Truly‐Biocompatible Gold Catalysis Enables Vivo‐Orthogonal Intra‐CNS Release of Anxiolytics

Abstract Being recognized as the best‐tolerated of all metals, the catalytic potential of gold (Au) has thus far been hindered by the ubiquitous presence of thiols in organisms. Herein we report the development of a truly‐catalytic Au‐polymer composite by assembling ultrasmall Au‐nanoparticles at the protein‐repelling outer layer of a co‐polymer scaffold via electrostatic loading. Illustrating the in vivo‐compatibility of the novel catalysts, we show their capacity to uncage the anxiolytic agent fluoxetine at the central nervous system (CNS) of developing zebrafish, influencing their swim pattern. This bioorthogonal strategy has enabled ‐for the first time‐ modification of cognitive activity by releasing a neuroactive agent directly in the brain of an animal.

Characterisation. NMR spectra were recorded at 300 K on a 500 MHz Bruker Avance III HD spectrometer. Chemical shifts are reported in parts per million (ppm) relative to the solvent peak ( 1 H NMR DMSO-d6 2.50 ppm; 13 C-NMR DMSO-d6 39.52 ppm). Data are presented as follows: chemical shift (ppm), multiplicity (s = singlet, d = doublet, t = triplet, m = multiplet, q= quartet), coupling constant J, and integration. Rf values were determined on Merck TLC Silica gel 60 F254 plates under a 254 nm UV source. Purifications were carried out by flash column chromatography using commercially available Biotage® Sfär silica column (60 m particle size -10 g). High-Resolution Mass Spectrometry was performed with a Bruker MicrOTOF II. The optical and fluorescent properties of NBD-NHEt (1) and POC-NBD (2) were analyzed by a NanoDrop™ 2000c Spectrophotometer and a NanoDrop™ 3300 Fluorospectrometer, respectively, both acquired from Thermo Scientific™. The purity of POC-NBD (used in cell and zebrafish studies) was >99.9%, as measured by HPLC using an Agilent 1260 Infinity II Preparative LC/MSD system coupled to an Evaporative Light Scattering detector (ELSD). HPLC method: eluent A: water and formic acid (0.1%); eluent B: methanol and formic acid (0.1%); and A/B = 95:5 to 5:95 in 4 min and isocratic 2 min (flow = 1 mL/min). The purity of prodrug 4 (used in cell and zebrafish studies) was >99%, as measured by HPLC using an Agilent 1260 Infinity II Preparative LC/MSD system coupled to an Evaporative Light Scattering detector (ELSD). HPLC method: eluent A: water and formic acid (0.1%); eluent B: acetonitrile and formic acid (0.1%); and A/B = 95:5 to 5:95 in 4 min and isocratic 2 min (flow = 1 mL/min). Prodrug-into-drug conversion experiments were conducted using the same HPLC equipment and method described above. Stock solutions (150 mM) were prepared in DMSO. The optical properties of Au-NPs were analyzed by a NanoDrop™ 2000c spectrophotometer (Thermo Scientific™), and Au loading of the implants was determined by ICP-OES (Perkin Elmer 8300 DV). The hydrodynamic diameter of the nanoparticles was determined by DLS using a Malvern Zetasizer Nano-S (Malvern). The Zeta potential (ζ potential) of the nanoparticles was performed on a Zetasizer Nano-ZS (Malvern). SEM and TEM images were obtained using a FEI Inspect F50 microscope equipped with an EDX analytical system, and a Titan (Thermofisher Scientic, formely FEI) with a Field Emission Gun operating at 300 kV, respectively.

Experimental Procedures and Characterizations
Synthesis of Au-NPs and testing of catalytic properties. The size, capping agents, and crystalline structure of metallic nanoparticles are factors that can affect their catalytic properties. In a preliminary screening, we synthesized monodisperse quasi-spherical particles in a size range of 15 to 150 nm in diameter by varying the Au:citrate ratio in the reaction parameters. [2] As shown in Table S1, smaller citrate-stabilized particles showed higher catalytic activity by fluorogenic studies compared to larger sizes. Therefore, having established that smaller size Au-NPs have superior catalytic activity, we tested a number of synthetic protocols to control NP growth and morphology by changing the reducing and capping agents (see Table S2). Monodisperse Au-NPs with different characteristics were prepared and analyzed by DLS, UV-vis, ζ potential and HAADF-STEM.
