DHPA-Containing Cobalt-Based Redox Metal-Organic Cyclohelicates as Enzymatic Molecular Flasks for Light-Driven H2 Production

The supramolecular assembly of predesigned organic and inorganic building blocks is an excellent tool for constructing well-defined nanosized molecular cavities that catalyse specific chemical transformations. By incorporating a reduced nicotinamide adenine dinucleotide (NADH) mimic within the ligand backbone, a redox-active cobalt-based macrocycle was developed as a redox vehicle for the construction of an artificial photosynthesis (AP) system. The cyclohelicate can encapsulate fluorescein within its cavity for light-driven H2 evolution, with the turnover number (TON) and turnover frequency (TOF) reaching 400 and 100 moles H2 per mole redox catalyst per hour, respectively. Control experiments demonstrated that the reactions were potentially occurred within the cavity of the cyclohelicates which were inhibited in the presence of adenosine triphosphate (ATP), and the redox-active NADH mimic dihydropyridine amido moieties within the ligands played an important role in photocatalytic proton reduction process.

The confinement of the cavity possibly enforced the proximity between the redox-active cobalt(II) centres and Fl, enhancing the PET efficiency to avoid unwanted energy transfer or reverse-ET reactions 26,27 . The mild redox couple of DHPA close to the H 2 /H + couple and the geometric position of DHPA close to the redox catalyst centre made this supramolecular system a more complete working model of AP systems 28,29 . Control experiments based on the reference compound that has the similar structural feature and almost same coordination geometries, as well as redox potential with that of the original, but without the DHPA fragments were also carried out for a comparison.

Results and Discussion
The ligand H 2 ZPB contianing two tridentate coordinated units was obtained from the reaction of 2-pyridyl aldehyde with malono-hydrazide in an ethanol solution. Evaporating a solution containing equivalent molar ratios of H 2 ZPB and Co(NO 3 ) 2 ·6H 2 O in the presence of NaClO 4 for several days led to the formation of the compound Co-ZPB. ESI-MS spectrum of the formed Co-ZPB solid exhibited intense peaks at m/z = 947. 19 and m/z = 996.66, with the isotopic distribution patterns separated by 0.50 ± 0.01 Da, and a comparison with the simulation results based on natural isotopic abundances suggested that these peaks are assigned to [Co 3 (HZPB) 3 ·ClO 4 ] 2+ and [Co 3 (HZPB) 2 (H 2 ZPB) ·2ClO 4 ] 2+ , respectively, indicating the successful assembly of a Co-based M 3 L 3 molecular macrocycle (Fig. 2a). Tridentate (N 2 O) coordinated units sharing two five-membered chelating rings are one kind of efficient building blocks that have been widely used to construct stable and functional discrete architectures with regular structure and high symmetry 30,31 . According to our previous work, each  of three cobalt centres typically coordinated with two planar tridentate N 2 O chelators to form a mer configuration molecular macrocycle with considerable stabilities. In the presence of Fl, the ESI-MS spectrum of Co-ZPB exhibited a new peak at m/z = 1112.22 that was assigned to [Co 3 (HZPB) 3 ·ClO 4 ⊃ Fl] 2+ through comparison with the simulation results obtained based on natural isotopic abundances (Fig. 2b), which indicates the ability of Co-ZPB to encapsulate Fl within its cavity. 1 H NMR spectrum of Co-ZPB recorded after the addition of a 1.0 molar ratio of Fl exhibited significant upfield shifts of protons H 3,6 (δ = 0.15 ppm) and other protons, suggesting that Fl was encapsulated within the electron-rich cavity of Co-ZPB ( Figure S5 in supporting information). UV-Vis titration of Co-ZPB upon addition of Fl caused a significant absorption enhancement at 510 nm. The titration curve of this band reflected the formation of 1:1 stoichiometric ratio of the host-guest complexation, with a calculated association constant of 2.19 × 10 5 M −1 ( Figure S9 in supporting information) 32 . It is postulated that the amide groups located within the positively charged macrocycle introduced geometric and functional properties that are beneficial to the recognition of the organic dye 33,34 .
