Atropine-functionalized gold nanoparticles binding to muscarinic receptors after passage across the intestinal epithelium

Gold nanoparticles have a high potential to be a treatment of diseases by their specific drug delivery properties and multivalent receptor stimulation. For the present project, spherical gold nanoparticles were synthesized and functionalized with the muscarinic receptor antagonist atropine (Au-MUDA-AT NPs). The diameter of the gold core could precisely be controlled by using different synthetic methods and reducing agents resulting in functionalized gold nanoparticles with diameters ranging from 8 to 16 nm. The ability to interact with intestinal muscarinic receptors is size-dependent. When using intestinal chloride secretion induced by the stable acetylcholine derivative, carbachol, as read-out, the strongest inhibition, i.e. the most efficient blockade of muscarinic receptors, was observed with 13 nm sized Au-MUDA-AT NPs. Functional experiments indicate that Au-MUDA-AT NPs with a diameter of 14 nm are able to pass the intestinal mucosa in a time-dependent manner after administration to the intestinal lumen. For example, luminally administered Au-MUDA-AT NPs inhibited contractions of the small intestinal longitudinal muscle layer induced by electrical stimulation of myenteric neurons. A similar inhibition of basolateral epithelial receptors was observed after luminal administration of Au-MUDA-AT NPs when using carbachol-induced chloride secretion across the intestinal epithelium as a test system. Thus, Au-MUDA-AT NPs might be a therapeutic tool for the modulation of intestinal secretion and motility after oral application in the future.


Introduction
Since 2020, lipid nanoparticles (NPs) gained enormous attention, as this type of NPs is currently being used worldwide as carriers for vaccines against SARS-CoV-2 in the vaccines of Pfizer-BioNTech and Moderna [1,2]. The lipid NPs have been shown to function as protective carriers for modified messenger ribonucleic acid (mRNA) because ribonucleases would otherwise degrade the mRNA inside the body before it could enter the cells and reach the point of action. Therefore, a coating with lipid NPs increases the stability of the mRNA.
One approach in the use of lipid-based NPs is to encage desired drugs in order to transport them through the body. These active substances often possess structures with hydrophobic properties, which makes it necessary to encapsulate them. Furthermore, it is possible to store drugs in a hydrophobic region of a polymer layer-coated Au NP ( figure 1, left). These have a hydrophilic outside, which provides sufficient stability of the NPs in aqueous media and ensures an improved transport across biological barriers [22]. Furthermore, a layer-by-layer coating can be applied, in which DNA and RNA may be encapsulated and transported [23]. These modifications may lead to enhanced gene delivery and have been used by Shahbazi et al. in their work on CRISPR-functionalized Au NPs [24]. Lee et al. used the layer-by-layer coating in their work with silencing RNA, resulting in a gene silencing effect within cancer cells [25].
