Electron-induced damage of biotin studied in the gas phase and in the condensed phase at a single-molecule level

Biotin is an essential vitamin that is, on the one hand, relevant for the metabolism, gene expression and in the cellular response to DNA damage and, on the other hand, finds numerous applications in biotechnology. The functionality of biotin is due to two particular sub-structures, the ring structure and the side chain with carboxyl group. The heterocyclic ring structure results in the capability of biotin to form strong intermolecular hydrogen and van der Waals bonds with proteins such as streptavidin, whereas the carboxyl group can be employed to covalently bind biotin to other complex molecules. Dissociative electron attachment (DEA) to biotin results in a decomposition of the ring structure and the carboxyl group, respectively, within resonant features in the energy range 0–12 eV, thereby preventing the capability of biotin for intermolecular binding and covalent coupling to other molecules. Specifically, the fragment anions (M–H)−, (M–O)−, C3N2O−, CH2O2−, OCN−, CN−, OH− and O− are observed, and exemplarily the DEA cross section of OCN− formation is determined to be 3 × 10−19 cm2. To study the response of biotin to electrons within a complex condensed environment, we use the DNA origami technique and determine a dissociation yield of (1.1 ± 0.2) × 10−14 cm2 at 18 eV electron energy, which represents the most relevant energy for biomolecular damage induced by secondary electrons. The present results thus have important implications for the use of biotin as a label in radiation experiments.


Introduction
Biotin (Bt, figure 1) is a water-soluble B vitamin, which exhibits an exceptionally high binding affinity to the proteins avidin, streptavidin (SAv) and neutravidin. The Bt-SAv binding is among the strongest non-covalent interactions known with a dissociation constant of the order of 10 −14 -10 −15 mol l −1 [1]. Biotinylation of biomolecules such as DNA and proteins has thus become a widely applied method in molecular biotechnology for purification, detection and immobilization purposes. A large variety of different SAv-modified probes, including gold nanoparticles [2], magnetic nanoparticles [3], fluorophores [4] and quantum dots [5], is nowadays commercially available and routinely used, for instance, in electron [2] and fluorescence microscopy of biological samples [5].
The high strength and specificity of Bt-SAv binding is also increasingly exploited in nanotechnology, e.g. in the controlled assembly of nanocrystal superstructures [6,7] or the fabrication of nanoscale protein patterns [8]. Due to the broad availability of biotinylated DNA oligonucleotides, the field of structural DNA nanotechnology also makes more and more use of Bt-SAv binding, e.g. for the precise arrangement of functional entities on DNA nanostructures [9][10][11], or the visualization of single-molecule chemical reactions on DNA origami templates by atomic force microscopy (AFM) [12][13][14].
Despite its tremendous technological importance, the physiological roles of Bt are still not completely understood. As a B vitamin, Bt is essential for the metabolism, for instance, as a prosthetic group in carboxylase enzymes and it also plays a role in cell signaling and in the expression of more than 2000 human genes [15]. Biotin can further affect the chromatin structure via biotinylation of histones and is thus believed to participate in gene silencing [16]. The same mechanism also seems to be associated with the cellular response to DNA damage [17][18][19] although the mechanisms and pathways are not yet known [16]. In human JAr choriocarcinoma cells for instance, biotinylation of histone H4 was observed to decrease in response to etoposide-induced double-strand breaks, which was interpreted as an early signaling event [18]. UV exposure of human T cells, on the other hand, was found to cause an increase in histone biotinylation, which might be a result of radiation-induced carboxylase degradation or a step during apoptosis [17].
Most of the radiation-induced damage to biomolecules is mediated via the radiolysis of water molecules which generates a large number of radicals and low-energy electrons (LEEs) along the radiation track [20]. In particular, reductive damage induced by secondary electrons with energies below 10 eV has recently been identified as the major source of DNA radiation damage [21]. Dissociative electron attachment (DEA) to DNA proceeds with high cross sections and can lead to the formation of single-and double-strand breaks [22][23][24]. LEE-induced bond breaking has thus received considerable attention and has been studied in a number of biologically relevant molecules, including amino acids [25][26][27], peptides [28,29], vitamins [30], DNA nucleobases [31][32][33], model compounds for the DNA backbone [34][35][36][37] and complete DNA nucleotides [38]. Here, we study LEE-induced damage of Bt in the gas phase by negative ion mass spectrometry, as well as in the condensed phase at the single-molecule level using AFM of DNA origami nanostructures [14].