The cyclic voltammogram of Co-ZPB in CH 3 CN exhibited broad peak at −0.88 V (vs. Ag/AgCl). Because redox potentials of the cobalt centers with same coordination environment in compound Co-QDB (vide infra) and of the DHPA moiety are very close to this value (Figures S20 and S21 in supporting information), the peak was assigned to the overlap of Co II /Co I reduction reaction with the reduction reaction of the DHPA moiety. Clearly, Co-ZPB is well suited to explore the redox-induced reactions that occur near the H 2 /H + couple (Fig. 3a). When the addition of increasing amounts of Et 3 NH + triggered the appearance of a new irreversible cathodic wave near the Co II /Co I response. Increasing the acid concentration raised the height of the new wave with a linear relationship and shifted it to more negative potentials. The new wave was assigned to proton reduction, suggesting that Co-ZPB can reduce protons in a catalytic reaction 35,36 . Moreover, as the oxidation potential of Fl in its photoexcited state (FI* → FI + + e − ) and ground state (Fl → Fl + + e − ) are −1.55 V and 0.87 V (vs SCE) 37 , respectively, the photoexcited state of Fl (Fl*) has sufficient capability to reduce Co(II) to Co(I) directly. In the meantime, Co-ZPB was also an efficient quencher of the photosensitizer Fl ( Fig. 4a and Figure S12 in supporting information). The addition of Co-ZPB to the solution of Fl (10 μM) in 1:1 CH 3 CN/H 2 O caused significant emission quenching. The quenching behaviour is considered a photoinduced electron transfer process from the excited state of Fl (Fl*) to Co-ZPB, enabling the activation of Co-ZPB by Fl for H 2 production in solution 38,39 . The photocatalytic activities of Co-ZPB (0.1 mM) assembled with Fl (0.1 mM) towards evolution of molecular hydrogen were evaluated in an acetonitrile/water solution at room temperature in the presence of 5% triethylamine (TEA) as the sacrificial electron donor 40,41 . The volume of H 2 was quantified at the end of the photolysis by GC of the headspace gases. Our system could work at pH range from 10.5 to 12.5, with maximal H 2 evolution at pH 11.0 (Figure S15 in supporting information). The initial calculated turnover frequency (TOF) was approximately 100 moles H 2 per mole catalyst per hour, with a turnover number (TON) of approximately 400 moles H 2 per mole of catalyst (Fig. 3b). Notably, the TON for Fl and the redox catalyst was obtained in a stoichiometric catalyst/photosensitizer ratio. Compared to the intermolecular systems in which the TON value of one component is optimized with the other component in greater excess, the TON in the stoichiometric system reflects the true activity of the AP system. Meantime, ESI-MS spectrum of the Co−ZPB after reaction exhibited intense peaks at m/z = 947.21, 1062.71 and 1112.20, with the isotopic distribution patterns separated by 0.50 ± 0.01 Da. The peaks were assigned to host and host-guest complex, respectively, indicating the Co−ZPB/Fl system has sufficient structural stabilities during the reaction ( Figure S2 in supporting information).
At a fixed Fl concentration (0.1 mM), the initial rates of H 2 generation increased with the [Co-ZPB] at lower concentrations (<0.1 mM) ( Figure S19 in supporting information). When [Co-ZPB] was fixed 0.1 mM and the Fl concentration was varied, the TOF plateaued at 0.1 mM; further addition of Fl did not increase the lifetime or TON of Co-ZPB. In all cases, the optimal conditions consisted of a constant molar ratio of Co-ZPB/Fl. An increase in the Co-ZPB/Fl ratio decreased the TON, and a decrease in the Co-ZPB/Fl ratio hardly increased the TONs of Fl or Co-ZPB. A 1:1 stoichiometric ratio of Co-ZPB/Fl complexation species apparently dominated the photosynthetic system. Control experiments demonstrated that Fl, Co-ZPB and light are essential for H 2 generation.
To confirm whether the photoinduced H 2 production occurred within the cavity of Co-ZPB or through a normal homogeneous system, the photocatalytic reaction was inhibited by the addition of a non-reactive species, adenosine triphosphate (ATP), to the reaction mixture because previous work showed that a cobalt-based cyclohelicate recognized ATP 42 . As expected, the presence of the molecular host Co-ZPB led to obvious upfield shifts of the aromatic protons on the adenosine ring, suggesting that ATP was encapsulated within the cavity of the macrocyclic complex ( Figure S6 in supporting information). The ESI-MS spectrum of Co-ZPB in the presence of ATP exhibited an intense peak at m/z = 1149.69, with the isotopic distribution patterns separated by 0.50 ± 0.02 Da. This peak was assigned to [Co 3 (HZPB) 2 (H 2 ZPB) ⊃ ATP] 2+ , indicating the stable existence of Co-ZPB in solution and the successful encapsulation of ATP within the cavity of Co-ZPB (Fig. 2c). Importantly, the addition of ATP to replace the photosensitizer or redox catalyst Co-ZPB did not result in any H 2 production, but the presence of 0.3 mM ATP effectively stopped the photocatalytic H 2 production of the Co-ZPB (0.1 mM)/Fl (0.1 mM) system (Fig. 4b). This competitive inhibition behaviour was described as enzymatic-like and suggested that the H 2 production possibly occurred within the cavity of Co-ZPB. The UV-Vis titration of Co-ZPB after the addition of ATP caused a significant decrease in absorption at 510 nm. The titration curve confirmed the 1:1 stoichiometric host-guest behaviour with an association constant of 3.64 × 10 6 M −1 ( Figure S17 in supporting information). This value was thirty fold larger than that of the encapsulation of Fl, demonstrating the possibility of ATP to substitute for Fl to encapsulate the cavity of the metallohelicate. The H 2 production likely occurred within the cavity of Co-ZPB, rather than in a normal homogeneous system 2,43 .
