Fluorescent Bis(guanidine) Copper Complexes as Precursors for Hydroxylation Catalysis †

: Bis(guanidine) copper complexes are known for their ability to activate dioxygen. Unfortunately, until now, no bis(guanidine) copper-dioxygen adduct has been able to transfer oxygen to substrates. Using an aromatic backbone, ﬂuorescence properties can be added to the copper(I) complex which renders them useful for later reaction monitoring. The novel bis(guanidine) ligand DMEG 2 tol stabilizes copper(I) and copper(II) complexes (characterized by single crystal X-ray diffraction, IR spectroscopy, and mass spectrometry) and, after oxygen activation, bis( µ -oxido) dicopper(III) complexes which have been characterized by low-temperature UV/Vis and Raman spectroscopy. These bis(guanidine) stabilized bis( µ -oxido) complexes are able to mediate tyrosinase-like hydroxylation activity as ﬁrst examples of bis(guanidine) stabilized complexes. The experimental study is accompanied by density functional theory calculations which highlight the special role of the different guanidine donors.


Introduction
A major topic in bioinorganic chemistry is the analysis, reproduction, and, ultimately, the improvement of the active sites of natural catalytic systems. This study focuses on dioxygenactivating copper complexes, mimicking the enzyme tyrosinase. Relatively, dioxygen itself is an inert molecule and usually has to be activated by selected enzymes [1]. Such enzymes like tyrosinase are biological oxygenation catalysts [2]. Tyrosinase is a type III copper enzyme, that catalyzes the ortho-hydroxylation of the amino acid tyrosine via the corresponding catechol as an intermediate to the subsequent quinone upon further oxidation [3,4]. These reactions are also responsible for the formation of melanin and other pigments in human or animal skin. Tyrosinase consists of a dicopper center that forms a peroxido complex when it reacts with dioxygen [5][6][7][8][9][10].
Single crystals for complexes C1-C5 were obtained by slow-cooling of acetonitrile or tetrahydrofuran solutions which were heated up beforehand or by gas-phase diffusion of diethyl ether. All complexes were analyzed by IR spectroscopy and mass spectrometry. The molecular structures of the complexes in the solid state were determined by single-crystal X-ray diffraction. The crystals obtained for complex C6 were not suitable for single-crystal X-ray diffraction.