Synthesis of negatively charged Au-NP-3 (a.k.a. Au-NPs in the main manuscript): NPs were prepared as previously described [4] with slight modifications. 50 mL glass vial was cleaned with aqua regia to dissolve and lift any traces of metal deposits off the glass surfaces. Distilled water (45 mL) was added and stirred vigorously with a cylindrical magnetic stir bar with pivot ring for the addition of aqueous sodium hydroxide (0.2 M, 1.5 mL) followed by tetrakis(hydroxymethyl)phosphonium chloride (THPC, 80 wt%) (85 mM, 1 mL). After 5 min stirring, freshly dissolved HAuCl4·3H2O (99.9% trace metals basis) (25 mM, 2 mL) was added and the colour changed from yellow to dark brown within seconds. The reaction mixture was stirred overnight at ambient temperature to allow for complete formation of gold cores, protecting the mixture from light to prevent the photothermal decomposition of the precursors. [5] In the particle nucleation mechanism, the in situ generation of formaldehyde from THPC under basic conditions favours the reduction of Au (III) to Au (0). The formaldehyde acts as an active reducing agent, oxidizing to formic, and generating and stabilizing Au-NPs with negative charge (average ζ potential value is -19.9 ± 4.2 mV), as previously reported. [4,6] Full characterization of Au-NP-3 is shown in Figure S2. Synthesis of HTPC-based Au-NPs from Table S1: Au-NPs were prepared as described above for Au-NP-3, maintaining the concentration of the Au (III) salt precursor and the addition volumes of all the reagents. To obtain different particle sizes, the concentration of the reducing agent and NaOH (THPC-NaOH) were modified as follows: 76.5mM:0.17M, 68mM:0.13M, 64mM:0.1M, resulting in Au-NP with sizes of 9.6, 19.7 and 25.4 nm, respectively. Of note, the larger NPs (19.7 and 25.4 nm) were stable for no more than 24 h. Therefore, after synthesis, the NPs were centrifuged, resuspended in water, and tested immediately with the corresponding pro-dye.
The catalytic properties of Au-NP-1, Au-NP-2 and Au-NP-3 described above were tested using the off/on fluorescent probe Pro-Res, which upon O-propargyl cleavage releases strongly fluorescent resorufin. For this screening, a concentration of nanoparticles of 40 µg/mL was added to an aqueous solution of Pro-Res at 40 µM. Either PBS or PBS + 10% of serum (FBS) were used as reaction media. Reactions were shaken at 700 rpm and 37 o C in a Thermomixer and fluorescence intensity analyzed after 24 h in a PerkinElmer EnVision 2101 multilabel reader (Ex/Em: 550 /580 nm). Samples were repeated in triplicate. The results are shown in Figure S1. Crystalline Au-NP-3 featuring decahedral structures demonstrated superior catalytic properties in comparison with larger nanoparticles, synthesized and stabilized by SC (Au-NP-1), and the small faceted Au NPs (twinned nature), synthesized and stabilized by SC and TA (Au-NP-2).
Based on these the fluorogenic studies, Au-NP-3 (Au-NPs) (simply named as Au-NPs from now on) were selected for preparing the implants.
Synthesis and characterization of Au-implants. Solid-supported gold catalysts were prepared by deposition of ultra-small Au-NPs on the surface of a PEG-grafted low-cross-linked polystyrene matrix. Briefly, 10 mg of TentaGel® HL amino-functionalized resins (75 microns) was dispersed in 1.5 mL of the freshly prepared Au-NPs suspension. After sonicating the mixture for 10 min, the resulting resins were collected by centrifugation (5 min at 8000 rpm). This procedure was repeated for a second time and the suspension was stirred at ambient temperature for 12 h in the dark using an eppendorf IKA rotary shaker loopster digital. Subsequently, the Au-implants were collected by centrifugation (5 min at 8000 rpm), washed several times with Milli-Q water, and re-dispersed in water until further use.