The incorporation of a DHPA group into the ligand backbone as the active site seemed being a powerful approach to adjust the overpotential of the metal sites for proton reduction by sharing the effect of electron gain, loss and distribution. To further investigate the important role of the NADH model in the proton reduction process, a new metallohelicate Co-QDB that has the similar molecular structural features and coordination geometries of cobalt centers with that of Co-ZPB, but without fragment of the DHPA group was synthesized and structurally characterized for comparison. The ligand H 2 QDB was synthesized through the reaction of 2-quiolinecarboxaldehyde with 5-(dibenzylamino)isophthalohydrazide according to the literature method (Fig. 5) 44 . Co-QDB was prepared in a yield of 75% by layering a methanol solution of Co(NO 3 ) 2 ·6H 2 O onto a dichloromethane solution of H 2 QDB in the presence of NH 4 PF 6 . The ESI-MS spectrum of Co-QDB exhibited intense peaks at m/z = 1088.05, 1119.55 and 1161.03, with the isotopic distribution patterns separated by 0.5 ± 0.01 Da, and a comparison with the simulation results based on natural isotopic abundances suggested that the peaks are assigned to [Co 3 (QDB) (HQDB) 2 ] 2+ , [Co 3 (HQDB) 3 (NO 3 )] 2+ and [Co 3 (HQDB) 3 (PF 6 )] 2+ , respectively, revealing the same structure and stability of the Co-QDB in solution ( Figure S3 in supporting information). Single-crystal X-ray analysis confirmed the formation of a pseudo-C 3 symmetric macrocyclic helicate with three cobalt ions and three deprotonated HQDB ligands connected in an alternating pattern ( Figure S1 in supporting information). Each cobalt centre was coordinated by two tridentate N 2 O chelating groups in a mer geometry with pairs of O atoms and amide N atoms each bearing a cis relationship, whereas the acetohydrazide N atoms were trans to each other that further indicate the mer configuration of the Co-ZPB. The measured C-O, C-N and N-N bond distances were all within the normal range of single and double bonds, pointing to the extensive electron delocalization over the entire molecular skeleton (Table S1 in supporting information) 45,46 . The separations between cobalt ions were 9.58 Å on average, and the average separation between the tertiary amine N atoms was 11.36 Å. The presence of four counter anions revealed that only two of the amide groups lost their protons during the coordination. These amide groups provided geometric and functional properties beneficial to the recognition of organic dyes, as observed in our previous works.
The cyclic voltammogram of Co-QDB recorded in DMF exhibited one reversible reduction of Co II /Co I at −1.08 V (vs. Ag/AgCl). This potential falls well within the redox range of reducing a proton in aqueous media 47 , enabling the host to be a redox catalyst for proton reduction ( Figure S21 in supporting information). Co-QDB was also demonstrated to be an efficient quencher of the excited state of Fl through photoinduced electron transfer ( Fig. 6a and Figure S14 in supporting information). Photolysis of a solution of 0.04 mM Fl and 0.08 mM Co-QDB in a solvent mixture containing TEA (5% v:v) in DMF/CH 3 CN/H 2 O resulted in H 2 generation, with optimal photocatalysis at pH 10.0 (Figure S16 in supporting information). As shown in Fig. 6b, the initial TOF was approximately 40 moles H 2 per mole catalyst per hour, with a TON of approximately 250 moles H 2 per mole of catalyst. The TON and TOF of the Co-QDB/Fl system is obviously lower than those of the Co-ZPB/Fl system.
Interestingly, the Co-QDB/Fl ratio is crucial: the TON plateaus at a 2:1 stoichiometric ratio of Co-QDB/Fl under the optimal conditions. At a fixed Co-QDB concentration (0.08 mM), the decrease in the Co-QDB/Fl ratio decreased the TON, and the increase in the Co-QDB/Fl ratio hardly increased the TON of Fl or Co-QDB, suggesting that a potential 2:1 stoichiometric ratio of the Co-QDB/Fl complexation species dominated the photosynthetic system (Fig. 6b). Additionally, glutathione (GSH), an important compound in natural systems that is inactive toward hydrogenation, was chosen as an inhibitor because our previous work showed that an isostructural cyclohelicate could recognize GSH well 44 . When the addition of 0.2 mM GSH to the 2:1 Co-QDB (0.08 mM)/Fl (0.04 mM) system directly stopped the photocatalytic H 2 production. Since GSH does not exhibit any suitable redox potential for H 2 production, this competitive inhibition suggested demonstrated that the H 2 production occurred within the cavity of Co-QDB.