Another approach is the attachment of the active substances on the ligand periphery (figure 1, right). A linker molecule anchors to the NP surface on one side, while various targeted moieties can be attached to its other side [5,9]. This procedure was performed by Brown et al. with the platinum anti-cancer agent oxaliplatin, which was attached via a supramolecular complex to the Au NPs [26]. Although the injected Au NPs showed higher cytotoxicity compared with the free drug, they were effective against colon carcinoma with an IC 50 of 0.495 nM in comparison with an IC 50 of 0.775 µM of the free drug oxaliplatin indicating an increase in affinity to the therapeutic target structures [26].
In addition to drugs, also other bioactive compounds, such as neurotransmitters or paracrine mediators, can be attached to the Au NPs. Gasiorek et al. showed that histamine bound to Au NPs with the linking unit 11-mercaptoundecanoic acid (MUDA) stimulates histamine receptors on epithelial membranes of a rat already in the subnanomolar range [9]. Furthermore, they observed an enormous potentiation of the receptor activation, as a concentration of 10 pM Au-MUDA-histamine was equieffective to 10 µM native histamine. Even when assuming that each Au-MUDA-histamine particle carried approximately 10 000 molecules of histamine, this corresponds to a strong increase in affinity which was attributed to multivalent interactions [9]. Such a potentiation was also observed in our previous studies on carbachol-functionalized Au NPs [5]. Carbachol (CCh) represents a pharmacologically interesting compound, since it is a stable derivative of acetylcholine, which is not degraded by acetylcholine esterase, the dominant enzyme for cleavage of the native substance, acetylcholine. Thus, muscarinic acetylcholine receptors can be successfully stimulated by CChfunctionalized Au NPs. These might offer potential therapeutic applications, as CCh itself is used therapeutically for glaucoma treatment to reduce intraocular pressure; it has also been used to stimulate the motility of the gastrointestinal tract during post-operative ileus or post-operative retention of urine via stimulation of the motility of the urogenital tract.
Whereas our previous studies were designed to attach neurotransmitters or hormones stimulating distinct G-protein coupled receptors [8], the present experiments focus on the attachment of the receptor antagonist atropine. When absorbed after oral administration, atropine competes with acetylcholine and acts as blocker of muscarinic receptors including M 1 and M 3 receptors present at the basolateral membrane of the epithelium involved in the activation of epithelial secretion [27] or M 2 and M 3 receptors involved in the stimulation of gastrointestinal motility [28]. Thus, there is interest in whether Au NPs can function as carriers for atropine and whether they still cross the epithelial barrier to act on the basolateral side of the epithelium or the smooth muscle layers below the mucosa.
Furthermore, atropine or other anticholinergic drugs such as hyoscine butylbromide are currently used to treat spasms in the gastrointestinal tract [29]. Thus, atropine-functionalized Au NPs (Au-MUDA-AT NPs) could have potential therapeutic applications, as atropine or related anticholinergic drugs are widely used in medicine, e.g. to treat bradycardia, as premedicamentation during narcosis, royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 or to overcome intestinal spasms [30]. Nanoparticles are not only passive drug vehicles. In particular, their effects are generally regarded to be strongly dependent on their physico-chemical properties such as e.g. shape, size, chemical composition [31,32]. Therefore, Au-MUDA-AT NPs of different sizes were synthesized and their effects on the intestinal tissue were investigated.