Molecular Structures in the Solid State
The molecular structures in the solid state of the complexes C1-C4 and the cationic unit of C5 are shown in Figure 1. The complexes C1-C4 consist of the DMEG 2 tol ligand that coordinates the central copper atom with the corresponding halide. In the structure of complex C5, a dinuclear dicationic, see Figure 1, unit is present where two ligands each coordinate with their guanidine moieties' two copper atoms. Selected values of the molecular structures, determined by X-ray diffraction are listed in Table 1, for the crystallographic data see Table 3.  [66]. [d] The angles between the planes represented by N gua , N amine , N amine and C gua , C alk , C alk . Average value of the four twist angles for each guanidine moiety of both ligands in the molecular structure. The dimethylethyleneguanidine (DMEG) moieties are bridged through a toluene, resulting in an asymmetric structure. To compare the influence of the adjacency to the aromatic part, the guanidine unit binding via the methylene group and the guanidine unit binding directly to the aromatic group are identified as N gua,aliph and N gua,arom , respectively. All Cu-N bond lengths of the complexes C1-C4 are in a range of 1.963(2) to 2.097(3) Å with the Cu(1)-N gua,arom bond being generally longer than the Cu(1)-N gua,aliph bond, with the exception of C1, where it is the opposite. This exception of the iodine complex has been observed earlier for the bis(tetramethylguanidino)toluene (TMG 2 tol) copper complex [56]. The largest bite angle of the metal center and the two N-donor atoms is 97.5(1) • for complex C1, and the lowest is 92.6(1) • for the complex C4, which has two halides at the metal center, explaining the lower bite angle. The trend of a decreasing bite angle from C1 to C3 is in agreement with the decreasing copper halide Cu-X bond lengths of smaller halides. To determine the coordination environment of the four-fold coordinated complex C4, the τ 4 factor was used [65]. A square-planar coordination environment has a τ 4 value of 0, whereas an ideal tetrahedral coordination environment is represented by a value of 1. The τ 4 value of 0.55 of complex C4 proposes a distorted tetrahedral geometry.
Complex C5, which consists of two ligands and copper atoms and the weakly coordinating triflate, crystallized in the space group Fdd2 and is the highest symmetry structure of all the complexes presented here. The four Cu-N gua bond lengths are the shortest with a range of 1.887(2) to 1.893(2) Å due to the small coordination number. The bite angles of the two N-donor atoms and the metal centers are 178.1(1) and 178.9(1) • indicating a close to linear coordination. The intermetallic interaction in the molecular structure of C5 is weak, as the Cu···Cu distance is only 2.732(1) Å, being shorter than Cu···Cu distances of the dinuclear cations of bis(guanidine) copper structures studied previously [62].
The strong N-donor effect of guanidines originates from the delocalized double bond. To compare the strength of the delocalization, the structural parameter was introduced [66]. The ratio of the C gua -N bond length of the N atom coordinating to the copper center and the two other C gua -N bond lengths of the guanidine moiety is characterized as the parameter . For all complexes C1-C5, the value of is higher for the guanidine unit closer to the aromatic unit. These results have also been found previously for the TMG 2 tol copper complexes [56]. Between the complexes C1-C5, the strength of delocalization is almost the same, except the Cu(II) complex C4 has a higher value because of the higher oxidation state [42].
The twisting angle of the guanidine units can be compared by determining the angles between the plane of the three N atoms N gua , N amine , N amine and the three carbon atoms C gua , C alk , C alk within the guanidine unit. The guanidine twist of the aliphatic guanidines for all complexes C1-C5 is between 16.6 and 17.8 • . The guanidines close to the aromatic units have a smaller twisting angle of 12.6 to 12.8 • , with the exception of the Cu(II) complex C4, which has an angle of 6.4 • . The free rotation of the guanidine units closer to the aromatic unit is smaller, resulting in a smaller twisting angle. The additional chloride in complex C4 increases the hindering effect and may cause the exceptionally smaller value.
The Cu(III) species of a bis(guanidine) is usually only stable at low temperatures [56,67] with an exception which is stable at ambient temperatures for several days [52]. The DMEG 2 tol bis(µ-oxido)copper(III) complexes are only stable long enough to be analyzed by UV/Vis and Raman spectroscopy at temperatures below 200 K. The Cu(III) species then decays. The formation and decay processes have been analyzed for Cu(I) complexes of the simple bis(guanidine) ligand btmgp [55]. The UV/Vis spectra of the complexes O1, O5, and O6 were observed when adding the complexes C1, C5, and C6 to an O 2 -saturated solution of dichloromethane (DCM)/acetonitrile (MeCN) (9:1, v/v) at 195 K. The UV/Vis spectra, measured after the maximum absorption was achieved (after 4 min), are represented in Figure 2. The UV/Vis spectrum shows two peaks at 275 and 365 nm (for O1) or 375 nm (for O5 and O6). These signals seem to be shifted from the expected characteristic peaks at around 300 and 400 nm, which are assigned to the π σ* → Cu 2 O 2 and the oxygen σ* → Cu 2 O 2 ligand-to-metal charge transfer bands (LMCT), respectively [9,68]. Comparing the extinction coefficients found for the 375 nm band of O5 and O6 of 7000 L mol −1 cm −1 to similar complexes of an earlier work with the TMG 2 tol ligand of 12,000 L mol −1 cm −1 , suggests that the oxido species does not fully form at 195 K because it already interferes with the decay of the oxido species. The formation of the bis(µ-oxido)dicopper(III) species O5 and O6 was followed by UV/Vis spectroscopy. The highly intense band at 275 nm gains its enormous intensity by π−> π* transitions of the aromatic system.
The UV/Vis spectrum of the bis(µ-oxido)dicopper(III) species reaches its maximum 4 min after initiation.

