Nucleobase pairing and photodimerization in a biologically derived metal-organic framework nanoreactor

Biologically derived metal-organic frameworks (bio-MOFs) are of great importance as they can be used as models for bio-mimicking and in catalysis, allowing us to gain insights into how large biological molecules function. Through rational design, here we report the synthesis of a novel bio-MOF featuring unobstructed Watson-Crick faces of adenine (Ade) pointing towards the MOF cavities. We show, through a combined experimental and computational approach, that thymine (Thy) molecules diffuse through the pores of the MOF and become base-paired with Ade. The Ade-Thy pair binding at 40–45% loading reveals that Thy molecules are packed within the channels in a way that fulfill both the Woodward-Hoffmann and Schmidt rules, and upon UV irradiation, Thy molecules dimerize into Thy<>Thy. This study highlights the utility of accessible functional groups within the pores of MOFs, and their ability to ‘lock’ molecules in specific positions that can be subsequently dimerized upon light irradiation, extending the use of MOFs as nanoreactors for the synthesis of molecules that are otherwise challenging to isolate.

T he structural versatility, permanent porosity, tunable pore surfaces, and periodic nature of metal-organic frameworks (MOFs) provide a unique opportunity to study their potential in applications ranging from gas storage and separation, to UV-induced coupling, catalysis, and sensing [1][2][3][4][5][6][7] . Biologically-derived MOFs (bio-MOFs) based on ligands such as amino acids, nucleobases, and oligosaccharides have shown a great potential in many research areas as they can be used as models for bio-mimicking and catalysis 8,9 . There is a significant interest in observing biological phenomena in porous bio-MOFs, as we can gain insights into their response upon the inclusion of guest molecules and how guest molecules behave within these confined spaces [8][9][10][11][12][13][14][15] . These insights can ultimately lead into the generation of novel bio-MOFs with tunable pore functionalization, allowing one to target specific chemical reactions and isolate desired products [16][17][18][19] . This has the potential to open up a new field of research related to drug design, discovery, delivery, and catalysis 20,21 .
Exploiting functional pore surface in MOFs has recently been utilized in carbohydrate separation and labeling and photochemical transformations [22][23][24] . Intrigued by the pore surface interactions with guest molecules, and the use of functional groups as structure-directing agents, herein we investigated the Watson-Crick (W-C) face of adenine (Ade) within a new MOF (hereafter called as  for nucleobase pairing with thymine (Thy), and solid-state photodimerization of Thy to Thy<>Thy, a molecule that is related to skin cancers such as melanoma 25 . For this concept, two prerequisites govern both the ability of the MOF to act as a nanoreactor, and the ability of Thy to undergo dimerization. First, the strength of Ade-Thy nucleobase pairing within the pores of the MOF is determined by its pore shape and size. Thy not only needs to be able to fit within the pore, but an optimal distance of <3 Å between Ade-Thy is required to allow for the formation of H-bonds via the W-C face of Ade 26 . Here, the structure directing ability of Ade within SION-19 can 'lock' Thy into a position close enough to another Thy for photodimerization to occur. Second, for successful dimerization, both the Woodward-Hoffmann and Schmidt rules would have to be fulfilled [27][28][29] . This implies that the C5-C6 and C5′-C6′ double bonds of Thy (according to its conventional atom numbering scheme) have to be <4.2 Å apart (center-to-center) 29 and the molecular orbital symmetry has to be conserved 28 . Ultimately, an overall balance between these requirements is needed, as pores too large or of improper shape can disfavor Ade-Thy base-pair interaction, while incorrect packing of Thy within the pores can affect their ability to dimerize due to poor orientation of the C5-C6 and C5′-C6′ double bonds.