Nanoparticle syntheses
Monodisperse, spherical Au NPs were synthesized in order to use them as starting particles for a further biofunctionalization with atropine. The citrate reduction method was employed, as described in our previous publications [5,8] Furthermore, smaller Au NPs (Au-MUDA 8 nm, Au-MUDA 9 nm (figure 2a), Au-MUDA 10 nm) were synthesized according to our published procedure [8]. Here, MUDA was used as the stabilizing ligand and t Bu-amine borane complex as a stronger reducing agent [33], resulting in slightly smaller Au NPs as synthesized using the citrate reduction method [34]. All Au NP dispersions were obtained in H 2 O and analytical characterizations revealed stable, spherical and monodisperse Au NP dispersions.

Ligand synthesis and nanoparticle functionalization
Hereinafter, the biofunctionalization of Au NPs with atropine is described. Atropine itself is found as the poison of the deadly nightshade Atropa belladonna (figure 3a) in nature. The chemical structure of this compound already contains an ester function (figure 3c, black). In order to immobilize it onto the Au NPs, atropine (purchased in pure form from Sigma Aldrich) was linked with MUDA as a thiol-containing bifunctional linker. An ester bond was formed with atropine's free hydroxyl group and MUDA's acid moiety (figure 3c). The esterification was performed under mild reaction conditions under inert atmosphere according to a modified procedure described by Neises & Steglich [35]; scheme 1, top displays the ligand synthesis. MUDA was dissolved in anhydrous dichloromethane (DCM) and catalytic amounts of 4-(dimethylamino)pyridine (DMAP) (0.1 eq.) were added. The coupling reagent diisopropyl carbodiimide (DIC) was added to the cooled solution and then the reaction mixture was treated with atropine. In particular, sterically demanding and acid-labile reactants may undergo reactions under mild conditions at room temperature in this approach. Remaining acid and other side products were removed in the subsequent aqueous work-up. The crude product was purified over a silica plug. Traces of the side product N,N 0 -diisopropylurea (DIU) remained and could not be removed completely during various work-up steps. However, these were tolerated, as the desired biofunctional ligand possesses a thiol moiety, which binds more favourably to the NPs.
The ligand was analysed using NMR and IR spectroscopy. The 1 H-NMR spectrum shows all expected proton resonances. Characteristic resonances in the range from 1.63 to 1.10 ppm correspond to the long alkyl chain of the thiol spacer MUDA. Furthermore, aromatic proton signals between 7.30 and 7.18 ppm can be assigned to the phenyl unit of atropine and suggest a successful ligand synthesis. IR royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 measurements were performed to verify the existing binding types within the product. The spectrum of MUDA-AT reveals distinct ν C-H resonances in the range of 3000-2800 cm −1 , which are significantly stronger than in the spectrum of AT and thus indicate the long alkyl chain of MUDA. Furthermore, MUDA-AT shows an intense ν C-S vibration at 629 cm −1 , suggesting the existence of a thiol.
With the synthesized ligand MUDA-AT, functionalization of the Au NPs was performed in a ligand exchange reaction (scheme 1, bottom). For this purpose, an Au NP dispersion of the preferred size was degassed with argon. Then, MUDA-AT was dissolved in DMSO and slowly added in a large excess under vigorous stirring at room temperature. The ligand was added in a dissolved form in order to ensure rapid and efficient homogenization. Furthermore, agglomeration of the Au NPs upon the addition of bulk material is prevented.
The obtained pink and stable Au NP dispersion was purified via dialysis in order to remove free ligands and further side products. All NP samples were characterized with TEM, DLS and UV/Vis. Furthermore, the organic surface structure was confirmed using NMR and IR spectroscopy. TEM images of Au-MUDA-AT NPs show predominantly spherical, monodisperse NPs (figure 4). No remarkable change in the morphology of the NPs occurred during their functionalization with MUDA-AT and they furthermore possess a small size distribution in their expected diameters (table 1) between Ø 8 nm (figure 4a) and Ø 16 nm (figure 4d). Moreover, Au-MUDA-AT NPs with Ø 14 nm (figure 4c) even arranged themselves hexagonally on the TEM grid, indicating a particularly high monodispersity. No agglomeration was observed in all approaches; thus, MUDA-AT reveals to be a suitable ligand to sufficiently stabilize Au NPs. The d hydr values all revealed to be slightly larger after the functionalization with MUDA-AT (table 1), which is expected, since a new sterically high demanding ligand is found on the Au NP surface. The UV/Vis spectra of all Au-MUDA-AT NPs (figure 5) show a distinct plasmon resonance with an absorption maximum λ max each in the range of around 530 nm. Since only one relatively narrow absorption band with a noticeable rise can be observed, a stable dispersion with monodisperse NPs can be assumed. The 1 H-NMR spectra of