Electron microscopy analyses. Au-microimplants were first infiltrated and embedded in a EMBed812 resin polymerizing at 60 o C for 24 h. Then, before curing the resin block, it was stained with OsO4 (4 wt.% in H2O) for 1 h to harden the Au-microimplants and ease sectioning. Then, semithin and ultrathin sections of the Au-microimplant were obtained with a diamond knife (ultra 35°, Diatome) using an Ultramicrotome (Leica EM UC7). For characterization by SEM, the semithin sections (500-1000 nm in thickness) were placed on a pin stub with carbon tape and coated with 15 nm of carbon. Previously, stubs were glow discharged (30s, 15 mA) to enable the sections to be mounted as flat as possible. Au-microimplant blocks after sectioning were also mounted on stubs and coated with 15 nm of carbon to be observed by SEM, using a FEI Inspect F50 microscope and 10-20kV of acceleration voltage. This microscope is equipped with different SEM detectors that obtain: 1) SEM images and composition by using a Back Scattering Electron Detector and 2) elemental chemical analysis by energy-dispersive X-ray microanalysis (EDS). The ultrathin sections (50-70 nm in thickness) were placed on a carbon film on copper grid (150 mesh) and allowed to dry in air. TEM observation was conducted using a Titan (Thermofisher) with a Field Emission Gun operating at 300 kV at the LMA-INA-University of Zaragoza facilities with the assistance of Dr Fernandez-Pacheco.
The microscope spherical aberration corrector (CESCOR-CEOS) allows a point resolution of 0.8 Å. The microscope is fitted with a High-Angle Annular Dark Field (HAADF) detector (Fischione) to operate in Scanning Transmission Electron Microscopy (STEM) mode with Z-contrast imaging.
Quantification of Au content by inductively coupled plasma optical emission spectrometry (ICP-OES). Au-microimplants of different batches (1 mg of sample) were digested with 1 mL of freshly-prepared aqua regia and the samples were left for digestion for 30 min. Afterwards, 50 μL of each sample plus 50 µL of aqua regia was diluted to 4 mL with Mili-Q water to achieve a final concentration of acid of 2.5 % v/v. Au quantification was done on a Perkin Elmer 8300 DV. Calibrations were performed employing Au standards in the same background solution (2.5 % aqua regia) with excellent correlations. Samples were measured in triplicate.

Synthesis and characterization of Pro-Res
Pro-Res was synthesized as previously reported, [1] with slight modifications to the purification method. The crude was purified by silica gel column chromatography MeOH (2.5%  5% v/v)-CH2Cl2 instead of via semipreparative TLC chromatography to give the desired product (7-Propargyloxy-3H-phenoxazin-3-one) as an orange solid (48 mg, 46% yield). Rf  Fluorogenic reaction with naked Au-NPs using Pro-Res. 1 mL solution containing the desired concentration of Au-NPs (40 μg/mL) and 40 μM of Pro-Res was prepared in an Eppendorf tube in PBS. Reactions were shaken at 700 rpm and 37 o C in a Thermomixer and fluorescence intensity measured at different time points in a PerkinElmer EnVision 2101 multilabel reader (Ex/Em: 550 /580 nm). Samples were repeated in triplicate. The results are shown in Figure S1. The concentration of resorufin product (μM) was calculated based on the standard curve of resorufin.
Fluorogenic reaction with Au-microimplants using compound prodye 2. 0.5 mL solution containing the desired concentration of Au-microimplants (0.05, 0.08, 0.1 and 0.5 mg/mL) and 50 μM of prodye 2 was prepared in an Eppendorf tube in PBS containing 30% of methanol (MeOH). The mixtures were shaken at 700 rpm and 37 o C in a Thermomixer for 24 h. From this initial screen (see Figure  S4c-d), the concentration of 0.1 mg/mL Au-microimplants was selected for further testing in the absence (PBS) and presence of serum (PBS + 10% FBS) using 50 μM of prodye 2. Note that 0.1 mg/mL of Au-microimplant is equal to approx. 10 μM of Au (exact calculation 9.65 μM). Fluorescence intensity measured at different time points in a PerkinElmer EnVision 2101 multilabel reader (Ex/Em: 485/535 nm). The conversion values were calculated from fluorescence intensity measurements at λex/em = 485/535 nm using the fluorescence intensity of 1 (50 μM) as 100%. Negative controls: 2 (50 μM) with or without Au-implants (0.1 mg/mL). Each experiment was performed at least in triplicate, and the values given correspond to the mean value  SD of n  3. Naked Au-NPs were tested alongside to determine the effect of serum on freestanding and polymer-supported Au-NPs. The results are shown in the main manuscript and in Figure S4c The results are shown in Figure S7.