In the DNA origami technique a long single-stranded DNA scaffold (in most cases M13mp18 with 7249 nucleotides) is folded into specific two-dimensional (2D) and threedimensional (3D) nanostructures simply by hybridization with a suitable set of short oligonucleotides [39,40]. In this way it is not only possible to form complex structures such as 2D rectangles and triangles [39] but also 3D boxes [41], spheres [42] and many others [40]. The main advantage of the DNA origami technique for the study of DNA radiation damage is the simple functionalization with well-defined oligonucleotides, and the possibility to study LEE-induced DNA strand break yields at a single-molecule level using AFM as an analytical tool [14]. The DNA origami structures serve only as a template to arrange DNA target structures within well-defined 'DNA nanoarrays', which allows for the analysis of electron-induced damage to multiple target sequences in one irradiation experiment [14].
We attempt to gain a complete picture of Bt damage by considering a wide range of electron energies, different damage pathways and environments (gas phase versus condensed phase with Bt bound to DNA). We first consider damage pathways at low electron energies (<12 eV) that proceed via negative ion resonances. However, the probability function for LEEinduced damage processes, i.e. the cross sections for DEA, electronic excitation and ionization weighted by the energy distribution of secondary electrons in water has a global maximum right below 20 eV [43]. Consequently, we complement the DEA experiments with a characterization of electron-induced damage of Bt on DNA origami templates using 18 eV electrons.

Gas-phase mass spectrometry
The DEA experiments on gas-phase Bt were performed in an ultrahigh vacuum (UHV) chamber (base pressure of 10 −8 mbar) by means of a crossed electron-molecular beam. Briefly, the experimental setup consists of an electron source, an oven and a quadrupole mass analyzer. An incident electron beam of well-defined energy (full-width at half-maximum ≈230 meV, electron current ≈10 nA) generated from a trochoidal electron monochromator orthogonally intersects with an effusive molecular beam of the target molecule. Under ambient temperatures, the Bt sample is solid. Therefore, it was directly deposited into a vessel inside the vacuum system. During the experiments the overall system was heated up to temperatures of 463-473 K, which is sufficient to generate an effusive molecular beam of Bt with a pressure of 10 −7 mbar as measured with an ionization gauge mounted at one of the flanges.
The negative ions are extracted from the collision area by a small electric field toward an entrance of the quadrupole mass analyzer and detected by single-pulse counting technique. The intensity of negative ions is recorded as a function of electron energy. The electron energy scale is calibrated by measuring the formation of SF − 6 /SF 6 or Cl − /CCl 4 ions with a pronounced resonance feature near 0 eV. The measurements were performed in the absence of the calibration gas to avoid unwanted reactions between the target molecules and anions that arise from electron attachment to the calibration gas. The DEA cross section has been estimated by comparing the ion yields of the fragments generated from Bt with the ion yield of Cl − from CCl 4 at the 0.8 eV resonance. The sample of Bt was obtained from Sigma-Aldrich at a stated purity of 99% and used as delivered.

