Photoisomerizable Guanosine Derivative as a Probe for DNA Base-Pairing in Langmuir Monolayers.

Mixtures of azo-functionalized amphiphilic derivatives of guanosine and of amphiphilic derivatives of other DNA nucleobases were deposited at an air-water interface and repeatedly irradiated with light of 340 and 440 nm wavelengths. The consequent switching between cis and trans configurations of the azobenzene moiety caused changes in the surface pressure of the film, which were analyzed using a model based on the two-dimensional Van der Waals equation of state. For mixed films of guanosine and cytidine derivatives, the analysis revealed a significant modification of the strength of intermolecular interaction caused by the optical irradiation, while no such modifications were identified in mixed films involving other nucleobases. The difference is attributed to light-induced breaking of the hydrogen bonding that is established only between specific nucleobases. The results demonstrate that photosensitive nucleoside derivatives can be used as an efficient probe for base-pairing in Langmuir monolayers.

Introduction

The Langmuir–Blodgett (LB) technique enables precise fabrication of large-scale two-dimensional materials,[1,2] which can be used in photovoltaics[3] and molecular electronics,[4,5] as model membranes[6] and functionalized coatings,[7] etc.[8] An attractive extension of the standard LB methodology is adding some dynamic properties to the LB films by constructing them from molecules containing a photoactive moiety, making it possible to influence their structural and consequently also chemical and physical properties via optical irradiation. Such most common moieties are azobenzene derivatives, which change their configuration from a stretched trans to a bent cis isomer when irradiated with light of one wavelength (typically in the UV spectral range) and back from a cis to a trans isomer when irradiated with light of another wavelength (usually in the visible spectral range). Alternatively, cis-to-trans back-isomerization can take place also via spontaneous thermal relaxation. Their high switching speeds, which are in the range of picoseconds,[9] and their photochemical stability[10] make azobenzene derivatives a popular choice for adding photoactive properties to various materials.

Photoisomerization in thin film structures at the air–water interface as well as on various solid substrates has been extensively studied experimentally.[11−32] In contrast, theoretical models of the corresponding phenomena are quite rare. Sekkat et al. have theoretically analyzed changes in the ratio of cis to trans isomers taking place when a photoactive film is irradiated with linearly polarized actinic light.[32] Toshchevikov et al. developed a model of modifications of mechanical stress during irradiation of photosensitive films.[33] In our work, surface pressure of films formed at the air–water interface was measured, so we developed a model describing photoinduced modifications of the surface pressure in such films. The model is based on the two-dimensional Van der Waals equation of state[34] and considers optical irradiation from several point sources of nonpolarized light.

The films in our experiments were composed of amphiphilic derivatives of different DNA nucleobases. Complementary nucleobases can form intermolecular pairs through multiple hydrogen bonds. While creation of such bonds is known to be hindered by the presence of bulk water,[35] it has been demonstrated that base-pairing can take place in Langmuir films of selected nucleosides when their complementary nucleobase is introduced either to the interface[36−39] or into the water subphase.[40−49] In addition, hydrogen bonding and the associated base-pairing can also be affected by photoisomerization; for example, it has been recently reported that photoisomerization of guanosine derivatives in solution caused breaking of the hydrogen bonds.[50] In this work, we show that a similar bond-breaking process most likely happens at the air–water interface as well and we demonstrate that this effect can be utilized to detect the existence (or absence) of base-pairing interactions in Langmuir films of nucleoside derivatives.

Experimental Section

Amphiphilic Nucleoside Derivatives

We synthesized an amphiphilic azo-functionalized guanosine derivative (GAzo) by reacting guanosine with 12-tricosanone to form a ketal (Figure , details are given in Supporting Information). The azobenzene moiety was then introduced at the 5′-hydroxy function in the form of an ester.

Figure 1

Synthesis of the azo-functionalized amphiphilic guanosine derivative GAzo. (i) (C11H23)2CO, p-TsOH, HC(OMe)3, dioxane, room temperature (rt), 5 h, 70%; (ii) (E)-4-(phenylazo)benzoic acid, N,N’-dicyclehexylcarbodiimide, N,N’-dimethylaminopyridine, dimethylformamide, rt, 48 h, 31%.