MOFs with unobstructed W-C faces of Ade reported in the literature are scarce and the generation of materials with a precise pore aperture for Thy uptake has proven to be challenging 17,30,31 . In addition, the dimerization of Thy and formation of Thy<>Thy has only been achieved either in liquid systems such as freezing water, in the presence of sensitizers such as acetone or when Thy is ordered in DNA struts [32][33][34] . Inspired by this challenge, and in order to highlight the structural tunability of MOFs, we demonstrate for the first time, that Ade-Thy binding can be achieved within the pores of SION-19, and Thy can be dimerized to Thy<>Thy in solid state at 40-45% Thy-loadings.

Results
Crystal structure determination of SION- 19 4.0 , crystallizes in an orthorhombic unit cell with the Ccce spacegroup symmetry and cell parameters of a = 10.7402(3) Å, b = 30.6236(7) Å, c = 42.6282(11) Å, α = β = γ = 90°with V = 14020.6(6) Å 3 (Supplementary Discussion 2-3). Within SION-19,  octahedral cages are constructed by four Ade ligands and six Zn II  tetrahedral ions, four of which are in the equatorial plane (Zn II  eq ) and two at the apical positions (Zn II ap ) (Fig. 1a). The Zn II eq link the cages by Zn 4 O clusters forming the adeninate columnar building units running along the a-axis. The tetrahedral geometry of Zn II eq is provided by the N3 and N9 atoms from Ade, one μ 4 -O bridge and one carboxylate O from a TBAPy ligand. For Zn II ap , the tetrahedral coordination environment is owed to two N7 atoms from two adjacent Ade ligands, as well as two monodentate carboxylate O atoms from two consecutive TBAPy ligands (Fig. 1b). Within SION-19, Ade acts as a deprotonated and bridging tridentate ligand, while the fully deprotonated and disordered TBAPy ligand coordinates in a monodentate fashion from each carboxylate (Supplementary Figure 2). The TBAPy ligands, inclined with respect to the a-axis, link the Zn II -based columnar units to yield a structure expanding in three dimensions. However, due to the geometrical features of these building units and TBAPy, infinite channels endowed with two markedly different chemical character are formed along the a-axis. These pores are classified as the acid-pore and the base-pore (Fig. 1c, d). The acid-pore in which the free carboxylate O atoms from the TBAPy ligand are pointing in, has a pore dimension of 7.9 Å × 4.9 Å (including van der Waals radii), while the base-pore contains the unobstructed W-C faces of Ade and its dimensions are 5.4Å × 6.9 Å (Fig. 1e). Four Ade ligands attached to the same Zn 4 O cluster point to four different channels, while two Ade linked to the same Zn II ap ion are arranged in the antiparallel fashion; therefore, even though all Ade are grouped relatively close to each other within the crystal structure, their spatial orientation allows for all of their W-C faces to be exposed to the channels. These pores, in turn, are occupied by disordered guest solvent molecules and cations that show no indication of preferred location. The accessible volume of SION-19 is 31.3%, and combining the accessible volume with the density of the static structure results in a pore volume of 0.287 cm 3  In order to probe any structural changes upon full activation of SION-19, 13 C CP/MAS NMR was utilized. The complete removal of the guest molecules from SION-19 is confirmed by (i) the broadening of the C-signals corresponding to the TBAPy and Ade ligands and (ii) the absence of the carbonyl carbon of DMF at 163.4 ppm (Fig. 2b). The NMR spectrum of SION-19′ shows that aliphatic carbons of [NH 2 Me 2 ] + are still present at 31.0 and 35.0 ppm as broad peaks. Combining our findings, it is apparent that the structural backbone of SION-19 is maintained when is activated to SION-19′.
Sorption properties of SION-19′. The permanent microporosity of SION-19′ was demonstrated by the type I N 2 adsorption isotherm collected at 77 K and 1 bar (Fig. 2c), and it was also found that SION-19′ is porous to CO 2 and CH 4 16). This also confirms the presence of [NH 2 Me 2 ] + in the acid-pore and thus they will not compete with other guest molecules for Ade binding in the base pore.