Raman Spectroscopy
The oxygenated complex C5 was analyzed by Raman spectroscopy to confirm the formation of the bis(µ-oxido)dicopper(III) species O5 since the characteristic 400 nm feature is shifted to 375 nm. The measurements were carried out with the dioxygen isotopes 16 O 2 and 18 O 2 in propionitrile at 188 K, see Figure 3. In the Raman spectra a shift of the Cu 2 O 2 signal from 603 cm −1 to 574 cm −1 was observed upon isotope substitution. This difference of 29 cm −1 , originating from oxygen-based vibration, is assigned to the breathing core of Cu 2 O 2 cores [69]. Raman measurements with dioxygen isotope substitution of the bis(guanidine)copper complexes [Cu(TMG 2 tol)I] [56] and [Cu(btmgp)I] [70] showed a difference in the Cu 2 O 2 peaks of 28 cm −1 and 25 cm −1 , respectively. The Raman shifts of the dioxygen isotopes 16

Density Functional Theory
For a detailed analysis of the electronic and vibrational characteristics of the desired systems, we performed density functional theory (DFT) [69,71,72] and natural population analysis (NBO) [73][74][75] calculations.
In previous publications, we recommended the use of the hybrid meta GGA functional TPSSh [76] together with the triple-valence basis set def2-TZVP [77] and empirical dispersion correction with Becke-Johnson damping [78,79] for the calculation of bioinorganic copper complex systems [55,62,80,81]. Table 2 summarizes the characteristic bond lengths of the optimized complexes, the NBO charges, and the charge-transfer energies. The optimized structures are in accordance with the molecular structures in the solid state: The Cu-N gua,aliph bond lengths are longer than the Cu-N gua,arom bond lengths for the complexes C1-C3 whereas in the cation of C5 these bond lengths are equal. This observation is in accordance with the experimental results. The NBO analysis yields NBO charges and the charge-transfer energies of the complexes. NBO charges do not describe real charges but, rather, the relative electronic situation. In all Cu(I) complexes, the guanidine N-donors possess a similar charge in the range of −0.67 to −0.71 e − units and also the copper atoms (0.72 to 0.77e − ). In the Cu(II) complex the N gua,aliph donor is slightly more negative than the N gua,arom donor and the copper atom possesses a NBO charge of 1.14 e − units. The Cu-N gua,arom donor coordinates with shorter bond lengths and this is in accordance with the charge-transfer energies: Shorter bond lengths correspond to larger charge-transfer energies. [a] X = I (C1), Br (C2), Cl (C3), Cl1/Cl2 (C4). [b] In these complexes, natural population analysis (NBO) identified the Cu-N bonds as covalent bonds and so no charge-transfer energies could be obtained.
In the Raman experiment we obtain the breathing mode at 603 cm −1 and, after isotope exchange, at 574 cm −1 . DFT predicts the breathing mode to be at 629 cm −1 , shifting to 604 cm −1 upon isotope exchange. The isotope shift hence amounts to 25 cm −1 which is in good agreement with the experimental value of 29 cm −1 . Figure 4 depicts the optimized structure of the bis(µ-oxido) species and the characteristic bond lengths of the core. In comparison to the analogous [Cu 2 O 2 (TMG 2 tol) 2 ] 2+ oxo complex cation, the Cu-N bond lengths are significantly shortened (Cu-N bond lengths in [Cu 2 O 2 (TMG 2 tol) 2 ] 2+ cation: 1.902 and 1.938 Å) and the Cu···Cu distance is slightly shortened (Cu···Cu in [Cu 2 O 2 (TMG 2 tol) 2 ] 2+ cation: 2.746 Å) [56]. The NBO analysis shows that the N gua,arom donor (−0.71 e − units) is slightly more basic than the N gua,aliph donor (−0.64 e − units). This trend has already been observed in the Cu 2 O 2 species with the TMG 2 tol ligand [56]. The NBO charge of the copper ions is 1.36 e − units and is in accordance with the literature [82].