Nonphotoactive amphiphilic derivatives of guanosine (dG(C10)2), cytidine (dC(C10)2), adenosine (dA(C10)2), and thymidine (dT(C10)2) (Figure ) were prepared by esterification of the 2′-deoxyribose units with decanoic acid derivatives. The synthesis and surface pressure vs area isotherms of Langmuir films of the latter compounds are described in our previous works.[51,52]

Figure 2

Structures of nonphotoactive amphiphilic deoxyribonucleoside derivatives used in the experiments and denoted dX(C10)2, where X stands for G (guanine), C (cytosine), T (thymine), or A (adenine).

Solutions of each compound with 0.125 mM concentration were prepared by dissolving them in CHCl3. These solutions were then stored at 3 °C and were stirred on a vortex mixer for several minutes before each use, to insure their homogeneity. Fresh spreading solutions of mixed compounds were prepared before every deposition onto the air–water interface by mixing pure solutions so that their molar concentrations were in the GAzo to dX(C10)2 ratios of 75:25, 50:50, or 25:75, where X stands for C, G, A, or T. These mixtures are denoted GAzo-X in the text.

Langmuir Film Preparation

Langmuir films from pure GAzo and from the described mixtures were prepared in a Langmuir–Blodgett trough (KSV Nima KN3002) equipped with two symmetrically driven hydrophilic barriers for monolayer compression. The maximum available surface area of the films was 549 cm2. The trough was filled with ultrapure water (Milli-Q, Millipore Simplicity, 18.2 MΩ cm), and the purity of water subphase was assessed by measuring the change in surface pressure for a standard compression/expansion cycle without a surface film and monitoring the water surface by a Brewster angle microscope (BAM). A total change in surface pressure of less than 0.5 mN/m was considered satisfactory. Surface pressure Π was measured by a conventional Wilhelmy plate method. All measurements were performed at room temperature (23 ± 1 °C).

In all experiments, the volume of the spreading solution was 415 μL. After spreading, the film was allowed to relax for 30 min in a noncompressed state to ensure that all of the solvent had evaporated and internal equilibrium had been established. The barriers were then compressed at a constant rate of 5 mm/min (3.75 cm2/min) until the final surface area of 81 cm2 was reached. The film was then left to relax for another 30 min to reach a new equilibrium before optical irradiation was applied.

Optical Irradiation

UV–visible spectra of chloroform solutions of compounds used in our experiments were recorded with an Agilent/HP 8453 UV–vis spectrophotometer. To irradiate the films, we used four UV light emitting diodes (LEDs; Zhuhai Tianhui Electronic Co., TH-UV340T3WA, peak wavelength of 340 nm, average total radiative power of 65 mW) and two blue LEDs (Chanzon, peak wavelength of 440 nm, average total radiative power of 500 mW), suspended approximately 3 cm above the film (see Figure b). The peak wavelength of the UV LED is close to the absorption peak of the trans isomer of GAzo and consequently induces a trans–cis isomerization of the molecules, while the blue LEDs emit radiation that induces a cis-to-trans transition (Figure a).

Figure 3

(a) UV–vis absorption spectra of the solution of GAzo in chloroform (black solid line, trans isomer and black dashed line, cis isomer) and emission spectra of UV and blue light sources used in the experiments in arbitrary units (violet and blue peak, respectively). Because absorbance values for wavelengths larger than 400 nm are an order of magnitude smaller than for those below 400 nm, a second absorbance axis with a different scale is used (drawn in the middle of the plot). (b) Illustration of the setup used in the experiments. The inset in the bottom-left corner shows a top-down view of the arrangement of blue and UV LEDs used to irradiate the film (marked with blue and violet circles, respectively).

In all experiments, the compressed film was at first irradiated with UV light for 90 min. Then, UV light was switched off and irradiation with blue light for 45 min followed. This was repeated several times for each film. The irradiation with blue LEDs was shorter because the considerably higher optical power of the blue LEDs resulted in faster cis–trans transition.

Brewster Angle Microscopy and Ellipsometry

BAM imaging and ellipsometry measurements were performed with an Accurion EP4 ellipsometer. Ellipsometric data were collected only for pure GAzo films by continually repeating the measurements every minute during the entire course of the LED irradiation experiment. The measurements were performed with a laser source of wavelength 658 nm and with a Xe lamp source filtered through a notch filter centered at 440 nm. The setup is illustrated in Figure b.

Results and Discussion

UV–vis Absorbance of Solutions

If hydrogen bonding is to occur at the air–water interface, it must most likely be detectable in solution as well.[53] We measured the UV–vis absorbance spectra of chloroform solutions of GAzo, dC(C10)2, and their mixture both before and after irradiation with UV light. If there are no interactions between the two compounds, we can expect the absorbance spectrum of the mixture to be equal to the sum of the absorbance spectra of solutions of its constituents. What we find is that this is the case for the irradiated mixture (i.e., cis) but not for the unirradiated one, i.e., trans (Figure ). This suggests that bonding between GAzo and dC(C10)2 in solution can be broken by optical irradiation.