Prodrug-into-drug conversion study. Prodrug 4 (4 μL, 150 mM stock solution in DMSO) was added to a H2O:MeOH 7:3 solution (496 mL of the reaction medium) in a 1 mL Eppendorf, containing 1 mg / mL of Au-microimplants. Reactions were carried out for 0, 18, 24, 32 and 44 hours with continuous stirring (700 rpm) at 37 o C using a Thermomixer. Afterwards, the Au-microimplants were collected by centrifugation (5 min at 8000 rpm) and the supernatant was analyzed by LC/MS (Agilent 1260 Infinity II) using an ELSD detector. Fluoxetine 3 (1.2 mM) was dissolved in H2O:MeOH 7:3 solution (0.5 mL) in the same conditions as described above and analyzed as a positive control. Each experiment was performed at least in triplicate, and the values given correspond to the mean value  SD of n  3. Each measurement was taken from distinct samples. R 2 is the coefficient of determination, used as statistical parameter of goodness of fit in the calibration curves. Data analysis was performed using OriginPro 8 statistical software. The results are shown in Figure S12-13.

Biological Studies
Cell culture. Lung adenocarcinoma A549 cells (a kind gift from Dr Wilkinson) and neuroblastoma SH-SY5Y cells (a kind gift from Prof. Kathryn Ball) were cultured in Dulbecco's Modified Eagle Media (DMEM) supplemented with serum (10 % of FBS) and L-glutamine (2 mM). Each cell line was checked for mycoplasma before use and maintained in normoxic conditions at 37 o C and 5% CO2. Cells were seeded in a 96-well plate at 1000 cells/well for A549 or 7500 cell/wells for SH-SY5Y and incubated for 24 h before treatment.
Study of the biocompatibility of Au-microimplants. The tolerability of cells to Au-microimplants was tested by performing doseresponse studies in A549 and SH-SY5Y cells. Cells were plated as indicated above. Each well was then replaced with 100 L of fresh media containing Au-microimplants at 0.6, 0.8, 1 and 1.2 mg/mL for both cells. After 1 week, PrestoBlue TM cell viability reagent (10 % v/v) was added to each well and the plate incubated for 90 min. Fluorescence emission was detected using a PerkinElmer EnVision 2101 multilabel reader (Ex/Em: 540/590 nm). Experiments were performed in triplicate. All conditions were normalized to the untreated cells (100 %). The results are shown in Figure S8 a-b.
Cell viability study of the drug and pro-drug. A549 and SH-SY5Y cells were plated as indicated above. The corresponding wells were then replaced with a solution of compound 3 or 4 at different concentrations (2.5, 10, 25, 50, 100 and 150 μM) containing 0.1% v/v of DMSO. Untreated cells were incubated with DMSO (0.1% v/v). After 1 week, PrestoBlue TM cell viability reagent (10 % v/v) was added to each well and the plate incubated for 90 min. Fluorescence emission was detected using a PerkinElmer EnVision 2101 multilabel reader (Ex/Em: 540/590 nm). Experiments were performed in triplicate.
Of note, see Figure S8c, there was a minor but significant reduction of cell proliferation after treatment with Fluoxetine 3. This is in agreement with a recent study that reports anticancer properties for 2. [7] This effect was not observed in cells treated with prodrug 4, which further demonstrates the inactivation of 4.

In Vivo Experiments
Zebrafish husbandry. Animals were bred and raised at the Queen's Medical Research Institute, University of Edinburgh, in accordance with the Animals in Scientific Procedures Act 1986 and under British Home Office project licence 70/8805. Embryos were obtained through natural spawning from the adult WIK wild type zebrafish line. [8] Embryos were maintained at low density (~40/50 mL zebrafish water) at 28 o C with a 14h light/10h dark cycle. For implant studies, on reaching larval stage 3 days post-fertilisation (dpf), animals were implanted with resins at room temperature then maintained as before until 5 dpf.