Catalysis
The presented copper complexes are model complexes for the enzyme tyrosinase, which is an effective oxygenation biocatalyst [3,4]. To test whether the complexes are active for substrate oxygenation, complex O1 was tested in the hydroxylation of 8-hydroxyquinoline, see Scheme 4. The reaction was carried out following a standard protocol [22,23]. First, the precursor O1 was prepared by adding complex C1 to a dioxygen saturated solution of tetrahydrofuran (THF) at 195 K. After the formation of the bis(µ-oxido)dicopper(III) species was confirmed via in-situ UV/Vis spectroscopy, 25 equivalents of 8-hydroxyquinoline and 50 equivalents of triethylamine were added.
After the addition of the precursor, the characteristic peak of 7,8-quinolinedione at 413 nm was observed, as shown in Figure 5. Taking the arguments of the unexpectedly lower extinction coefficient, shown in Section 2.1, into account, we can assume that the bis(µ-oxido) complex has already partly decayed and is only present at 58%, when the maximum extinction was measured. The concentration of the formed product was calculated with the known extinction coefficient of 1000 L mol −1 cm −1 at 413 nm for 7,8-quinolinedione [83]. This results in a turnover number of 23 for the conversion of 8-hydroxyquinoline. The timeframe between the last measurement of the formation of O1 and the first measurement after the addition of the substrate is 5.5 s. The process of hydroxylation and subsequent oxidation was extremely fast, as it occurred in less than 5.5 s, even at a temperature of 195 K.

Fluorescence Measurements
The complex C1 and two other bis(guanidine) ligands were investigated by fluorescence spectroscopy, see Figure 6. The complexes [Cu(btmgp)I] [62,64] and [Cu(TMG 2 tol)I] [56] have been synthesized and characterized in earlier studies, but, in this study, the fluorescence has been measured with newly prepared complexes for a better comparison.

Discussion
Ligand and complex synthesis are completed straightforward but the hydroxylation activity was completely unexpected since phenolate hydroxylation tests with the related TMG 2 tol system completely failed [56]. A further remarkable feature of the reported DMEG 2 tol stabilized bis(µ-oxido) dicopper(III) complexes is the shifted UV feature at 375 nm. Normally, this should appear at 400 nm whereas the LMCT of the isomeric side-on peroxido dicopper(II) species can be found at 350 nm [14]. However, the Raman measurements support the identification of these species as bis(µ-oxido) species through the characteristic Cu 2 O 2 breathing mode cores [69]. Since, already, at −80 • C, the decay of the species competes with its formation, no detailed kinetics can be reported. The related TMG 2 tol bis(µ-oxido) species is stable at −80 • C for a longer period [56]. The most notable feature is the hydroxylation activity for the special substrate 8-hydroxyquinoline. Up to now, only some bis(pyrazolyl)methane stabilized peroxido species are able to hydroxylate this challenging substrate to its corresponding quinone [19,37,38]. It must be noted that the hydroxylation velocity seems to be enormous since the reaction is accomplished in a few seconds at −80 • C. Further kinetic analysis was prevented by the limited stability of the Cu 2 O 2 species.
With regard to the desired fluorescence properties, the substitution of TMG against DMEG in the ligand yields a tenfold increase of the fluorescence intensity of the corresponding copper iodide complex, which is useful information for ligand design in general. Moreover, it makes the presented copper(I) complexes valuable for the upcoming studies. Currently, the utilization of the DMEG 2 tol system for laser-induced fluorescence measurements for the detection of mass transfer reactions is under investigation.