Condensed phase experiments on DNA origami templates
Triangular DNA origami structures have been prepared according to the procedure described by Rothemund [39]. Details on the DNA origami assembly can be found elsewhere [14]. Three of the staple strands have been modified at the 5 end with Bt, and another three staple strands have been modified with a sequence containing a disulfide linker (S-S) and two T bases: 5 -Bt-TT-S-S. The DNA origami structures are exposed to LEEs, and the prolonged sequences constitute the target structure whose dissociation yield (DY) is determined. The DY of the disulfide containing sequence was determined previously [14] and in the present experiment it served as a reference sequence. The DNA origami structures have been adsorbed on SiO 2 /Si substrates by incubation for 1 h in 10× TAE buffer containing 100 mM MgCl 2 and subsequent rinsing with ethanol/water (1/1). The triangular shape of the DNA origami templates was chosen to minimize intermolecular interactions and thus clustering between the individual DNA origami structures. In this way, a rather homogeneous coverage of DNA origami templates on the SiO 2 /Si substrates is obtained. The dry samples were introduced into the UHV chamber of a commercial time-of-flight secondary-ion-mass-spectrometry instrument, and irradiated with LEEs by means of a flood gun operated at 20 V acceleration voltage. The charging of the SiO 2 surface was determined to be ≈1.7 eV by recording the shift of the electron current onset measured on the aluminum sample holder and the SiO 2 /Si substrates. Thus, the effective energy of electrons arriving at the sample was ≈18 eV. The electron fluence was determined from the measured electron current I, the irradiation time t and the irradiated area A: = I · t/A. Five samples were irradiated at a fluence of ≈3 × 10 12 cm −2 , and two samples were used as a control. The irradiated samples were removed from the UHV chamber and rinsed with 4 ml ethanol/water (1/1) to remove fragmentation products. Subsequently, the samples were incubated for 5 min with a solution of 50 nM SAv and again rinsed with an ethanol/water mixture. AFM imaging of the dried samples was performed in air using a Bruker MultiMode 8 microscope operated in ScanAsyst-HR mode and SCANASYST-AIR-HR probes (Bruker).The DNA origami technique allows for the direct comparison of electron-induced damage to different target structures in one irradiation experiment. Consequently, a determination of the fluence dependence of Bt damage is not required [14]. Instead, the DY can be determined by comparing the damage of Bt to the number of strand breaks (N SB ) in the reference sequence (SSTT) and to the corresponding DY:  by electron attachment to the molecule to form a transient negative ion (TNI). The TNIs are unstable and dissociate into a fragment anion and one or more neutral counterparts. Figure 2 shows the yield of the dehydrogenated parent ion [M-H] − , which is formed within a broad resonance centered at 1.3 eV and with lower intensity at 0 eV. This fragment ion is most likely generated by hydrogen loss from the carboxyl group of Bt and represents a stable closed shell anion. Similar to other organic acids such as formic acid [44] and halogenated organic acids [45,46], the resonance at 1.3 eV is assigned to a π * shape resonance, i.e. the incoming electron occupies a formerly empty π * orbital located on the COOH group. For the dissociation of the O-H bond the π * shape resonance has to couple to the σ * (O-H) orbital. Recently, it was suggested that the σ * (O-H) orbital can be directly accessed at low energies due to the large width of the σ * shape resonance [47,48]. Nevertheless, we cannot completely exclude the generation of [M-H] − due to the loss of an H atom from the N site in a similar way that has been previously reported from thymine [49]. The origin of the lower energy signal is not clear, but it may be attributed to hot band transitions, i.e. transitions involving vibrationally excited molecules. Due to the reciprocal energy dependence of the electron attachment cross section, the intensity of the signal at threshold can be high despite a moderate population of higher vibrational states.

Dissociative electron attachment to gas-phase biotin
In most biotechnological applications Bt is bound to other molecules via the carboxyl group, usually with an amide bond. From the esters of simple organic acids and amino acids it is well-known that the electron-induced formation of the carboxylate ions observed in free acids is preserved in DEA to the corresponding esters [50,51]. Hence, also in a situation where Bt is covalently bound within a more complex environment it is likely that the carboxylate ion is formed by electron attachment at 0.5-2.5 eV to the Bt unit and subsequent cleavage of a C-O bond.
The most intense fragment anion was detected at m/z 80 and is visible within a sharp resonance located at 0.3 eV (figure 2). Since the resonance is located well below the π * shape resonance of the carboxyl group, it is assigned to a shape resonance located on the Bt ring structure. The composition of this ion is not clear, but it can be assigned to the sum formula C 3 N 2 O − , which is accompanied by a complete decomposition of the Bt ring. Further fragment anions that are associated with a decomposition of the ring structure are the OCN − and CN − ions.
The OCN − fragment ion is formed within two resonant features centered at 5.2 and 7.8 eV, respectively. The corresponding ion yield curve is shown in figure 2. The OCN − ion can be excised from the Bt ring by simultaneous dissociation of three chemical bonds. The cleavage of two C-N bonds and one N-H bond requires approximately 6.2 and 4.0 eV (taking the bond dissociation energies of (CH 3 ) 2 -NC 6 H 5 and H-N(CH 3 ) 2 ) [52], which is compensated by the electron affinity of the OCN radical (3.609 eV) [53]. Taking into account the above numbers the thermodynamic threshold for OCN − formation would be 6.6 eV. In fact we observe the resonance at lower energy at 5.2 eV. Hence, the thermodynamic threshold has to be lowered by the formation of a C N double bond within the OCN − ion in order that DEA still becomes energetically accessible. The formation of the TNI at higher electron energies, i.e. at 5.2 and 7.8 eV must be associated with electronic excitation in the parent molecule and the corresponding resonances are referred to as core-excited resonances.