Figure 4

UV–vis absorbance spectra of the chloroform solution of a mixture of GAzo and dC(C10)2 in 1:1 ratio before (blue lines) and after irradiation with UV light (orange lines). The dotted lines show the sum of spectra of GAzo solution and dC(C10)2 solution. Absorbance values for blue lines are displayed on the left axis, while those for the orange lines are displayed on the right axis.

Surface vs Area Isotherms

Surface pressure versus mean molecular area Π(s) isotherms for all of the studied films are shown in Figure . The surface pressure in the film from pure GAzo reaches a maximum at a mean molecular area s of around 79 Å2. At this point, the surface pressure does not exhibit a typical collapse behavior, in which the surface pressure drops sharply, but instead, it plateaus at a more or less constant surface pressure.

Figure 5

Π(s) isotherm for all of the studied mixed films. Each plot contains isotherms with different nonphotoactive molecules, with colors of the lines indicating different relative concentrations of GAzo.

Similar behavior can be observed for the film from pure dG(C10)2 (violet line in Figure b), except that the maximum of Π is reached at s ∼ 63 Å2 and lower value of surface pressure. The surface pressure measured in films from other three amphiphilic nucleoside derivatives does not exhibit a pronounced plateau but, instead, rises monotonously until the maximum compression, with compressibility increasing as the available area is shrunk. It is possible that a plateau would be observed at lower temperatures.[54]

Most isotherms for mixed films closely resemble a linear combination of the isotherms obtained for the two pure compounds (Figure a,c,d). The exceptions are GAzo-G films, for which the isotherms of mixed films behave quite differently (Figure b), which is normally interpreted as an indication that the two compounds do not ideally mix with each other.[55]

Figure a shows the Π(s) isotherms of the Langmuir films prepared from a solution of GAzo that was stored in the dark and a solution that had been irradiated with the actinic UV light for 1 h prior to deposition. Conventional UV–vis spectrometry was used to confirm that the nonirradiated suspension contained predominantly trans isomers, while the irradiated suspension contained predominantly cis isomers. The surface pressure of the cis GAzo film plateaus at slightly higher values of Π than surface pressure of the trans GAzo film, which suggests that cis GAzo isomers more efficiently reduce the surface tension of the water subphase. There is a large hysteresis present in the compression–expansion cycles, indicating attraction between the molecules in the compressed films. The rise in Π in the second compression cycle starts at a slightly lower value of s, meaning that some of the molecules were lost during the first compression–expansion cycle. However, this difference in s is much smaller than the one associated with the hysteresis, leading to the conclusion that most of the molecules indeed dissociate upon expansion.

Figure 6

(a) Π(s) isotherms of Langmuir films fabricated from UV-irradiated (orange lines) and nonirradiated GAzo (blue lines). The arrows indicate the sequence of the compression–expansion cycles. (b) Π(t) plots for different mixed GAzo-C films during repeated subsequent irradiations with UV and blue light. Vertical lines indicate the beginnings of intervals of blue or UV irradiation.

An example of modifications of surface pressure during several irradiation cycles is shown in Figure b. In the beginning, the nonirradiated film is compressed from its extended state to the mean molecular area s ∼ 25 Å2 and left to relax for 30 min. We used this relatively high compression to maximize the chances of hydrogen bond formation. Then, at t ∼ 1.5 h, the first interval of UV irradiation is started, followed by blue irradiation, and so on. The initial surface pressure (Π0) at each cycle remains nearly constant, meaning that the films are relatively stable even after 15 h of measurements. On the other hand, the photoinduced modification of surface pressure (ΔΠ0) drops on average for 3% after every cycle, which signals some fatigue of the materials. Each time the film is irradiated with UV light, the surface pressure rises, which is consistent with the isotherms shown in Figure a. Conversely, irradiation with blue light returns the surface pressure to its previous value, indicating that the film can be reversibly switched from one equilibrium state to another. Irradiation of the film composed solely of nonphotoactive molecules has no effect on its surface pressure (violet line in Figure b), demonstrating that heating of such films by the LEDs is negligible.