Materials and Methods
All reagents were obtained by TCI GmbH (Eschborn, Germany), Sigma-Aldrich GmbH (Taufkirchen, Germany), ABCR GmbH (Karlsruhe, Germany), Fisher Chemicals (Fisher Scientific, Schwerte, Germany), and Merck KgaA (Darmstadt, Germany) and used as purchased. Acetonitrile, dichloromethane, and propionitrile were heated under reflux over CaH 2 and tetrahydrofuran and diethyl ether were heated under reflux over sodium. The solvents were then distilled under nitrogen for purification. The solvent was transferred into an inert-gas glovebox for solution preparation. N,N -dimethylethylenechloroformamidinium chloride (DMEG-VS) was synthesized as described in the literature [53].

Fluorescence Spectroscopy
The fluorescent emission spectra were recorded on a FL-2500 Fluorescence Spectrophotometer, Hitachi High-Technology Co., Ltd. (Tokyo, Japan). A 3 mL quartz cuvette was used as sample cell. The fluorescence intensity was measured with an excitation at 370 nm, an emission range from 220 to 800 nm and a resolution of 1 nm. The spectrophotometer slits for excitation and emission were set at 10 nm.

Elemental Analysis
The elemental analyses were performed with an an Elementar varioEL (Langenselbold, Germany).

X-ray Diffraction Analysis
The single crystal diffraction data for C1-C5 are presented in Table 3. The data for C1-C5 were collected on a Bruker D8 goniometer with APEX CCD detector (Bruker, Karlsruhe, Germany). An Incoatec microsource with Mo-Kα radiation (λ = 0.71073 Å) was used and temperature control was achieved with an Oxford Cryostream 700 (Oxford, UK). Crystals were mounted with grease on glass fibers and data were collected at 100 K in ω-scan mode. Data were collected with SMART [84], integrated with SAINT and corrected for absorption by multi-scan methods with SADABS [85]. The structure was solved by direct and conventional Fourier methods and all non-hydrogen atoms were refined anisotropically with full-matrix least-squares based on F 2 (XPREP [86], SHELXS [87] and ShelXle [88]). Hydrogen atoms were derived from difference Fourier maps and placed at idealized positions, riding on their parent C atoms, with isotropic displacement parameters U iso (H) = 1.2 U eq (C) and 1.5 U eq (C methyl ). All methyl groups were allowed to rotate but not to tip.
In C5 it was not possible to model the disordered solvent molecules (1/2 molecule THF) in an adequate manner, and the dataset was treated with the SQUEEZE routine as implemented in PLATON [89,90].

Raman Spectroscopy
A Tsunami Ti:Sapphire laser system, model 3950-X1BB (Spectra Physics Lasers Inc., CA, USA) was used for the Raman measurements. The fundamental laser line was frequency doubled to 360 nm with an FHG, model GWU2 23-PS (GWU-Lasertechnik Vertriebsges. mbH, Erfstadt, Germany). The laser beam was widened with a spatial filter and then focused on the cuvette inside the cryostat. The focus spot size is around 20 µm. The scattered light was then collected with the UT-3 Raman spectrometer [91]. All spectra were normalized to 1 s integration time and 1 mW laser power. The background was subtracted, and the resulting spectra were corrected with the spectrometer sensitivity for the respective wavelength regions. The experiments were conducted in a clean room with constant temperature (20.0 ± 0.5 • C) and humidity (45% ± 3%). For the measurements, a custom-made half-height Suprasil glass cuvette (Hellma Analytics, Müllheim, Germany) with 1.7 mL sample volume was used. The complex was prepared in an oxygen-and water-free atmosphere (<0.5 ppm) inside a LABstar glovebox (MBraun, Garching, Germany) with a concentration of 5 mM in 99% propionitrile (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany). The solvent was degassed by repetitive "freeze-pump-thaw" cycles and dried over a 3 Å molecular sieve. A newly designed cryostat for Raman measurements at the UT-3, which uses a Proline RP890 chiller ethanol chiller (Lauda, Lauda-Königshofen, Germany) with a Peltier module (TEC4-97-49-17-7-05, Thermonamic, China), was used to cool the sample inside the cuvette to a temperature of around −85 • C before oxygenation was performed [70]. 18 O 2 (Campro Scientific, Berlin, Germany) was used for oxygenation.

