Synthesis and Evaluation of Pseudoglucosinolates Releasing Isothiocyanates in the Presence of Azoreductases

Abstract

Naturally occurring isothiocyanates (ITCs) display multiple interesting bioactivities, but their medicinal exploitation is very limited as ITCs are non‐druglike compounds with problematic pharmacokinetic properties. The concept of pseudoglucosinolates provides a novel prodrug approach for the release of ITCs which can be adjusted to different triggers such as enzymatic and chemical microenvironments. Herein, the first adaptation of this concept toward the release of ITCs in the presence of azoreductases within a turn‐on fluorescence probe is reported and its applicability in covalent protein labeling is underlined.

An expansion of the concept of pseudoglucosinolates (psGSLs) is reported introducing the first azoreductase‐responsive psGSL‐based probe releasing an isothiocyanate under simultaneous fluorescence turn‐on in the presence of AzoR, an azoreductase from Escherichia coli. The synthesis, biochemical evaluation, application for protein labeling and conversion in the presence of Escherichia coli are reported.

1. Introduction

Isothiocyanates (ITCs) are potent electrophiles^{[ 1} , ² , ^{3 ]} that naturally occur as mustard oils, which are released from glucosinolates (GSLs) by the thioglycosidase myrosinase. This process is part of a herbivore‐defense mechanism in plants of the Brassicales order, such as radish, broccoli, and white mustard (Figure  1A).^{[ 4} , ^{5 ]} Several naturally occurring ITCs display a portfolio of highly interesting bioactivities. Beyond their primary feeding deterrent activity toward herbivores,^{[ 6 ]} in particular volatile ITCs are utilized by the producing plants to attract pollinators.^{[ 7 ]} Furthermore, ITCs display potent and rather selective antimicrobial activity against pathogenic bacteria^{[ 8} , ^{9 ]} such as methicillin‐resistant Staphylococcus aureus,^{[ 10 ]} enterohemorrhagic Escherichia coli ^{[ 1 ]} and fungal pathogens such as Nosema ceranae.^{[ 11} , ^{12 ]} In addition, numerous reports have been published on different biological effects of ITCs on mammalian cells^{[ 13} , ¹⁴ , ¹⁵ , ¹⁶ , ^{17 ]} including chemoprotective and anti‐inflammatory properties in line with the known health benefitting effects of Brassicales rich diets.^{[ 18} , ¹⁹ , ^{20 ]} For example, sulforaphane (1) and sulforaphane (2, Figure 1 A),^{[ 13 ]} two ITCs released from the GSLs glucoraphenin (3) and glucoraphanin (4) respectively, are known^{[ 21} , ^{22 ]} to covalently inhibit CRM1 (chromosomal region maintenance 1).^{[ 23} , ²⁴ , ^{25 ]} This protein is involved in the nuclear export of compounds from the mammalian nucleus into the cytosol of cells, a pathway heavily overexpressed in different cancer cell types, and hence considered to be a valuable anticancer target.^{[ 26} , ²⁷ , ^{28 ]}

Figure 1.

A) Structures of sulforaphane (1), sulforaphane (2), glucoraphenin (3), and glucoraphanin (4) and GSL breakdown by myrosinase. B) Nitroreductase‐responsive pseudoglucosinolates (psGSLs) by Klahn and co‐workers.^{[ 36} , ^{59 ]} C) Azoreductase‐responsive pseudoglucosinolates (psGSLs).

However, the interesting bioactivities of 1 and 2 (Figure 1A) have only been explored to a very limited extent so far, as ITCs are in general bad drug candidates from a pharmacokinetic perspective. ITCs are hydrolysis sensitive and show high tendency to react with biological nucleophiles.^{[ 4} , ²⁹ , ^{30 ]} Although there are multiple approved drugs bearing electrophilic moieties such as β‐lactams (e.g., the antibiotic Cefalotin, Keflin), α‐ketoamides (e.g., the virostatic telaprevir, Incivo), epoxides (e.g., the proteasome inhibitor carfilzomib, Kyprolis), or Michael acceptors (e.g., the catechol‐O‐methyltransferase inhibitor entacapone, Comtess, the immunomodulatory diethylfumerate, Tecfidera, or the Bruton's tyrosine kinase (BTK) inhibitor ibrutinib, Imbruvica),^{[ 31} , ³² , ^{33 ]} the overall reactivity for hydrolysis of these examples is comparably lower than for ITCs.^{[ 29 ]}

The major drawback of GSLs as naturally occurring prodrugs of ITCs is that their ITC release is dependent on the thioglucosidase myrosinase, a nonhuman enzyme, strongly limiting the exploitation of ITC bioactivities.