Brewster Angle Microscopy Images

Figure shows BAM images of a pure GAzo film captured at different stages of the compression process. First, a structure perforated with holes is formed (Figure a). The holes get smaller in size and fewer in number as s decreases. When the surface pressure plateaus, one begins to see bright spots on the surface (Figure b). With continued compression, the number of these spots increases and when the film is left to relax, they start to coalesce into larger structures (Figure c).

Figure 7

BAM images of the pure GAzo film at different stages of the experiment (a–c). The scale bar in (a) shows the scale for all images. (d) Plot of Π(t) (blue line) and s(t) (dashed line) for the same GAzo film during compression. Approximate times when images in (a)–(c) were taken are marked with their respective letters above the blue line.

BAM images of mixed GAzo-C films compressed to the surface area s ∼ 25 Å2 and relaxed for 30 min are shown in Figure . In the film with n = 0.75, where n denotes a relative fraction of the GAzo molecules, the spots observed in the pure GAzo film are replaced by a more connected and denser structure (Figure a). For n = 0.5, a weblike structure is formed with some of the spots still remaining in the gaps. If the film was left undisturbed for a longer period of time, the number of spots dwindled, suggesting that they slowly merged into the “web” that one can see in Figure b. For n = 0.25, the film surface is again covered by individual spots (Figure c), which is also what can be seen in the film composed solely of dC(C10)2 molecules (Figure d).

Figure 8

BAM images of mixed GAzo-C films, where n = 0.75, 0.50, 0.25, and 0 indicates the relative fraction of the GAzo molecules. The images were captured just moments before the first UV irradiation. The scale bar in image (a) shows the scale for all images.

Figure shows BAM images of mixed GAzo-A films. For n = 0.75, a pattern of more and less densely packed spots is formed (Figure a), resembling the one seen in the pure GAzo film. For n = 0.5 and 0.25, the surface is uniformly covered by bright spots, the only difference being that they appear to be smaller in the latter case (Figure b,c). In the pure dA(C10)2 film (n = 0), only a few spots can be observed on the film surface (Figure d).

Figure 9

BAM images of mixed GAzo-A films, where n = 0.75, 0.50, 0.25, and 0 indicates the relative fraction of the GAzo molecules. The images were captured just moments before the first UV irradiation. The scale bar in image (a) shows the scale for all images.

The surfaces of mixed GAzo-G films appear similar to those of GAzo-A mixed films, in the sense that they are covered with numerous spots (Figure a,b). However, for n = 0.25, these spots are combined into a larger interconnected structure (Figure c). Just as pure dA(C10)2, pure dG(C10)2 also exhibits only a few visible bright spots on its surface (Figure d). However, in this case, one can notice a large flake floating on the surface as well. Surface molecules aggregating into such dense flakes can explain why only a few clusters can be observed in BAM images of both pure dA(C10)2 and pure dG(C10)2 films.

Figure 10

BAM images of mixed GAzo-G films, where n = 0.75, 0.50, 0.25, and 0 indicates the relative fraction of the GAzo molecules. The images were captured just moments before the first UV irradiation. The scale bar in image (a) shows the scale for all images.

BAM images of mixed GAzo-T films are shown in Figure . In contrast to derivatives of other nucleobases, the spots covering the surface of pure dT(C10)2 film condensed into larger flakes over time (Figure d). The mixed films also appear strikingly different, forming increasingly denser structures as more dT(C10)2 is present in the film (Figure ).

Figure 11

BAM images of mixed GAzo-T films, where n = 0.75, 0.50, 0.25, and 0 indicates the relative fraction of GAzo molecules. The images were captured just moments before the first UV irradiation. The scale bar in image (a) shows the scale for all images.

By comparing BAM images of different films captured before and after the UV irradiation, we were unable to discern any significant changes in their structure, apart from those that seem to happen regardless of irradiation, e.g., merging of smaller structures into larger ones. This means that trans–cis isomerization does not produce any systematic modifications in the optical appearance of the films.

Ellipsometry

Ellipsometric measurements were performed only on films composed of pure GAzo. Several compression and irradiation cycles were analyzed. We recorded the values of the two ellipsometric parameters Ψ and Δ during compression; these values are associated with the amplitude ratio and phase shift between reflected beams of two orthogonal polarizations, respectively. While there was a clear change in these values during the compression of the film (not shown), we were unable to detect any significant differences between the irradiated and the nonirradiated film. We believe that this is because the wavelength of the laser source (658 nm) utilized for ellipsometric measurements is far away from any molecular resonance. We, therefore, tried to perform ellipsometric measurements with a noncoherent light source with a wavelength closer to the resonance. In this case, however, we did not have sufficient intensity for the signal to overcome the noise of the measurement.