. Ussing chamber experiments
Crossing the intestinal barrier is a major obstacle for many drugs, limiting their therapeutic use. For a future nanoparticle-based oral drug delivery, which might increase the absorption of drugs, it is necessary that Au-MUDA-AT NPs can overcome the intestinal barrier to reach the muscarinic receptors located on the basolateral cell side of the epithelium [36].
The stable acetylcholine derivative carbachol (CCh) is well known to stimulate muscarinic receptors and induce intestinal chloride secretion which can be measured as increase of short-circuit current (I sc ) in Ussing chamber experiments [37]. This CCh-induced increase of I sc is inhibited by atropine [38]. The ability of Au-MUDA-AT NPs to diminish the CCh-induced effects in the intestine after mucosal application was measured in Ussing chamber experiments and compared with the response obtained with native atropine. For this purpose, segments of rat jejunum were incubated with Au-MUDA-AT NPs (Ø 14 nm, 2 × 10 −10 mol l −1 ), atropine (2 × 10 −8 mol l −1 ) or equivalent volumes of buffer on the mucosal (i.e. the luminal) side of the tissue for different periods of time (0.5, 1 or 2 h). This was   royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 followed by administration of CCh (5 × 10 −5 mol l −1 ) at the serosal side to stimulate basolaterally located muscarinic receptors at the epithelial cells. The CCh-induced current, measured as increase in I sc above the baseline just prior to administration of CCh (ΔI sc ), was registered (figures 7 and 8). Tissues pre-treated with Au-MUDA-AT NPs or atropine showed a significantly reduced response ( p < 0.05) to CCh when compared with time-dependent control tissues, which were not treated with cholinergic antagonists, but received the buffer for NP administration, only. Furthermore, the inhibition of the CCh response by Au-MUDA-AT NPs was intensified, when incubation time was increased from 0.5 to 1 or 2 h; a phenomenon, which was not observed for native atropine. This suggests that native atropine is absorbed quicker by the epithelium than atropine bound to Au NPs, as native atropine shows its maximum inhibitory effect already at an earlier time point. Apparently, the Au-MUDA-AT NPs crossed the intestinal epithelium in a time-dependent manner: the longer the incubation period lasted, the more NPs reached the basolateral side of the epithelium. At the end of each experiment, forskolin (9 × 10 −6 mol l −1 at the serosal and mucosal side), a cAMP-dependent activator of chloride secretion [39], was administered. No significant differences in the change of I sc (ΔI sc ) were observed in the forskolin response between all groups, indicating that Au-MUDA-AT NPs did not unspecifically affect the secretory capability of the epithelium. In addition, 4000 transmittance (arb. units)  royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 tissue conductance (G t ) was not altered after drug application, indicating that neither atropine nor Au-MUDA-AT NPs affected the viability of the intestinal epithelium (data not shown).