Implantation procedure. Implantation of Au-microimplants (catalytic devices) or OH-microimplants (catalytically inactive; negative control) was carried out at 3 dpf. Larvae were anaesthetised with 2.5 mM ethyl 3-aminobenzoate methanesulfonate (Tricaine) in zebrafish water (pH 7.2) before being immobilised in 2% (w/v) low melting point agarose (in zebrafish water). The head was exposed, and an incision made in the skin between the developing eye cups using an optical surgery scalpel. Implants (60-75 m in diameter) were positioned inside the incision using custom-prepared glass needles pulled from borosilicate glass. Larvae were released from the agarose and allowed to recover in zebrafish water for 1 h before being incubated in drugs of interest at 28 o C on a 14 h light/ 10 h dark cycle until 5 dpf. Full procedure of the implantation method is shown in Figure S9.
Confocal studies using prodye 2. Au-microimplants were implanted as described and larvae incubated in 1% DMSO (negative control) or with 2.5 μM of 2 in 1% DMSO for ~44h. The zefrafish were then imaged using a Olympus FV3000 Confocal Laser Scanning Microscope using a 20x objetive. The setting of the confocal microscope were as follows: exc= 488 nm and em= 514-553 nm.
Incubation for behavioral studies. In initial trials, without implants, control recordings were obtained in 1% DMSO in zebrafish water. Larvae were then pre-incubated in drug of interest in 1% DMSO for 2h prior to treatment recording. For Au-microimplants and control microimplants trials, devices were implanted as described and larvae incubated in 1% DMSO or prodrug 4 in 1% DMSO for ~44h.
Behavioural assay and data acquisition. Spontaneous swimming behaviour of 5 dpf larvae was observed in each treatment group. Larvae were acclimatised to room temperature for 1h prior to recording and screened for normal developmental appearance. [9] For recordings with implants, larvae in which the implant was not centred in the head were discarded. Individual larval swimming was tracked in a cell culture dish (35mm x 10mm) using EthoVision XT 7 software (Noldus Information Technology) via a Sony ExwaveHAD B&W video camera. The preparation was lit from below by a light box. Recordings were of 10 min duration in each condition. Acquisition of total distance (mm) and mean and maximum speed (mm s -1 ) was automatically performed by the software. Swimming parameter data are reported ± standard error of the mean (SEM). All statistical analysis was carried out on raw data. Data were assessed for normality and analysed with appropriate parametric or non-parametric tests as described in the results. All statistical tests were performed using GraphPad Prism 8.3 (GraphPad Software, LLC); ns -no significance, *P<.05, **P<.01, ***P<.001, ****P<.0001. Where data are presented as percentage of control in the text, the raw data are also stated.

SUMMARY OF RESULTS:
Influence of psychotropic drugs in larval zebrafish behavior. To select compounds for further study, initial trials were performed in intact larvae (no implants). Distance travelled and speed of spontaneous larval swimming in a 10-minute window was measured to determine the effects of NMDA, GABA and Fluoxetine 3 exposure on normal behaviour. All solutions were prepared in 1% DMSO in zebrafish water. Following control recordings in 1% DMSO, 5 dpf larvae were treated with 150 M GABA, 100 M NMDA or 50 M Fluoxetine 3 for 2 h (see Figure S10). 2 h incubation in 100 M NMDA had no effect on measured swimming parameters (Wilcoxon Signed-Ranks test: P=.8408; n=20). 150 M GABA produced a decrease in average speed (2.90mms -1 ± 0.26 mms -1 to 2.04 mms -1 ± 0.28 mms -1 ; paired t test: t(19)=2.169, P=0.0430; n=20) and a small but insignificant decrease in distance (t(19) = 2.039, P=.0556). 50 M Fluoxetine 3 significantly altered both swimming parameters, as described below and was thus selected for further investigation.