Thymine loading and dimerization. Based on the structural uniqueness of SION-19 and accessibility of the W-C face of Ade pointing towards the base-pore, room temperature Thy solution isotherms were performed in a mixture of EtOH:MeCN solution (20:80) with loadings ranging from 20 to 100%. These loadings were calculated based on the accessible volume of each formula unit in SION-19 (242.0 Å 3 ) and the molecular volume of Thy (142.2 Å 3 ). Following a loading of 100%, the equilibrium was established via UV/vis spectroscopy to be for 24 h (Fig. 3a, b) 39 . SION-19′ can uptake~1.1 molecules of Thy at 100% loading (Fig. 3b, inset), and this is in agreement with the formula derived from EA:    42,43 . Dynamic, finite temperature behavior of Thy within the pores of SION-19′ was investigated with classical MD simulations at room temperature (298 K). In these simulations, base-pair H-bonding was modeled using threebodied potentials using parameters designed for accurate H-bonding in bio-molecules 44 . With this potential, we observe a good energetic agreement with the DFT binding energy computed for Ade-Thy in SION-19′, being 1.4 kcal/mol higher in energy (−19.6 vs −18.2 kcal/mol).
Exposure of pyrimidine nucleobases like Thy to UV light can induce a [2 + 2] cycloaddition between the C5-C6 and C5′-C6′ double bonds of two pyrimidine rings (Supplementary Scheme 1) 33 . While dimerization of Thy in a liquid medium is a viable method to afford Thy<>Thy, its solid-state dimerization has not been fully studied. To confirm this, Thy crystals were exposed to UV light (254 nm) for 24 h and 1 H NMR revealed that no Thy<>Thy was obtained (Supplementary Figure 27). To investigate the atomistic features of the Thy loading experiments in SION-19′, room temperature MD simulations were performed at loadings of 25, 44, 56, 74, and 100%. Due to the relatively tight pores, and the strong H-bonding potentials used in the simulation, 5 cycles of 2 ns annealing were performed at each loading. Statistics on the orientations of Thy with respect to themselves, and Ade in SION-19′ were collected over the span of 10 ns at intervals of 2 ps. It was hypothesized that in order to enable the formation of Thy<>Thy, the Thy molecules must resemble the transition state of a [2 + 2] cycloaddition, and follow the Schmidt rules. That is, the C5-C6 and C5′-C6′ must be nearly parallel, and separated by <4.2 Å 29 . Thus, the trajectories for the potential formation of a fourmembered ring were analyzed, specifically when the C5-C6 double bond of adjacent Thy molecules were (i) within 4.2 Å of each other, and (ii) nearly aligned, by determining the dihedral angle formed between C5-C6 and C5′-C6′ is below 30°2 9 . Figures 3c, d and 4 demonstrate that Thy molecules appear to be more favorably oriented for dimerization at 44% loading and above. This is compared to the peak in base-pair formation, which occurs at a 44% loading and then subsequently deteriorates. At higher Thy loadings, Thy packs in a dense sterically unfavorable manner within the base pore disfavoring base-pair formation (Fig. 3c). This correlation suggests that the Ade ligands within SION-19′ are structure-directing agents, which fix the position of Thy within the pores, and can allow for Thy<>Thy formation from two nearby 'locked' base-paired Thy molecules. A plot of the trajectories at a 44% loading show that Thy molecules form base pairs with Ade that are two Ade units apart along the a-axis, and on opposite sides of the pore (Fig. 3d).
To validate this experimentally, samples of SION-19@Thy (20-80%) were continuously exposed to UV for 24 h 45 5%). The non-quantitative nature of this dimerization is thought to be due (i) to the low population of an excited state (singly excited, 1 SE), which proceeds through an internal conversion, and requires overcoming an energy barrier 27 , and (ii) to the poor penetration of light throughout the solid material, and thus some Thy molecules do not have enough energy to overcome this barrier and dimerize. Here, this mechanism is governed by both conformational control, and electronic effects [27][28][29] . The inability of lower Thy loadings (20-30%) to dimerize might be due to the increased spatial distance between Thy molecules as they slowly diffuse through the channels, while at higher loadings (50-80%), there is a decrease of H-bonding stabilization between the framework Ade and Thy, leading to an unfavorable packing within the base-pore.