The ESI-MS spectrum of Co-QDB in the presence of Fl exhibited intense peaks at m/z ~2360.27 assigned to {K[Co 3 (HQDB)(QDB) 2 ] 2 ⊃ Fl} 2+ , providing additional proof for the 2:1 stoichiometric complexation behaviour ( Figure S3 in supporting information). After irradiating the system for 6 hours, ESI-MS spectrum of the Co−QDB also exhibited intense peaks at m/z = 1088.05, 1161.03, and 2341.27, with the isotopic distribution patterns separated by 0.50 ± 0.01 Da. The peaks are assigned to host and host-guest complex, respectively, indicating the Co−QDB system also has sufficient structural stabilities in the reaction process ( Figure S4 in supporting information). UV-Vis titration of Co-QDB after addition of Fl supported the 2:1 stoichiometry of the host-guest complexation, with an association constant of 8.32 × 10 9 M −2 ( Figure S10 in supporting information). The 1 H NMR spectrum of Co-QDB after the addition of a 0.5 molar ratio of Fl exhibited significant upfield shifts of protons H 3,6 (δ = 0.13 ppm) and other protons, reflecting the encapsulation of Fl within the cavity of the macrocycle Co-QDB ( Figure S7 in supporting information). Of course Co-QDB was able to recognize GSH in similar aqueous media ( Figure S8 in supporting information). UV-Vis absorption titration of Co-QDB after the addition of GSH also induced quenching and suggested the formation of a 1:1 stoichiometry of the host-guest complexation with an association constant of 5.05 × 10 5 M −1 ( Figure S18 in supporting information). At a fixed Co-QDB concentration of 0.08 mM, the presence of GSH could substitute for Fl to occupy the cavity of Co-QDB. It is hypothesized that Co-ZPB and Co-QDB are true molecular flasks 48,49 , within which AP systems are assembled through encapsulation of an organic dye as a photosensitizer.

Conclusion
In summary, we have reported the preparation of a redox-active cobalt-based macrocycle through the incorporation of an NADH mimic within the ligand backbone and a new strategy for the construction of AP systems. The metal-organic cyclohelicate is an enzymatic molecular flask and encapsulated Fl within its cavity for light-driven H 2 evolution with a TON and TOF that reached 400 and 100 moles H 2 per mole redox catalyst per hour, respectively. The reaction was inhibited by the presence of ATP and occurred within the cavity of the cyclohelicate. The control experiments indicated that the redox-active dihydropyridine amido group of the NADH mimic was helpful for the photocatalytic proton reduction process. By incorporating other redox-active or photoactive functional groups, this strategy can be extended to highly active AP systems.

Materials and Methods
Materials. All chemicals were reagent grade, obtained from commercial sources and used without further purification. The elemental analyses of C, H and N were performed on a Vario EL III elemental analyser. 1 H NMR spectra were measured on a Varian INOVA 400 M spectrometer. ESI mass spectra were obtained on an HPLC-Q-TOFMS instrument using methanol as the mobile phase. UV-Vis spectra were measured on an HP 8453 spectrometer. The solution fluorescence spectra were obtained using an FLS920 spectrometer (Edinburgh Instruments). Both the excitation and emission slit widths were 2 nm. The solutions of Co-ZPB (1.0 × 10 −3 M) and Co-QDB (4.0 × 10 −3 M) were prepared in CH 3 CN and DMF, respectively. Stock solutions of Fl (1.0 × 10 −3 M) were prepared directly in CH 3 CN and were excited at 460 nm.
All electrochemical measurements were carried under nitrogen at room temperature on a CHI 1130 (CH Instrument Co., Shanghai) electrochemical analyser with a conventional three-electrode system consisting of a homemade Ag/AgCl electrode as the reference electrode, a platinum silk electrode with a 0.5 mM diameter as the counter electrode, and a glassy carbon electrode as the working electrode. TEA as the sacrificial electron donor at pH 11.0, and the Co-QDB sample contained Co-QDB (8 × 10 −5 M), Fl (4 × 10 −5 M) and 5% TEA as the sacrificial electron donor at pH 10.0. The flask was sealed with a septum, protected from light, and degassed by bubbling nitrogen for 15 min under atmospheric pressure at room temperature. Next, the samples were irradiated by a 500 W xenon lamp; the reaction temperature was maintained at 293 K using a water filter to absorb heat. The generated photoproduct of H 2 was characterized on a 7890 T GC instrument with a 5 Å molecular sieve column (0.6 m × 3 mm), a thermal conductivity detector, and nitrogen as the carrier gas. The amount of hydrogen generated was determined by the external standard method. The hydrogen in the resulting solution was not measured, and the slight effect of the hydrogen gas generated on the pressure of the Schlenk bottle was neglected in the calculation of the volume of hydrogen gas.