomers to IMS-MS, by using a 1000 m/s traveling wave velocity and a 40 V wave height and 90 mL/min nitrogen buffer gas velocity. Under these conditions, the two isomers separate with arrival times of 7.27 ms for T[c,s]T and 7.76 ms for T[t,a]T (Figure 2f). Although the centroids of arrival times change for each isomer compared to previous experiments where they were measured separately, the trend that T[c,s]T elutes earlier than T[t,a]T does not change.

We also submitted a mixture of the two isomers that had been adjusted to give equal peak heights to an IMS-MS measurement on a Bruker timsTOF Pro2 mass spectrometer, which has higher resolving power for mobility than the Synapt.^27,28 We achieved baseline separation as seen in Figure 3. The 1/K₀ of T[c,s]T is 0.981 (Vs cm⁻²), whereas the 1/K₀ of T[t,a]T is 1.021 (Vs cm⁻²). The calculated cross sections are 204 Ų and 213 Ų, respectively. The ion-mobility resolving power of timsTOF Pro2 is greater than that of the Waters Synapt G2, making the former instrument more appropriate for future characterization of more complex photoproduct mixtures, the need for which is discussed above. Taking advantage of the ion-mobility resolving power and sensitivity of timsTOF Pro2. The separation of other isomers is underway and will be reported later.

Figure 3.

Ion mobility spectra demonstrating the baseline separation of T[c,s]T and T[t,a]T on tims-TOF after optimization of the resolving power. The 1/K₀ of T[c,s]T is 0.981 (Vs cm-2), The 1/K₀ of T[t,a]T is 1.021 (Vs cm-2).

Predicted and experimental cross sections are similar:

We calibrated the experimental cross sections because the TWIMS used by Waters on the Synapt G2 and the TIMS adopted by Bruker on the timsTOF Pro2 measure ion mobility indirectly,^29–33 unlike a drift tube IMS. To allow comparison of the experimental and theoretical cross sections, we calculated the cross sections by using a combination of density functional theory (DFT) and trajectory methods embedded in the IMoS software.³⁴ DFT energy minimizations with Gaussian09w software showed that sodium binds near the oxygen-rich sites as would be expected (see Figure S6 and S7 in section S3 for details). We obtained the theoretical cross sections by IMoS calculations on the energy-minimized structures obtained from the DFT calculations; for the T[c,s]T and T[t,a]T isomers; they are 232 Ų and 240 Ų, respectively (a difference of 3.1%) (Table 1 compares the experimental cross sections measured on Bruker timsTOF Pro2, Waters Synapt-G2 and the theoretical cross sections). The calibrated cross sections determined by the Synapt-G2 are smaller than those measured by the timsTOF and predicted by IMoS. One possible explanation is that the helium gas (He) occupying the cell in front of the IMS cell in the Synapt-G2 instrument configuration mixes with the nitrogen gas (N₂) in the IMS cell; ions have smaller mobilities in mixtures of N₂ and He than in He.³⁵ Nevertheless, the relative order of ion mobilities is consistent with both ion mobility mass spectrometers and the theoretical calculations of the T[c,s]T isomer, showing it to be more compact than the T[t,a]T isomer in a nitrogen buffer gas.

Table 1.

Measured and predicted cross section of the isomeric thymidine CPDs.

	T[c,s]T	T[t,a]T	Difference (%)
Computational cross section (Å2)	232	240	3.1%
Experimental cross section (Å2) (Waters-SynaptG2)	190	199	4.7%
Experimental cross section (Å2) (Bruker timsTOF)	204	213	4.0%

The calculation was carried out by IMolS.³⁴

Tandem MS-ion mobility (MS/MS-IMS) of sodiated thymidine dimers:

The Synapt-G2 ion mobility mass spectrometer is equipped with an argon-containing trap and transfer cells that enable us to carry out collision-induced dissociation (CID) of the thymidine dimers. To maximize the yield of fragment ions, the collision voltages applied at the entrance of the trap cell were kept at 40 V. The product-ion spectra are different quantitatively for T[c,s]T and T[t,a]T. T[c,s]T, for example, gives rise to two ions of m/z 391.12 and 275.07 by losing one followed by a second anhydrodeoxyribose (Figure 4a) with arrival times of 7.27 and 5.82 ms, respectively (Figure S8). The T[t,a]T isomer, however, yields nearly exclusively the ion of m/z 391.12 (Figure 4b) with an arrival time of 7.48 ms (Figure S8). This can be compared to what was previously observed for the protonated T[c,s]T and T[t,a]T (m/z 485), which both produced predominantly protonated anhydrodeoxyribose (m/z 117) at a collision energy of 25 V. In contrast, 253) in equal yield while the trans,anti isomer the cis,syn isomer also produced protonated Thy[c,s]Thy (m/z fragmented predominantly to protonated thymine (m/z 127).⁹

Figure 4.