The CN − anion was exclusively observed from a low-energy resonance located at 1.7 eV, which is in striking contrast to the observation of the OCN − ion. The only possible way for its generation is the expulsion of the CN fragment from the Bt ring, which is most likely accompanied by a complete degradation of the ring. The CN − ion formation requires a complex reaction, since five covalent bonds need to be broken prior to its formation (N-H, two C-N bonds and a C O double bond). Nevertheless CN − is often observed after electron attachment to various molecules such as amino acids [51,54], trifluoroalanine [46], N-acetyl-glycine [55], hexafluoroacetone azine [56], formamide [57] and various amide derivatives [58] even at very low energies (0-3 eV). The low-energy thresholds for CN − formation are usually explained by the high electron affinity of the CN radical (3.8 eV), the additional two bonds that are formed within CN − , and a number of new bonds that must be formed in byproducts in the course of the reaction.
In general, the fragment ions are observed with very low count rates indicating a small DEA cross section. As an example we determined the DEA cross section for OCN − formation at 7.8 eV to be 3 × 10 −19 cm 2 using the Cl − signal from CCl 4 as a reference. The damage of Bt by DEA is thus much less effective (about two orders of magnitude) than the damage to DNA. OCN − is a common fragment anion observed in DEA to DNA nucleobases such as e.g. uracil. However, from uracil OCN − is formed with cross sections up to 1.5 × 10 −17 cm 2 [59]. Recently, the cross section of the most intense fragmentation channel in DEA to the DNA nucleobase thymine (T)-i.e. loss of H from the parent ion resulting in [T-H] − -was accurately determined to be 7.9 × 10 −17 cm 2 [60].  A ubiquitous reaction in DEA to organic acids is the formation of OH − . It is generated from a single C-O bond cleavage, and within a larger molecular network this reaction would correspond to a dissociation of the Bt unit. In simple organic acids, OH − was only observed from core-excited resonances at 8-12 eV [61,62]. On the other hand, at low energies (<1 eV) the OH − anion was observed from more complex molecules such as N-acetyl-glycine [55] and the sugar d-ribose [63], indicating that rearrangement reactions within the neutral fragments are required to decrease the thermodynamic threshold.
The fragment anion at m/z 46 (CH 2 O − 2 ) is ascribed to the excision of the complete carboxyl group accompanied by hydrogen transfer and localization of the negative charge on the carboxyl group. It is formed close to threshold in a similar way as in DEA to acetic acid [61], propanoic acid [62], several amino acids [51] and chlorodifluoroacetic acid [45]. The structure of this fragment is presently not known. In principle, it can be assigned to the formic acid anion (HCOOH − ); however, this anion only exists as a short-lived scattering state and hence cannot bind the extra electron to form a thermodynamically stable anion.

Characterization of electron-induced damage of biotin using the DNA origami technique
The most relevant energy regime for secondary electron damage is around 18 eV [43]. However, at this energy no negative ions have been observed in DEA to Bt, indicating that other processes such as electronic excitation and ionization may become relevant. In the following, we consider electron-induced Bt damage using the DNA origami technique. Figure 4(a) shows a scheme of the DNA origami design used in the present study. The pure Bt modifications of the staple strands are placed on the right side of the trapezoids A total number of 5439 molecules have been analyzed (1860 molecules from control samples and 3579 molecules from irradiated samples), and the corresponding histograms are shown in figure 5.
In the samples irradiated with 18 eV electrons at a fluence of 3.1 × 10 12 cm −2 , the number of bound SAv is markedly reduced. In the histogram for the positions of the SSTT sequence, the maximum of the number of bound SAv is shifted to two. In the case of pure Bt the counts of the lower numbers of bound SAv are also increased, although most of the DNA origami templates still carry three SAv. The reduced number of bound SAv is due to the electron-induced damage to the protruding species (i.e. the Bt moiety at the positions of pure Bt, and the SSTT sequence including the Bt label at the other positions). The histograms clearly indicate that the electron-induced damage of the SSTT sequence is considerably higher than that of the pure Bt label. Thus, most of the SSTT damage is ascribed to an electron-induced strand breakage of the disulfide bond, which was previously determined to be 7.1 × 10 −14 cm 2 [14]. Using equation (1), a DY for the Bt label at 18 eV electron energy is determined to be (1.1 ± 0.2) × 10 −14 cm 2 .