Discussion
To further confirm the structural uniqueness of SION-19′ to 'lock' Thy into a specific orientation and transfer energy for dimerization, Thy loading and dimerization in the pores of adenine-containing . These results further confirm that not only the presence of the W-C faces of Ade, but also the optimal Ade-Thy binding ratio and the subsequent electronic effects/conformation control, play an important role in the dimerization process. That is, the stabilization of Thy via W-C H-bonding with Ade in SION-19 allows it to conform to both the Schmidt and Woodward-Hoffmann rules.
In conclusion, we synthesized a novel biologically derived MOF, SION-19, possessing a new kcs1 topology and Adefunctionalized pore surface, that acts as a nanoreactor for the dimerization of Thy via site-specific binding to Ade. We successfully showed that the orientation of Ade within SION-19 permits Thy binding (40-45% loading) that resembles the transition state of [2 + 2] cycloaddition required for Thy dimerization. Thy loading and dimerization in the pores of SION-19, and the specific relationship between Ade-Thy binding demonstrate that taking advantage of functional groups that decorate the pore surface of MOFs can be used as an effective strategy for their utilization as nanoreactors for photo-induced organic reactions.
The ability to photo-induce the dimerization of solid state Thy within the constrained pores of SION-19 can ultimately provide new fundamental insights on the governing mechanism of this process. Our approach can be applied to a variety of MOFs or other porous materials with comparable structural features to 'lock' molecules in specific positions, and can open up new avenues for the synthesis of organic molecules that may otherwise be difficult to obtain through traditional routes. Furthermore, using bio-MOFs as nanoreactors to form biologically relevant molecules can bridge multiple research disciplines, which can aid in the development of new methods for drug discovery and delivery. These strategies can offer an opportunity for interdisciplinary work, that is, to develop new synthetic methodologies and study the activity of new bio-related products. ) 0.5 ), resulting in an accessible volume of 242.0 Å 3 per FU. The volume of a single Thy molecule, calculated from its crystal structure, was found to be 142.219 Å [3 46 , and therefore, theoretically 1.70 molecules of Thy can fit per FU of SION-19′. Based on these calculations, the concentration of Thy was varied in order to keep the proportion of filling between 20 and 100% (0.56-2.80 × 10 −4 M). SION-19′ was also immersed at higher concentrations (up to 7.11 × 10 −4 M) to check the saturation uptake. Through UV-vis, the time dependency for equilibrium of 100% Thy loading was established to be 24 h. Subsequently, once both the equilibrium time and calibration curve of Thy loaded material were established, SION-19′ was loaded with known concentrations of Thy (0.56-2.80 × 10 −4 M). The absorption of the supernatant was collected and analyzed by UV-vis to quantify the amount of Thy absorbed. Here, the absorption coefficient of thymine in an EtOH:MeCN solution at 263 nm is 8217 cm −1 /M which is comparable to that of thymine in H 2 O (7900 cm −1 /M) 47 .

Methods
Approximately 50 mg of SION-19@Thy (20-80% loadings of Thy) is exposed to UV continuously for 24 h (Fig. 5). The material was then collapsed in a 6 mL 0.5 M K 2 CO 3 solution, sonicated for 10 min, diluted with 50 mL of DMSO, and subsequently dried with MgSO 4 , filtered and concentrated using a rotary evaporator 48 . Each sample was dissolved in 2 mL of DMSO and further prepared according to the procedure in Supplementary Discussion 7.4 for UHPLC-ESI/MS.

Data availability
The X-ray crystallographic coordinated for the SION-19 structure reported in this study has been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition number 1855564. This data can be obtained free of charge from the Cambridge Crystallographic Data Centre via www.ccdc.ac.uk/data_request/cif.