Fragmentation patterns determined by MS/MS for (a) T[c,s]T and (b) T[t,a]T (An-dR refers to anhydrodeoxyribose).

To examine the energy dependence of the fragmentation of a sodiated species, we recorded the product-ion spectrum as a function of collision voltage to create energy-dependent breakdown plots (Figure 5). The m/z 391 ions were generated at a lower DC voltage for T[c,s]T than for T[t,a]T, and the m/z 275.08 ions were not detected from T[t,a]T even at the highest collision voltages used here. To ensure this trend is reproducible, we repeated the CID experiments on a different date (see Figure S9 for more details) and found a similar trend.

Figure 5.

Breakdown graphs of sodiated (a) T[c,s]T and (b) T[t,a]T as a function of ion-activation voltage. Blue curves T[ ]T, orange curves T[ ]T - an-dR, and green curves T[ ]T – 2(an-dR).

The observed differences in fragmentation of the two isomers are likely due to the intrinsic stabilities of the ions in the gas phase and the transition-state energies for fragmentation. To test that notion, we calculated by using DFT the gas-phase heats of formation of the sodiated m/z 391 ions generated from T[c,s]T and T[t,a]T as well as the activation energies for the 1^st and 2^nd fragmentation steps. Based on the DFT results, we propose a mechanism for the stepwise eliminations of two molecules of anhydrodeoxyribose (Figure S12) and the location of the Na⁺. Theory shows that the energy barriers for T[c,s]T are lower than for the trans isomer for losing anhydrodeoxyribose in both steps (Figure 6). The reason for lower energy barriers for cis isomer is the proximity of the two deoxyribose units. The eliminations lead to products in a keto form that stabilizes the sodium ion better than an enol form. Details can be found in section 4 in the supporting information.

Figure 6.

(a) Potential energy surface associated with fragmentation of thymidine dimers. Blue curves for T[c,s]t, red curves T[t,a]T. (b) The activation energy for elimination of each deoxyribose.

CONCLUSIONS

We demonstrated that IM-MS and MS/MS methods can differentiate isomeric photoproducts of thymidine (i.e., T[c,s]T and T[t,a]T isomers) formed from photo-irradiated nucleic acids. Furthermore, improved baseline separation of the dimers was achieved by using timsTOF Pro2, which offers higher mobility resolving power. The T[c,s]T isomer has a smaller drift time than T[t,a]T, similar to the trend observed for HPLC elution. The fragmentation patterns are also distinctive and reflect intrinsic differences in the stabilities of thymidine dimers the fragmentation chemistry and gas-phase structures of CPDs and pave the way for future characterization of more complex photoproduct mixtures by using IMS and MS/MS.

EXPERIMENTAL SECTION.

Authentic thymidine CPD samples were produced by triplet-sensitized UVB (300 nm) irradiation of 355 mg thymidine (Sigma-Aldrich, Saint Louis, MO) in 260 mL of nitrogen degassed 36% aqueous acetone in an RPR-100 Rayonet reactor fitted with 16 RPR-3000A UVB lamps.²⁵ After irradiating for 22 h with continuous nitrogen bubbling, the solution was evaporated to dryness in a rotary evaporator, resulting in a yellow oil. The products were separated by reversed-phase HPLC (XBridge BEH C18 column, 3.5 μm, 4.6 ×75 mm) on a System Gold HPLC system with a binary gradient Model 125 pump and a Model 166 UV detector (Beckman Coulter, Inc., Fullerton, CA). Photoproducts were eluted with 1 mL/min 100% solvent A (50 mM triethylammonium acetate, pH 7.5) for 3 min, followed by a linear gradient of 0–20% B (50% acetonitrile in 50 mM triethylammonium acetate, pH 7.5) in solvent A for 3–53 min, and detected by spectrophotometric absorbance at 260 nm. The cis,syn and trans,anti-containing fractions in D₂O were identified by ¹H NMR analysis on a 300 MHz Varian Oxford NMR spectrometer incorporating comparison with published data.²⁵ Fractions containing thymidine cis,syn and trans,anti T=T CPDs were diluted with 50/50 v/v water and acetonitrile for analysis by mass spectrometry.