However, suitable prodrugs releasing ITCs in a timely and spatial manner may very well become novel drug types. In the course of our ongoing efforts to explore the use of artificial GSLs,^{[ 34} , ^{35 ]} we have recently reported the concept of pseudoglucosinolates (psGSL) (Figure 1B),^{[ 36 ]} in which we have decoupled the natural mechanism of ITC release from the myrosinase by substituting the glucose unit with an autoimmolative para‐aminobenzylthiol unit masked as a nitro‐reductase‐responsive para‐nitrobenzyl moiety. In the presence of NfsB, a nitroreductase from E. coli, these compounds were demonstrated to release the corresponding ITCs in vitro and in vivo.^{[ 36 ]} The psGSL concept allows for the adjustment of ITC release toward noncanonical enzymes and may represent a feasible prodrug approach to harness and exploit the broad bioactivity portfolio of naturally occurring ITCs, allowing for the incorporation of masked, bio‐responsive ITC moieties as secondary warheads in drugs, or the use of ITCs as for enzyme‐responsive covalent‐binding probes in chemical biology.

In this context azo‐reductases as key players in xenobiotic metabolism^{[ 37 ]} in microbes and higher eukaryotes are of particular interest as release‐triggering enzymes. Azoreductase‐activated compounds have been demonstrated to allow for the development of prodrugs targeting colon‐related diseases^{[ 38} , ^{39 ]} or diagnostic probes targeting the tumor‐microenvironment in vivo.^{[ 40 ]}

Here, we now report on the first adaptation of the psGSL concept for azoreductase‐responsive psGSLs (Figure 1C) as imaging probe allowing for the detection of the enzyme's activity and providing a proof of principle for ITC release in the presence of these enzymes as starting point for future development of targeted drug delivery approaches.

2. Results and Discussion

To generate an azoreductase‐cleavable masking group we aimed to merge the para‐aminobenzylthiol into an azobenzene moiety (Figure 1C). Therefore, para‐aminobenzyl alcohol (5) was diazotated in the presence of nitrosonium tetrafluoroborate in acetonitrile at 0 °C and subsequently coupled to N,N‐dimethylaniline (6) in a diazo coupling yielding azobenzene 7 in 87% over two steps (Scheme  1 ). An Appel reaction in the presence of tetrabromomethane and triphenylphosphine gave access to the corresponding bromide 8 in quantitative yield. Substitution of the bromide with potassium thioacetate in dimethylformamide  (DMF) led to formation of the thioacetate 9 in 47% yield. Finally, acidic hydrolysis of the thioacetate gave access to the desired thiol‐containing azobenzene 10 in 48% yield based on recovered starting material, which was stored at −20 °C under argon atmosphere to avoid disulfide formation. Attempts to access compound 10 via basic saponification of the thioester 9 led to exclusive formation of the corresponding disulfide. Disulfide reduction in the presence of dithiothreitol^{[ 41} , ^{42 ]} or Na₂S^{[ 43 ]} might be possible and, however, was not attempted with in this work.

Scheme 1.

Synthesis of azobenzene masked para‐aminobenzyl thiol 10, psGSL_PEG(Azo)–N₃ and psGSL_PEG(Azo)–BODIPY.

Compound 10 was then coupled to the chloro oxime 11 in the presence of diisopropylethylamine to obtain the thiohydroximate 12 in 93% yield.

The chloro oxime 11 was generated in situ from known oxime 13 in the presence of NCS in DMF under light exclusion. Installation of the O‐sulfonate in the presence of SO₃ pyridine complex at 60 °C and subsequent treatment with aqueous potassium bicarbonate solution led to the formation of ps GSL _PEG (Azo)–N ₃ in 60% yield. This product represents a potentially azoreductase‐responsive psGSL bearing an azide moiety for further functionalization.

Additionally, by copper(I)‐mediated azide‐alkyne click reaction with boron‐dipyrromethene (BODIPY) fluorophore 14, we obtained ps GSL _PEG (Azo)–BODIPY in 78% yield. Although labeled with a BODIPY fluorophore, ps GSL _PEG (Azo)–BODIPY is a nonfluorescent compound, as the fluorescence of the BODIPY is quenched by an overlap of the emission band of the BODIPY dye with the excitation band of the attached azobenzene moiety via Förster resonance energy transfer (FRET). Thereby, we have generated a potentially fluorogenic compound, which should lead to a fluorescence turn‐on upon enzymatic cleavage of the FRET acceptor azobenzene unit and formation of a fluorescent ITC.