Theoretical Model

To obtain a relationship between the ratio of trans-to-cis isomers of the photosensitive derivative and the corresponding surface pressure, we recall the two-dimensional Van der Waals equation of state[34]Here Π is the surface pressure of the film, S is its total surface area, N0 is the total number of all molecules forming the film, kB is the Boltzmann constant, and T is the temperature of the system. The parameters A and B account for interactions between the molecules and for their finite size, respectively. By isolating Π from eq , one obtainsWe explicitly marked the terms that are changing during optical irradiation as functions of time. Equation is generally valid only at low densities,[34] so it cannot predict the entire isotherm. However, since light-induced modifications of surface pressure are usually quite small, we believe that it is reasonable to use it to describe those modifications in a more densely packed system. Setting the start of irradiation to t = 0, the variation of Π can be described asWe assume that A is proportional to the strength of interaction between the molecules forming the film and B is proportional to their total surface area.

To describe A(t) and B(t), we introduce two parameters n and m defined so that Np = nN0 is the total number of photoactive molecules and Nc = mnN0 is the number of photoactive molecules that are in the cis state. In irradiation experiments, n is constant, while m is changing with time. We assume that its value is zero at the beginning of UV irradiation intervals, i.e., m(t = 0) = 0. We introduce three constants bc,t,r that represent effective surface areas of different molecules, with indices c and t signifying the cis and trans isomers of the photoactive derivative and index r signifying the nonphotoactive (regular) molecules. Consequently, B can be written aswhere Nt = (1 – m)nN0 denotes the number of photoactive molecules in the trans state and Nr = (1 – n)N0 is the number of regular molecules. In experimental data, the parameter s is usually shown, which is the average area available to a single molecule in the film. Consequently, in further derivation, we define relative surface areas of different molecules b̂c,t,r as b̂c,t,r = bc,t,r/s.

To express A(t), which is associated with the interaction between neighboring molecules, we define six parameters acc, act, att, acr, atr, arr that denote the strengths of different possible pairwise interactions between the molecules. Each index denotes one type of molecule in the pair, e.g., acc describes the interaction between two cis isomers, acr describes the interaction between the cis isomer and a regular molecule, etc. Consequently, A is described aswhere xc,t,r = Nc,t,r/N0 denotes relative fractions of different kinds of molecules.

Further on, if we recall that the change in Π due to the irradiation is small, a linear approximation of eq can be usedHere c2, and c1, are constants dependent on the composition of the film. The full derivation of this expression is given in the Supporting Information.

The function m(t) describes the time dependence of the fraction of cis isomers and is given as[56]The integration takes place over the entire surface area S of the film. We used Φ to denote the quantum yield of the trans–cis isomerization process, and the constant κ0 is defined as κ0 = 9P0ε ln 10/2πh2EγN, where P0 is the total radiative power of each LED, ε is the absorbance of the photoactive molecules in the film as measured in a solution, h is the height at which the LEDs are suspended above the film, Eγ is the energy of a photon of actinic light, and NA is the Avogadro constant. Further details of this calculation are described in the Supporting Information.

Approximations and Characteristic Properties

In the above-described model, we neglect molecular diffusion in the film. In order for diffusion to be important during the irradiation cycle, the molecules would have to travel far enough so that a change in illumination intensity (described by f(x,y) in eq ) would become significant. By using 10–7 cm2/s as an estimate for the diffusion constant,[57] we obtain the average distance traveled by a molecule to be 0.5 mm. At this distance, the relative change in f(x,y) in our irradiation setup can be at most 6%. Next, we assume that the rate of spontaneous cis–trans back-isomerization is much smaller than the rate of UV-irradiation-induced trans–cis isomerization, so we consider that at the end of UV irradiation intervals all photosensitive molecules are in the cis state, i.e., that the situation m = 1 is reached. Finally, we assume that molecular reorientation that happens after isomerization[32,33] occurs on much shorter time scales than the process of isomerization itself.

One of the most characteristic features of the system is the total change of surface pressure for long UV irradiation times, i.e., for conversion from m = 0 to 1. For this case, eq can be used to derive the following expressionThe two newly introduced parameters describe the change in interaction strength: Δaazo = acc – att and Δar = 2(acr – atr). In the limiting case of n = 1, i.e., for a film composed solely of photosensitive molecules, this expression becomesIn other words, the change in surface pressure is proportional to the change in effective size of the molecules (the term in the numerator) and the corresponding effect is additionally enhanced if the film is highly compressed, i.e., when either b̂c or b̂t goes toward 1. The second term simply states that an increase in the strength of attractive interactions between the molecules leads to a decrease of surface pressure.