Prodrug 4 does not elicit a reduction in larval swim distance and speed. Distance travelled and speed of spontaneous larval swimming in a 10-minute window was measured to determine the effects of 4 exposure on normal behaviour. Larvae were treated with 50 M drug 3 or prodrug 4 in solution with 1% DMSO in zebrafish water. Control recordings were made in 1% DMSO in zebrafish water. As shown in Figure S11, 2 h incubation in 50 M drug 3 reduced mean total distance travelled from 2285mm ± 142.70mm to 859.50mm ± 92.01mm (paired t test: t(19) = 11.49, P<.0001; n=20) and mean swim speed from 3.81mms -1 ± 0.24 mms -1 to 1.43 mms -1 ± 0.15 mms -1 (t(19) = 11.5049, P<.0001). Treatment with 50 M prodrug 4 had no effect on these swimming parameters (Wilcoxon Signed-Ranks test: P=0.4304). These data indicate that drug 3 can modulate zebrafish larval swimming behaviour, and that its propargyl carbamate derivative 4 is effective in preventing its action in vivo.
OH-microimplants are ineffective in deprotecting prodrug 4 in vivo. Behavioural assays were repeated in zebrafish larvae grafted with carboxyl-functionalized Tentagel resins (OH-microimplants) to confirm that deprotection of prodrug 4 in vivo requires the presence of Au, see Figure S14c   Prepared as previously described, [2] using sodium citrate (SC) for the reduction of HAuCl4 in water.
5.2 ± 0.27 nm at 518 nm -24.9 ± 4.5 mV Au-NP-3 Prepared as described above, using THPC for the reduction of HAuCl4 in water.
Prepared as described above, using THPC for the reduction of HAuCl4 in water under basic conditions.

RESULTS HIGHLIGHTS:
The effects of psychotropic drugs on larval zebrafish behavior were varied. Incubation with NMDA had no effect on measured swimming parameters. 150 M GABA produced a decrease in average speed (83.85% ± 16.35% of that measured with DMSO controls) and a small but insignificant decrease in distance (86.61% ± 16.31% of that measured with DMSO controls). 50 M fluoxetine significantly reduced the mean total distance to 37.79% ± 3.22% and the mean swimming speed to 37.81% ± 3.21% of that measured with DMSO control values. Figure S11. Analysis of the pharmacological properties of Fluoxetine 3 and prodrug 4. a-c) Agonist activity in 5-HT2B-expressing cells. a) Prodrug 4, b) Fluoxetine 3, and c) BW-723C86 (positive control). Dose response curve and IC50 value were generated by OriginPro 8 software. Error bars: ± SD from n = 2. d,e) Effects of 3 and 4 on 5 dpf larval zebrafish swimming. Larvae were treated with 50 M of 3 or 4 in solution with 1% DMSO in zebrafish media for 2 h or 44 h, respectively. Control recordings were made in 1% DMSO in zebrafish media. d) Distance travelled and e) speed by zebrafish in a 10-minute window following 2 h pre-incubation with 3 or 44 h pre-incubation with prodrug 4. Error bars: ± SEM; n = 20.
RESULTS HIGHLIGHTS: 50 M drug 3 reduced mean total distance to 37.79% ± 3.22% of that measured with DMSO controls. Mean swim speed was 37.81% ± 3.21% lower of that measured with DMSO controls. Treatment with 50 M prodrug 4 had no effect on these swimming parameters. These data indicate that drug 3 can modulate spontaneous larval swimming behaviour, and that its propargyl carbamate derivative 4 is effective in preventing its action in vivo.   RESULTS HIGHLIGHTS: Au-implants alone had no effect on measured parameters compared to control (1% DMSO). 50 M prodrug 4 treatment of zebrafish grafted with Au-microimplants reduced mean total distance to 50.34% ± 6.92% of that measured with Au-microimplant controls. Mean swim speed was 53.53% ± 7.52% of that measured with Au-Implant control values. Bioorthogonal intracranial release of 3 by Au-microimplants had similar effect to direct treatment with fluoxetine in modulating spontaneous zebrafish larval swimming. Neither mean total distance nor mean speed of larvae grafted with OH-microimplants was influenced by 44 h incubation with prodrug 4.