For the proof of principle for azoreductase‐responsive psGSLs, we decided to employ AzoR,^{[ 44 ]} the azoreductase from E. coli, known for its high substrate promiscuity,^{[ 45 ]} which is usually expressed under various stress conditions.^{[ 45} , ⁴⁶ , ⁴⁷ , ^{48 ]} Because AzoR is not commercially available, we overexpressed AzoR in E. coli BL21 (DE3) from vector pET28^{[ 49 ]} and purified it via immobilized metal affinity chromatography taking advantage of the N‐terminal His₆‐tag.

The resulting purified protein was obtained with an overall yield of 239.5 mg L⁻¹ of medium and stored in TRIS (tris(hydroxymethyl)aminomethan) buffer pH 7.4 at −30 °C. With the enzyme and psGSLs in hand, we then investigated the release of the ITC 15 (Figure  2A) from ps GSL _PEG (Azo)–N ₃ in the presence of AzoR and its cofactors nicotinamide adenine dinukleotide (NADH) and flavin mononucleotide (FMN) via liquid chromatography mass spektrometry (LCMS) analysis. As shown in Figure 2, ps GSL _PEG (Azo)–N ₃ (see Figure 2, A: structure, B1: R _t 4.85 min and C1: mass 693 m z⁻¹ for [M‐K + H + Na]⁺, mass 671 m z⁻¹ for [M‐K + 2H]⁺, and mass 591 m z⁻¹ for [M‐K‐SO₃ + H]⁺) is converted by AzoR at 37 °C within 2 h and the formation of the corresponding ITC 15 (see Figure 2, A: structure, B2: R _t  = 4.22 min, and C2: mass 358 m z⁻¹ for [M + Na]⁺ and 336 m z⁻¹ for [M + H]⁺) is observed.

Figure 2.

LCMS analysis after incubation of ps GSL _PEG (Azo)–N ₃ with azoreductase AzoR from E. coli and subsequent derivatization with NH₄OH solution. A) Structures of ps GSL _PEG (Azo)–N ₃ , ITC 15 and thiourea 16. B) UV chromatogram at 254 nm of B1) pure ps GSL _PEG (Azo)–N ₃ (500 μM) in TRIS buffer (25 mM, pH 7.4), B2) incubation of ps GSL _PEG (Azo)–N ₃ (500 μM) with AzoR (10 μM), NADH (5 mM), and FMN (20 μM) in TRIS buffer (25 mM, pH 7.4) after 2 h at 37 °C, and B3) A2 and addition of aqueous NH₄OH solution (30 w%, 10 μL) at 23 °C. C) ESI + mass analysis at C1) 4.85 min of B1, C2) 4.22 min of B2, and C3) 2.45 min of B3. D) Control experiments following (B2) or (B3) in absence of D1,D3) AzoR or D2,D4) the cofactors NADH and FMN.

Furthermore, by adding an excess of concentrated aqueous ammonia, ITC 15 was rapidly converted into the corresponding thiourea 16 within minutes (see Figure 2, A: structure, B3: R _t  = 2.33 min, and C3: mass 375 m z⁻¹ for [M + Na]⁺, 353 m z⁻¹ for [M + H]⁺).

Next, we performed several control experiments (see Figure 2D1–4) for the conversion of ps GSL _PEG (Azo)–N ₃ in the absence of the enzyme or cofactors NADH and FMN, with or without subsequent addition of an excess of 30% aqueous ammonia (10 μL≈157 μmol≈6284‐fold excess compared to ps GSL _PEG (Azo)–N ₃ ). Similar to the observed chemical stability for nitroreductase‐responsive psGSLs,^{[ 36 ]} ps GSL _PEG (Azo)–N ₃ was highly stable against strongly basic conditions and showed no decomposition in the presence of an excess of aqueous ammonia over 15 min at 25 °C.

Mechanistically, it is known that reduction of substrates by AzoR proceeds via a ping‐pong bi‐bi mechanism.^{[ 37} , ⁵⁰ , ^{51 ]} This includes two 2‐electron reduction steps from the azobenzene to the corresponding hydrazine derivative and the spontaneous hydrazo bond lysis via resonance stabilization.^{[ 52 ]} Therefore, the respective substrates and the cofactor NADH must alternatingly bind to the active center of AzoR bearing the prosthetic group FMN as illustrated in Figure  3A. For ps GSL _PEG (Azo)–N ₃ , the required reduction step to the hydrazine 17 and the spontaneous hydrazo bond lysis to amine 18 (Figure 3B) seemed to be fast as none of the intermediates could be detected over the course of the reaction.