Another characteristic property of the system is the rate of change of Π at the beginning of UV irradiation. This is obtained by calculating the time derivative of eq , along with the expression for m(t) from eq Here we introduced Δãazo = 2(act – att). The constant Π represents the surface pressure before irradiation in the case when there were no attractive forces between the molecules, and σ is the dimensionless constant that represents the average value of f(x,y) and has a value between 0 and 2.

Aside from the changes of surface pressure induced by the trans–cis isomerization of the photosensitive derivative, the measured surface pressure is also affected by evaporation of water from the LB trough. For photosensitive films, long-term modifications of the system temperature might come into play as well. We approximate those effects with a linear drift function, so that the measured Π(t) dependencies are considered to have a formHere Π0 is the surface pressure before irradiation, ΔΠ(t) is the variation of surface pressure due to trans–cis isomerization (eq ), and k is the drift rate.

Data Analysis

Data analysis was focused solely on the UV-irradiation-induced modifications, as their kinetics were slower than for blue-light-induced changes and consequently more details could have been resolved. The measured Π(t) dependencies for each film (see Figure b for examples) were first divided into subsets. Each subset contained a 90 min time interval of UV irradiation. Then, the values of t and Π at the beginning of each interval were set to zero. The data from the first isomerization cycle was discarded because some of the films had not yet equilibrated by the time of first irradiation. By taking into consideration time dependences of surface pressure during UV irradiation for the following six cycles, we obtained 120 data sets and fitted them with a function of the formHere c1, and c2, are the constants defined by eq . As the exact molecular orientation at the surface is not known, the approximation that all photosensitive molecules have the same orientation was used to calculate m(t) from eq .

The experimentally obtained values for ΔΠ0(n) were deduced by taking the average modification and its standard deviation as resolved from the six consecutive irradiation cycles. We thus obtained ΔΠ0 for five different values of n (0, 0.25, 0.50, 0.75, and 1) and for four different types of mixed films (GAzo-C, GAzo-G, GAzo-A, and GAzo-T films) and fitted those data with the expression from eq . In principle, there are five fitting parameters for every set of data points ΔΠ0 (n): b̂r, b̂t, b̂c, Δar, and Δaazo. However, only parameters b̂r and Δar are unique to each data set; therefore, we simultaneously fitted all four obtained data sets with a total of 11 free parameters: b̂1–4, Δar,1–4, b̂t, b̂c, and Δaazo.

The initial slopes dΔΠ0/dt| (n) were deduced from linear fits of the first three measured points in the ΔΠ(t) curves. As with the ΔΠ0 values, the reported values are average values for six consecutive cycles, and the uncertainties are the standard deviations of those values. The obtained results were fitted to eq . All fitting procedures were performed using the Mathematica 11.1 software package.

Fitting Results

Figure shows examples of measured ΔΠ(t) curves obtained during irradiation with UV light for different mixed films. The red lines in the plots represent the best fit with the theoretical function assuming that all photosensitive molecules have the same orientation. The details are described in the Supporting Information.

Figure 12

Examples of ΔΠ(t) behavior for different concentrations of GAzo in GAzo-C (a), GAzo-G (b), GAzo-A (c), and GAzo-T (d) mixed films. Red lines show best fits obtained by assuming that all photosensitive molecules have the same orientation.

One can notice that GAzo-G, GAzo-A, and GAzo-T mixed films (Figure b–d, respectively) all exhibit a similar behavior: i.e., ΔΠ(t) asymptotically approaches its final value determined by the constant drift due to water evaporation and/or temperature variation. The data for the GAzo-C mixed film, however, show a very different behavior (Figure a). Their most striking feature is a noticeable kink present for n = 0.5. A similar, but much smaller, kink was observed also in some of the irradiation cycles for n = 0.75. In both cases, the initial rates of change dΔΠ0/dt|(n) are considerably larger than for derivatives of other nucleobases; an example of Π(t) behavior during irradiation for n = 0.5 is shown in Figure . In addition, while the total change in surface pressure ΔΠ0 in Figure b–d is more or less proportional to n, this is not the case in Figure a.

Figure 13

Examples of ΔΠ(t) behavior during irradiation with UV light for all types of mixed films at n = 0.5. Points of different colors represent the measured data, and lines of the corresponding color show best fits obtained by assuming that all photosensitive molecules have the same orientation.