Computational Details
Density functional theory (DFT) calculations were performed with the program suite Gaussian 16, revision A03 [92]. The geometries of the C1-C4, the cation of C5, and the oxido species, see Figure 4, were optimized using the nonlocal hybrid meta GGA TPSSh functional [76] and the triple-zeta basis set def2-TZVP [77] as implemented in Gaussian on all atoms. Furthermore, empirical dispersion correction was used with Becke-Johnson damping factors (GD3BJ) [78,79,93]. Frequency calculations did not show imaginary values. NBO calculations for the complexes were accomplished by using the program suite NBO 6.0 [73][74][75].

Computational Details
Density functional theory (DFT) calculations were performed with the program suite Gaussian 16, revision A03 [92]. The geometries of the C1-C4, the cation of C5, and the oxido species, see Figure  4, were optimized using the nonlocal hybrid meta GGA TPSSh functional [76] and the triple-zeta basis set def2-TZVP [77] as implemented in Gaussian on all atoms. Furthermore, empirical dispersion correction was used with Becke-Johnson damping factors (GD3BJ) [78,79,93]. Frequency calculations did not show imaginary values. NBO calculations for the complexes were accomplished by using the program suite NBO 6.0 [73][74][75].

Synthesis of the Oxido Complexes O1, O5, and O6
The synthesized copper(I) complex was dissolved in 5 mL distilled acetonitrile (C1 (0.20 mmol, 0.101 g), C5 (0.20 mmol, 0.122 g), or C6(0.20 mmol, 0.121 g) in the glovebox and transferred into a 500 μm gas-tight Hamilton syringe. A total of 9.5 mL DCM was added to a Schlenk measurement cell and cooled to 195 K. The solvent was saturated with dioxygen by bubbling dry dioxygen gas (O2 99.994%) for approximately 10 min through the solvent against ambient pressure. The copper(I) precursor solution (40.0  10 −3 mol/L) was then injected into the oxygenated solution, resulting in a DCM/MeCN (v/v 9:1) solution with an oxido complex concentration of 1.0  10 −3 mol/L. The formation of the oxido species was followed by UV/Vis spectroscopy.
[Cu(MeCN)4](PF6) (0.093 g, 0.25 mmol) was added to acetonitrile (4 mL), and the solution was heated to approximately 70 °C. Then, a mixture of the DMEG2tol ligand (0.085 g, 0.27 mmol) in 2 mL acetonitrile was added to the hot [Cu(MeCN)4](PF6) solution. Tetrahydrofuran (4 mL) and diethyl ether (4 mL) were slowly added to the surface of the solution, consecutively. The impurities of the formed brownish crystals were removed through recrystallizing with acetonitrile (4 mL) and diethyl ether (6 mL). Overnight, the product was obtained in the form of yellow crystals (0.148 g, 0.12 mmol, 48%). 1 4 ](PF 6 ) (0.093 g, 0.25 mmol) was added to acetonitrile (4 mL), and the solution was heated to approximately 70 • C. Then, a mixture of the DMEG 2 tol ligand (0.085 g, 0.27 mmol) in 2 mL acetonitrile was added to the hot [Cu(MeCN) 4 ](PF 6 ) solution. Tetrahydrofuran (4 mL) and diethyl ether (4 mL) were slowly added to the surface of the solution, consecutively. The impurities of the formed brownish crystals were removed through recrystallizing with acetonitrile (4 mL) and diethyl ether (6 mL). Overnight, the product was obtained in the form of yellow crystals (0.148 g, 0.12 mmol, 48%). 1