IM-MS instrumentation and measurements:

All Waters Synapt-G2 measurements were carried out in the positive-ion mode (Waters, Milford, MA). The mass spectrometer was calibrated in the m/z 50–1500 range by using the clusters of sodium iodide formed from solution prior to the ion mobility measurements. The thymidine CPD solutions were directly infused into the mass spectrometer by using a syringe-pump with a flow rate of 0.8 μL/min. The source and desolvation temperatures were 120 °C and 150 °C, respectively; the capillary voltage was 3.0 kV. The sampling cone and extraction cone voltages were 40 V and 4 V, respectively. Other ESI parameters were optimized to attain the largest mass spectrometric signals. Ions of m/z 507.2 ± 1, corresponding to a sodiated thymidine dimer, were selected in the MS/MS mode for IM-MS experiments. The traveling wave ion mobility (TWIM) components were operated by using the default settings with a 0.4 mL/min flow of argon gas to the trap and transfer cells, a 180 mL/min flow of helium gas to the collision cell, and a flow of 90 mL/min nitrogen gas to the ion mobility cell. The resolving power of the TWAVE was optimized to achieve optimum separation of isomeric thymidine CPDs with a wave height of 35 V and a wave velocity of 1000 m/s for all the ion mobility experiments. To assess the precision of CCS measurements and account for experimental variability we conducted multiple replicate experiments under identical measurement conditions and analyzed the consistency of CCS values obtained. Each experiment was repeated three times under the same parameters. The observed differences in arrival time were within ±0.07 ms, indicating good experimental precision.

The tims-TOF measurements were on a timsTOF Pro2 (Bruker Daltonics, Billerica,MA) in the positive-ion mode. The thymidine CPD solutions were infused at a rate 1 μL/min. The electrospray capillary voltage was set to 2.5–2.8 kV. The pressure in the entrance of the TIMS tunnel was maintained at 2.58 mbar. The ions were accumulated for 50 ms prior to separation, and ion mobility separation was carried out over 100 ms. The ion-mobility scanned range of 1/K₀ is from 0.95 to 1.2 Vs cm^-2.

MS/MS experiments:

A Waters Synapt G2 was used to record tandem MS-IMS (MS/MS-IMS) spectra. MS/MS experiments were conducted upon collisional activation (argon gas at 1×10⁻⁴ mbar) of mass-selected sodiated thymidine CPDs ions of m/z 507.17 by applying a DC voltage to the trap cell. The trap collision voltage was tuned to 40 V so that both the precursor and fragment ions were observed in the same spectrum.

Cross-section calibration:

The cross-section calibration was performed by using a mixture of tetraalkylammonium (TAA) bromide salts. Three ions (tetrahexylammonium, tetraheptylammonium, and tetraoctylammonium) were used as calibrants. Solutions of the TAAs were prepared by dissolving the ammonium salts in 0.1% formic acid in 50/50 v/v water/MeOH to a final concentration of 20 μM. Each experiment was repeated by three times. IMS spectra were recorded by using the same measurement settings as for thymidine dimers. Calibration curves were generated according to the procedures of Ruotolo and Robinson.²⁹

Theoretical calculations of thymidine CPDs:

ChemOffice software (ChemDraw, and Chem3D (PerkinElmer Inc), was used to create the 3D structures of isomeric thymidine dimers. The molecular structures of the neutral thymidine CPDs were then pre-optimized by the PM3 semi-empirical algorithm³⁶ by using Gaussian 09w software. All atom DFT calculations were carried out with Gaussian 09w (Gaussian Inc., Wallingford, CT) by using the B3LYP functional with a 6–31G(d,p) basis set. Structure optimizations and thermodynamics calculations using DFT were employed to determine the lowest energy sodium adduct location on thymidine CPDs (see section S3 in supporting information for details). After DFT, IMoS³⁴ v.1.10 software was used to predict theoretical cross sections from the optimized 3D structures in N₂ buffer gas. The molecular coordinates and partial charge information needed for the IMoS calculations were extracted from the DFT output file. A temperature setting of 304 K was used for all the cross section calculations, and the drift gas-ion interactions were described with the Trajectory Method (TM) and Lennard-Jones potentials.³⁷ The total number of orientations used for the TM method was three, and the number of nitrogen molecules per orientation was 300,000. The CCS of each low energy conformation of the thymidine CPDs was calculated three times with a standard deviation of zero.