Organ bath experiments and electrical field stimulation
As the modulation of intestinal motility would be an interesting therapeutic application, e.g. for the treatment of intestinal spasms, organ bath experiments were performed to measure the change of isometric force of the tunica muscularis after Au-MUDA-AT NPs administration. First, the general ability of Au-MUDA-AT NPs to modulate jejunal contractility was shown after serosal application. To do so, intestinal segments in their natural geometry (i.e. not inverted as done in later experiments, see below) with the tunica muscularis in direct contact to the organ bath solution, were used, so that the Au-MUDA-AT NPs administered to the bath solution had direct access to the muscarinic receptors within the muscle layer. Au-MUDA-AT NPs (Ø 14 nm) at a concentration of 5 × 10 −12 mol l −1 led to a significant decrease ( p < 0.05) of CChinduced (5 × 10 −7 mol l −1 at the serosal side) isometric force by approximately two-thirds in comparison with the untreated group (0.52 ± 0.13 g versus 1.66 ± 0.38 g; mean ± s.e.m., n = 8; data not shown). Second, the modulation of contractility after intestinal uptake of the NPs was investigated. For this purpose, segments of the jejunum were inverted, so that the mucosa faced the outer surface of the resulting tube and atropine or Au-MUDA-AT NPs, which were administered to the organ bath, first had to pass the mucosa before they could reach muscarinic receptors on the smooth muscle cells of the tunica muscularis. These inverted jejunal segments were electrically stimulated by electric field stimulation (current pulses of 250 mA, pulse duration of 1 ms, frequency 10 Hz, train duration 10 s), which was repeated every 2 min. These pulses evoke an increase in isometric force due to the release of excitatory neurotransmitters from myenteric neurons innervating the smooth muscle cells (figure 9a). These so-called on-contractions, which occur simultaneously to electrical field stimulation (EFS), are described to be atropine-sensitive [40]. Indeed, atropine (2 × 10 −5 mol l −1 at the mucosal side) led to a reduction of contraction strength under EFS approximately more than 60% after a 30 min incubation period (figure 9b, table 2). The mucosal application of Au-MUDA-AT NPs required a higher concentration of 2 × 10 −10 mol l −1 to be effective. In this concentration, the atropine-functionalized NPs inhibited the EFS-induced contractions significantly in a time-dependent manner. For statistical analysis, the EFS-induced increases of isometric force prior to drug application for each muscle segment were averaged over 30 min (i.e. the average of 15 on-contractions during the untreated control period (pre-drug period) was calculated for each jejunal segment) and set as 100%. All following changes of isometric force were averaged in periods of 10 min (atropine) or 30 min (no drugs, Au-MUDA-AT NPs) and compared in % with this reference value calculated for the pre-drug period. In the time-dependent control series, i.e. without administration of any drugs, these EFS-induced contractions did not decrease but had in contrast the tendency to increase slightly over time (table 2). When atropine (2 × 10 −5 mol l −1 at the mucosal side) was administered, a fast decrease of the EFS-induced contractions was observed, which reached a stable plateau already 10 min royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 after administration of this muscarinic antagonist (table 2). A significant decrease in the EFS-induced contractions was also observed after administration of Au-MUDA-AT NPs (2 × 10 −10 mol l −1 at the mucosal side), which developed, however, more slowly in comparison with native atropine (figure 10, table 2). This indicates that Au-MUDA-AT NPs cross the intestinal epithelium in a time-dependent manner as already suggested from the data obtained in the Ussing chamber experiments. Taken together, the findings from the uptake studies indicate that Au-MUDA-AT NPs (Ø 14 nm) can cross the intestinal barrier in a time-dependent manner and bind to muscarinic receptors on the basolateral side of the enterocytes. Further studies are necessary to understand how Au-MUDA-AT NPs overcome the intestinal epithelium. Generally, there are different strategies to cross membranes including endocytosis, which plays a major role as cell entry mechanism for NPs [41]. By binding specific ligands to the NPs' surface, the NPs can actively target specific cell receptors and enter the cells via receptor-mediated endocytosis (RME) [42]. For example, insulin-loaded polymeric NPs whose surface was coated with the protein-ligand transferrin were internalized after specific receptor binding and induced hypoglycaemia in rats [43]. Furthermore, Au NPs  whose surface was not loaded with a specific ligand were endocytosed via RME after non-specific serum protein adsorption onto the nanoparticles' surface [44]. Another strategy for crossing the intestinal barrier is paracellular trafficking, e.g. after tight junction opening [45,46]. As there is a great diversity in the physico-chemical properties of NPs in general (e.g. composition, stabilizing agent, surface modification, surface charge, size, shape, etc.) which influence their cellular uptake [47][48][49], the specific cell entry of an NP has to be evaluated individually. Therefore, future studies are necessary to investigate how Au-MUDA-AT NPs exactly cross the intestinal barrier. For a possible in vivo application, also environmental factors as the acidic pH in the stomach, the enzymes of the digestive tract or the intestinal mucus will determine the uptake of the NPs and have to be considered during synthesis. Additionally, the NPs' biodistribution in the organism after crossing the intestinal barrier has to be evaluated and can be affected by the chemistry of the NPs [50,51]. In 2020, Enea et al. performed in vitro studies with Caco2-cells and also 11-MUDA acid-coated gold nanospheres [52]. By contrast to our results, neither 15 nm nor 60 nm sized NPs could cross the cells but only accumulated intracellularly. This indicates that the intestinal crossing of our 14 nm sized Au-MUDA-AT NPs occurred due to the slight size difference or the surface functionalizsation. Apparently, Au-MUDA NPs induce only low cytotoxic effects [52,53], even high-dose oral application of Au NPs over 14 days did not cause severe toxicity [54], offering a promising perspective for a possible future oral drug delivery of Au-MUDA-AT NPs. Not only the local treatment of intestinal diseases, as intestinal spasms, is of therapeutic interest, also systemic delivery of different functionalized Au NPs would be beneficial. Thus, the results of our uptake studies might help in the development of novel nanoparticle-based therapies.