The values of ΔΠ0(n) and their fits to eq for all mixed films are shown in Figure . The fitting was performed under the constraint that b̂t has to be the same for all films and b̂c has to be the same for all films but the GAzo-C mixed film. This exception was permitted to verify that the aberrant results obtained for the GAzo-C mixed films, assumed to be caused by a change in A, could not be adequately modeled only by assuming a change in the effective size of the GAzo molecule.

Figure 14

Fits of ΔΠ0(n) data for all mixed films. The type of mixed film is mentioned in the bottom-right corner of each plot.

Because we only have five data points per subset ΔΠ0(n), the best fit is not unique: while most fits have values of b̂r, b̂t, and b̂c that are close to 1, there are multiple pairs of fitting parameters (Δar, Δaazo) that produce similarly satisfying fits of the data. We expect b̂r ≈ b̂t ≈ b̂c because the observed effective areas of the molecules are quite similar, and we also expect b̂c,r,t ≈ 1 because the film is highly compressed. If we assume that there are no large differences in intramolecular interactions due to change of the isomerization state of the azobenzene group, we can set Δar = Δaazo = 0. This produces good fits for all of the mixed films except for the GAzo-C film. For this film, a satisfactory fit can only be achieved for Δar = acr – atr < 0 (Figure a). The associated decrease in interaction strength is attributed to light-induced breaking of molecular bonding between the constituent compounds and signifies that in the investigated Langmuir films such breaking occurs only in mixed films of GAzo with trans GAzo. Note that the f(x,y) function (eq ) is no longer present in the expression for ΔΠ0(n), so no assumptions about the orientation of the molecules have to be made. The values of the parameters obtained in these fits are presented in Table .

Table 1

Values of Fitting Parameters for Fits in Figure and Figure

	Figure 14				Figure 15
	1 – b̂r [10–4]	1 – b̂t [10–4]	1 – b̂c [10–4]	Δar [mN/m]	Φσ [10–2]	Δãazo [mN/m]
dC(C10)2	22	5.2	4.8	–2.64	1.99	–1.12
dG(C10)2	6.4	5.2	4.8	0	1.99	–1.12
dA(C10)2	7.4	5.2	4.8	0	1.98	–1.12
dT(C10)2	5.0	5.2	4.8	0	1.78	–1.12

The values of b̂c, b̂t, and Δar obtained from the above-described fitting procedure were used to fit the data for dΠ/dt| (n) to eq . The value of Δãazo was taken to be the same for all films, and we assumed that the product Φσ is not a function of n but can have different values depending on the type of molecules present in the film, since the shape of the f(x,y) function can depend on the type of molecule. The fits obtained in this way are presented in Figure .

Figure 15

Fits of the rate of change of surface pressure at the start of irradiation (dΠ(t)/dt|) for different mixed films. The type of nonphotoactive molecule in the film is mentioned in the bottom-right corner of each plot. Once again, the values of the parameters obtained in these fits are presented in Table . Note that the values of the product Φσ are very similar for all of the tested mixed films; assuming that quantum yield is constant, this indicates that the orientations of GAzo molecules also remain the same regardless of the type of other molecules in the film.

The obtained results once again show evident similarities among the GAzo-G, GAzo-A, and GAzo-T mixed films (Figure b,c,d), for which convex fitting curves are obtained. The behavior of the GAzo-C mixed film is clearly opposite, i.e., it leads to a concave fitting curve (Figure a).

After observing the above-described specific effect for addition of cytidine derivatives to the surface of the subphase, we also investigated the effect of adding standard water-soluble cytosine molecules to the water subphase. Two different concentrations were tested: 1 and 10 mM, but no effects of such additives on the rate of isomerization or surface pressure were observed.

Discussion

The behavior of films from pure nucleobase derivatives that was observed in our experiments is reminiscent of the one observed by Haycraft et al. who studied Langmuir films of so-called bulge amphiphiles, i.e., molecules with a large group attached to the hydrophobic end of the hydrocarbon chain.[58,59] They also observed a plateauing of surface pressure after the collapse, accompanied by the appearance of bright spots in the BAM images of the film, which in time coalesced into larger formations. Moreover, they also observed cases of large hysteresis in the compression–expansion cycles. In contrast to their work, however, the large “bulge” group in our case is associated with nucleobases that represent the “hydrophilic heads” of the molecules.