The above-mentioned DEA experiments indicate that at 18 eV no direct electron attachment to Bt is possible. The electron impact at 18 eV can either lead to ionization or electronic excitation of Bt followed by dissociation of the Bt moiety, thereby preventing subsequent SAv binding. An alternative pathway is the generation of low-energy secondary electrons from the Si substrate followed by electron attachment to the Bt unit. The electron ionization cross sections of DNA bases at 18 eV are in the order of 10 −16 cm 2 [64,65], and it can be assumed that the ionization cross section for Bt is of the same order. In comparison, the DEA cross sections determined in the present study are of the order of 10 −19 -10 −18 cm 2 . Since the secondary electron yield of oxidized Si surfaces at 18 eV primary energy is only around 0.5 [66], the dominating process at 18 eV electron energy is most likely initiated by ionization of Bt.
The histograms of the control samples in figure 5 show two additional features that are worth mentioning: 1. In the control samples, a considerable number of DNA origami structures carry less SAv than three, i.e. the maximum number of labels. This is due to a limited purity of the prolonged staple strands, i.e. the number of strand breaks without irradiation simply corresponds to the number of staples that do not carry a Bt marker due to errors in the oligonucleotide synthesis. This error is larger for the SSTT sequences than for the oligonucleotides that carry only the Bt marker. 2. A considerable number of DNA origami templates do not carry a single SAv at the position of the pure Bt target ('Zero counts'). This effect is particularly obvious for the control samples, where the number of zero counts for the SSTT sequence is close to zero, but is close to 150 for the pure Bt sequence. The number of DNA origami structures without SAv is for the pure Bt position almost as high as for 1 SAv molecule per DNA origami structure. This observation can be ascribed to the adsorption geometry of the flat triangles on the Si substrate. It was shown previously that about 17% of the triangular DNA origami structures adsorb 'face-down' [14], i.e. with the protruding strands pointing toward the Si surface. The protruding SSTT strands are long enough to expose the Bt linker in a way that SAv can in most cases still bind to Bt in this configuration. However, the Bt linker without protruding strand is apparently too short, so that Bt is buried underneath the DNA, thereby blocking the binding to SAv. Since the DY is determined from the difference of decomposed molecules from the control sample and the irradiated sample, this adsorption geometry is not expected to influence the DY considerably.

Conclusions
The electron-induced damage to Bt was characterized with two complementary methods, i.e. in the gas phase with a crossed-beam setup and in the condensed phase with the DNA origami technique. The DEA experiments revealed the formation of a series of fragment anions that are either associated with decomposition of the carboxyl group or the Bt ring structure. The deterioration of the ring structure destroys the capability of Bt to bind to proteins such as SAv via intermolecular hydrogen bonding. A decomposition of the carboxyl group leads to a desorption of the Bt unit when it is bound to other molecules. In general, the DEA cross sections are rather small, and exemplarily we determined the cross section for OCN − formation at 7.8 eV to be 3 × 10 −19 cm 2 . With the DNA origami technique, we studied electron-induced damage of Bt at 18 eV electron energy and determined a DY of (1.1 ± 0.2) × 10 −14 cm 2 . In previous experiments the DYs of 5 -Bt-TT-SS and 5 -Bt-TT sequences have been determined to be 7.1 × 10 −14 and 1.7 × 10 −14 cm 2 , respectively [14]. Along with the present results, we can conclude that the cleavage of a TT sequence (without Bt) proceeds with a yield of 0.6 × 10 −14 cm 2 , whereas the contribution that can be ascribed to a cleavage of the disulfide bond (S-S) is 5.4 × 10 −14 cm 2 .
These experiments indicate that Bt damage has to be considered when using Bt as a label in single-molecule experiments using the DNA origami technique for the determination of DNA strand break yields. This concerns, in particular, not only the energy regime where ionization takes place (>10 eV) but also the energy range where DEA is most effective, i.e. below 1 eV.