Impact of Au NP core sizes
As the effects of NPs are strongly dependent from their properties [31,32], the effects of Au-MUDA-AT NPs with different diameters (8-16 nm) were tested in Ussing chamber experiments ( figure 11). After 20 min of Au-MUDA-AT NPs incubation (2 × 10 −10 mol l −1 ) at the serosal side of the jejunum, CCh (5 × 10 −5 mol l −1 at the serosal side) was administered. Again, forskolin (9 × 10 −6 mol l −1 at the serosal and mucosal side) was administered as viability control at the end of each experiment (figure 11) [39]. For a better stabilization of I sc prior to drug application, tissues were pre-treated with the nonsteroidal anti-inflammatory drug indomethacin (10 −6 mol l −1 at the mucosal and serosal side). Indomethacin inhibits stimulation of epithelial secretion via release of prostaglandins [55]. Au-MUDA-AT NPs with a diameter of 13 nm showed the strongest inhibition of the carbachol-induced I sc with approximately more than 75% compared with control tissues. The strength of inhibition decreased steadily with smaller and larger diameters of Au-MUDA-AT NPs ( figure 12). In none of the experimental series, the secretory response evoked by the cAMP-dependent secretagogue forskolin (9 × 10 −6 mol l −1 at the serosal and mucosal side) was significantly lower than in the untreated control royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 group, clearly excluding unspecific inhibition of the transporters involved in anion secretion by the NPs. The tissue conductance was not altered after NPs treatment, too (data not shown). These results prove that the effects of Au-MUDA-AT NPs are strongly size-dependent with 13 nm sized Au NPs showing the strongest effects. This might be explained by the different curvature of different-sized NPs [56]. The given curvature influences ligand-receptor interactions as a specific angle or arrangement of the ligands on the NP surface can promote or hinder the formation of multiple receptor-ligand complexes at the same time. Other studies have also focused on the relation between the effects of the NPs and their size, but mainly larger size distributions were investigated. The focus on this small size range from 8 to 16 nm shows that fine-tuning of the NPs' properties is very important to reach the most promising effects. As Jiang et al. have described in their study in 2008, not only the surface or the functionalization determines the NPs' interaction with cells but also the NPs and their specific properties themselves [32]. These are important findings that must be regarded when NPs are used in therapeutic application.

Conclusion
In conclusion, the successful synthesis and the proof of biological effectiveness of atropine-functionalized gold nanoparticles with a narrow size distribution were shown. Functional studies demonstrated that Au-MUDA-AT NPs can cross the intestinal barrier and modulate the secretion and motility of the gut. royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 The size-dependent interaction with muscarinic receptors emphasized how important the nanoparticles' properties are. Our findings suggest a possible use of functionalized gold nanoparticles as a future treatment of intestinal diseases after oral application.

Experimental section 4.1. Chemical, materials and laboratory techniques
All chemicals were purchased from Acros Organics, Alfa Aesar, Carl Roth, Fisher Scientific, Fluka, Merck, Santa Cruz Biotechnology, Sigma Aldrich or TCI and used without further purification. Organic solvents were distilled before use or purchased in an anhydrous state and stored over molecular sieves. Chromatographic purifications were performed using Merck silica gel 60 (0.040-0.063 mm). Thin layer chromatography (TLC) was performed on Merck aluminium-backed plates with silica gel and fluorescent indicator (254 nm). For indication, UV light (λ = 254 nm/365 nm) was used.
Reactions were performed under inert conditions (argon atmosphere 99.9999%, Air Products) using standard Schlenk line techniques with oven-dried glassware unless stated otherwise. All water-based (nanoparticle) syntheses were performed in demineralized water or arium water. All glass vessels were washed with aqua regia and demineralized water prior to use. For dialysis of the NP solutions, membranes of regenerated cellulose 'ZelluTrans' with different pore sizes (MWCO 6000 and MWCO 12 000) were used. These were purchased from Carl Roth GmbH. The dialysis membrane was immersed in the dialysing solvent for 30 min prior to use. All dialyses were performed at room temperature.