Our analysis of the data implies that a drop in interaction strength upon irradiation is present in the GAzo-C mixed films but not in any of the other mixed film containing noncomplementary nucleobases. We propose that photoinduced breaking of H-bonds, which are present between complementary but not between noncomplementary nucleobases, is the explanation for the observed behavior of surface pressure during irradiation. The mechanism by which this happens is that the cis isomer of the azobenzene moiety stretches into the region where H-bonds are formed and thus interferes with the bonding: this was demonstrated to occur in solution for similar compounds.[50,60]

In our previous works, we have already reported that amphiphilic guanosine derivatives behave very differently from derivatives of other nucleobases.[51] It was speculated that in guanosine films guanosine molecules may form so-called Hoogsteen base pairs that are interlinked by two instead of three hydrogen bonds present in the G–C pair.[61] While in the present work the isotherms of GAzo-G mixed films were also found to exhibit quite a unique behavior, the irradiation-induced changes in those films were found to be the same as those in films with noncomplementary nucleobases (GAzo-A and GAzo-T). This suggests that in the GAzo-G films base-pairing either does not take place or it cannot be broken via the trans–cis isomerization of the guanosine derivative.

The strength of intermolecular interaction in the GAzo-C mixed films can be deduced from the obtained value of Δar = 2(acr – atr), which is −2.6 mJ/m2 for the n = 0.5 (50:50) mixture. The total number of molecules deposited on the film was 3.12 × 1016, but a portion of these molecules is lost during compression. Using the isotherms shown in Figure , we can approximate the average effective surface area of the molecules in the monolayer film in question as 100 Å2; dividing the total surface area of the compressed film by this number, we find that the lower bound for the number of molecules remaining in the compressed film is 8 × 1015. Multiplying Δar by the final surface area of the film and dividing this value by half the number of molecules in the film, we can estimate that the energy change per G–C pair is at most 0.23 eV. Since the binding energy of a G–C pair in vacuum is 0.73 eV,[35] we can assume that the difference in energy comes from an increase in interaction strength of water molecules at the surface. In turn, an increase in surface area requires less energy, which means that, by definition, the surface pressure increases. We can speculate that the bound G–C pairs allow more water molecules to reach the air–water interface, which is energetically less favorable than them being in contact with the surfactant, or that the interaction between the water molecules and the bound G–C pair itself is energetically less favorable than the interaction with separate nucleobases. Determining the exact microscopic behavior of water and surfactant molecules at the interface would require a numerical simulation of the system.[62,63]

The quantum yields Φ obtained from fitting the measured dependencies ΔΠ(t) to the theoretical model (Figure ) lie between 0 and 8%. For comparison, the values of quantum yields for isomerization of azobenzene derivatives in commonly used solvents (hexane, toluene, ethanol, etc.) are between 20 and 40%.[9] However, since molecules in a compressed film are sterically hindered, lower values of Φ are reasonable[64] and similarly reduced values of quantum yield in Langmuir films were also observed by Maack et al.[15] With further analysis of the values of Φσ obtained from the fits of dΔΠ0 /dt| shown in Figure , we got Φ ≅ 5%, which lies in the interval predicted by the fits of ΔΠ(t).

To further improve the analysis, one could attempt to obtain the orientation of molecules through fitting, using some appropriate angular distribution function. However, this is quite complicated, as due to inhomogeneous illumination, the theoretical behavior of ΔΠ(t) is described by a non-analytical function, making the calculation computationally demanding. In addition, various angular distribution functions can produce similar dependence of f(x,y), meaning that very accurate measurements of ΔΠ(t) are required to distinguish between them.

Summary and Conclusions

In summary, the behavior of the GAzo-C mixed film during photoisomerization is different than that of other films where noncomplementary nucleobases were mixed. Our analysis implies that the difference stems from an irradiation-induced drop in the interaction strength between the molecules in the GAzo-C mixed film, suggesting that hydrogen bonds between guanosine and cytidine derivatives are broken during irradiation. Such experiments hence provide a convenient tool for investigation as well as for manipulation of base-pairing and other intermolecular interactions in Langmuir films.

In this work, the developed theoretical model was applied only on the Π(t) data. Naturally, the fitting procedure can be made more accurate by including additional complementary measurements. For instance, measurements of optical absorbance would allow one to determine Φ or m(t) more directly, while measurements of surface potential and experiments with linearly polarized irradiation can provide more information on the orientation of molecules. In addition, an IR spectroscopy method adapted for measuring at the air–water interface[65,66] could be employed to more directly confirm the presence and breaking of H-bonds in the studied Langmuir films.