Analytical methods
1 H-NMR and 13 C-NMR spectra were recorded on Bruker Avance 400 MHz (AV 400), Bruker Avance II 300 MHz (AV 300) and Bruker Avance II+ 600 MHz (AV II 600) spectrometers at the Institute of Organic Chemistry, University of Cologne. All measurements were performed at room temperature. Chemical shifts are given in ppm relative to respective solvent peaks. 1 H-NMR data are reported as follows: chemical shifts (multiplicity [ ppm], classification). Multiplicity is recorded as s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet. All spectra were displayed with the software MestReNova. 1 H-NMR spectra of all Au NP dispersions were recorded. The Au NP dispersion (4-5 ml) was dried in vacuo and subsequently dissolved in D 2 O in order to prepare the samples. To ensure stable dispersions during the measurements, a base (usually NEt 3 ) in D 2 O was added. The measured NMR spectra were then compared with the spectra of the initial Au NP dispersions as well as the synthesized ligands. Impurities of D 2 O and NEt 3 originate from the sample preparation. Mass spectra were recorded on an ESI mass spectrometer (spectrometer (micrOTOF) from Bruker Daltonics (Bremen, Germany)) at the Justus Liebig University Giessen or on an ESI mass spectrometer of Agilent Technologies, model LC/MSD Vl, at the research group of Prof. Berkessel at the University of Cologne. IR measurements were performed on a Perkin Elmer FTIR-ATR (UATR TWO) at room temperature with a maximum resolution of 1 cm −1 . Absorption bands are given in cm −1 . TEM images were taken on a Zeiss LEO 912 (300 kV, LaB 6 -cathode) equipped with a GATAN digital camera. For sample preparation, a 10 µl NP solution was placed on a carbon-coated copper grid. Determination of the average particle size and standard deviation was achieved by measuring 200 individual particles using Image J Fiji. Dynamic light scattering (DLS) measurements were performed on a NanoZS from Malvern. UV/Vis spectra were recorded with an UV-1600PC spectrophotometer from VWR. Elemental analyses for C, H, N and S were acquired by Mr Dirk Pullem on a Eurolab EA Elemental Analyzer at the University of Cologne.

Ligand synthesis 4.3.1. Synthesis of MUDA-AT (according to a method by Neises & Steglich [35])
MUDA (150 mg, 0.69 mmol, 1.0 eq.) was dissolved in anhydrous DCM (10 ml). DMAP (9 mg, 0.07 mmol, 0.1 eq.) was added under argon counter flow and stirred at 0°C. DIC (150 µl, 0.97 mmol 1.4 eq.) was added quickly and stirred for 5 min at 0°C. Then, AT (200 mg, 0.69 mmol, 1.0 eq.) was added and stirred for 15 min at 0°C. Afterward, the reaction mixture was stirred at room temperature for 16 h under argon atmosphere. During the work-up process, the reaction mixture was filtered and the royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 solvent was evaporated. The residue was redissolved in DCM and washed with diluted HCl (2 × 10 ml) and with saturated NaHCO 3 solution (10 ml). The combined organic layers were dried over Na 2 SO 4 , filtered and the solvent was evaporated. The crude product was purified over a silica plug (eluted with DCM). The product could be obtained as a colourless solid in a good overall yield (283 mg, 84%) and was stored at 4°C in the dark.

Animals
Female and male Wistar rats with a body mass of 200-250 g were used for the Ussing chamber and organ bath experiments. The animals were bred and housed at the Institute for Veterinary Physiology and Biochemistry of the Justus Liebig University Giessen at an ambient temperature of 22.5°C and air humidity of 50-55% on a 12 h : 12 h light-dark cycle with free access to water and food until the time of the experiment.

Tissue preparation
Animals were killed in CO 2 narcosis by cervical dislocation followed by exsanguination. The jejunum was extracted and washed with ice-cold buffer solution. For Ussing chamber experiments, the jejunum was placed on a small plastic rod and opened longitudinally before it was fixed between the Ussing half chambers. For the contractility studies with serosal drug application, the jejunum was fixed on a small plastic rod and 2 cm long segments were ligated before they were transferred into organ bath chambers to measure isometric contractions of the longitudinal muscle layer. For isometric force measurement after mucosal drug administration, the jejunum was placed on a small plastic rod with a notch on one end. A ligature was set in the notch to fix the jejunum on the rod and the intestine was carefully inverted so that the mucosa formed the outer surface of the resulting tube. After closing the lower end of the inverted tube with a ligature, the length of the jejunal segments was standardized to 10 cm. Each 10 cm segment was filled with 2.5 ml ice-cold buffer solution. Finally, segments of 2 cm length were ligated from the 10 cm tube and incubated in the organ bath chambers. For each experiment, two segments of the jejunum of each rat were prepared.

Short-circuit current measurements
The mucosa-submucosa preparation was fixed in a modified Ussing chamber bathed with a volume of 3.5 ml on each side of the jejunal tissue. The tissue was incubated at 37°C and short-circuited by a computer-controlled voltage-clamp device (Ingenieur Büro für Mess-und Datentechnik Mussler, Aachen, Germany) with correction for solution resistance. Short-circuit current (I sc ) was continuously recorded on a chart-recorder. I sc is expressed as µEq h −1 cm −2 , i.e. the flux of a monovalent ion per time and area, with 1 µEq h −1 cm −2 = 26.9 µA cm −2 . Tissue conductance (G t ; in mS cm −2 ) was measured every minute by the voltage deviation induced by a current pulse (± 50 µA, duration 200 ms) under open-circuit conditions. Baseline of I sc and G t were calculated as an average of I sc and G t over 3 min prior to each drug application. The maximum increase or decrease of I sc and G t was measured as difference to the baseline during 10 min after drug application.

Isometric force measurements and electrical field stimulation
Change of force (in gram) generated by the jejunal segments due to relaxation and contraction of intestinal smooth muscle cells in the longitudinal layer of the tunica muscularis was recorded under isometric conditions. The organ bath chamber was filled with carbogen-gassed buffer solution, constantly warmed up to 37°C. First, a pre-tension of 2 g was applied, then the tension was manually reduced to 1 g. Subsequently, the following protocols were used for isometric force measurement after serosal (1) and mucosal (2) drug application using either non-inverted (1) or inverted segments of small intestine (2): (1) After an equilibrium period of at least 15 min, Au-MUDA-AT NPs or equivalent volumes of buffer were applied followed by the administration of CCh (5 × 10 −7 mol l −1 at the serosal side) 20 min later. KCl (15 × 10 −3 mol l −1 at the serosal side) served as viability control at the end of each experiment. The baseline prior to drug application was calculated over 1 min, while the maximum increase of royalsocietypublishing.org/journal/rsos R. Soc. Open Sci. 9: 220244 isometric force was calculated as difference to the baseline within a period of 3 min after drug application. (2) After an equilibrium period of 20 min, the inverted tissues were stimulated with an electrical field which was applied by two electrodes. Electrical stimuli were applied with an electric current of 250 mA, a pulse duration of 1 ms over 10 s, a frequency of 10 Hz and were repeated every 2 min. After 30 min of electrical stimulation, the first drug was applied. The baseline was calculated as average of the isometric tension over 10 s prior to each electrical stimulation. EFS-induced change of force was calculated as the difference between the maximum increase of isometric tension within 30 s after EFS and the baseline.

Statistics
Results are given as mean ± standard error of the mean (s.e.m.) with the number (n) of investigated tissues. For the comparison of two groups either Student's t-test or Mann-Whitney U-test was applied. An F-test decided which test method had to be used. When more than two groups had to be compared, an analysis of variances (ANOVA) was performed. If an F-test indicated that variances between the groups were significantly larger than within the groups, Tukey post hoc test was performed. p < 0.05 was considered to be statistically significant.
Ethics. Animal experiments were approved by the named animal welfare officers of the Justus Liebig University (administrative numbers 561_M and 577_M). All experiments were performed according to the German and European animal welfare law.
Data accessibility. Supporting information concerning synthesis and analysis of the nanoparticles and isometric force data referred to as 'data not shown' are deposited at Dryad Digital Repository: https://doi.org/10.5061/dryad.dz08kps05 [57].