Figure 3.

A) Illustration of the substrate binding pocket with FMN as prosthetic group on AzoR from E. coli (PDB: 2D5I)^{[ 44 ]} created with SeeSAR version 13.0.1; BioSolveIT GmbH, Sankt Augustin, Germany, 2023, www.biosolveit.de/SeeSAR. B) Proposed mechanism for the reduction of ps GSL _PEG (Azo)–N ₃ catalyzed by AzoR from E. coli via ping‐pong bi‐bi mechanism and conversion to ITC 15.

Next, we wanted to prove the fluorescence turn‐on upon formation of ITC 20 from ps GSL _PEG (Azo)–BODIPY in the presence of AzoR.

Therefore, ps GSL _PEG (Azo)–BODIPY was incubated with AzoR as before and aliquots of the reaction mixture were diluted with TRIS buffer. The fluorescence emission at 511 nm upon excitation of the sample at 495 nm was measured in 10 min steps over a period of 60 min (Figure  4A). After only 10 min a significant fluorescence that increased overtime was observed, indicating that the FRET acceptor unit azobenzene was cleaved and the fluorescent ITC 20 was released (Figure 4B) as identified by LCMS analysis (Figure S5, Supporting Information). In the absence of either AzoR or the cofactors, no fluorescence was observed.

Figure 4.

A) Fluorescence turn‐on of ps GSL _PEG (Azo)‐BODIPY in the presence of AzoR: aliquots of the incubation of ps GSL _PEG (Azo)–BODIPY (500 μM) with AzoR (10 μM), NADH (5 mM) and FMN (20 μM) in TRIS buffer (25 mM, pH 7.4) at 37 °C, were diluted in TRIS buffer and fluorescence emission was measured at t = 0 (absence of AzoR), 10, 20, 30, 40, 50, 60, and 70 min at 510 nm upon excitation at 495 nm. B) Structure of ITC 20 (low resolution mass spectrometry), see Figure S5, Supporting Information).

Furthermore, we used the fluorescence turn‐on to assess the Michaelis‐Menten kinetics for the conversion of ps GSL _PEG (Azo)–BODIPY. K_cat was determined to be 5.8 s⁻¹ while K_M was found to be 29.2 mM (see Table S3, Supporting Information) indicating a comparably slow conversion of the probe by AzoR (k_cat/k_m = 2.01 s⁻¹  M ⁻¹ × 10³) compared to simple azo dyes (k_cat/k_m ≈ 7–100 s⁻¹  M ⁻¹ × 10³).^{[ 52 ]} Although the conversion of sulfonate containing azo dyes by AzoR is known,^{[ 53 ]} these dyes differ significantly due to the absence of the thiohydroximate O‐sulfonate moiety present in ps GSL _PEG (Azo)–BODIPY.

The sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) analysis (Figure  5 ) revealed slightly fluorescent bands at the height of the enzyme when excited at 366 nm indicating a covalent labeling of AzoR (MW: 23819.7 Da, Uniprot P41407 plus His₆‐containing tag) most likely through covalent thiourea linkage of the ITC 20 with lysine residues of AzoR.^{[ 36 ]}

Figure 5.

A) SDS–PAGE gel analysis of the conversion of ps GSL _PEG (Azo)–BODIPY in the presence of azoreductase AzoR from E. coli, left: visible light/Coomassi blue stain, right: UV light (366 nm), 1) molecular weight marker (3 μL, 2 mg mL⁻¹), 2) azoreductase AzoR (10 μL, 2 mg mL⁻¹), 3) 10 μL of ps GSL _PEG (Azo)‐BODIPY (500 μM) with AzoR (10 μL, 2 mg mL⁻¹), NADH (5 mM), and FMN (20 μM) in TRIS buffer (25 mM, pH 7.4) after 2 h at 37 °C. B) SDS–PAGE gel analysis of the labeling of bovine serum albumin (BSA) with ps GSL _PEG (Azo)–BODIPY in the presence of azoreductase AzoR from E. coli, left: visible light/Coomassie blue stain, right: UV light (366 nm), 1) molecular weight marker (3 μL, 2 mg mL⁻¹), 2) azoreductase AzoR (10 μL, 0.5 mg mL⁻¹), 3) BSA (10 μL, 2 mg mL⁻¹), and 4) 10 μL of ps GSL _PEG (Azo)–BODIPY (500 μM) with AzoR (10 μM), NADH (5 mM), and FMN (20 μM) in TRIS buffer (25 mM, pH 7.4) and BSA (2 mg/mL) after 2 h at 37 °C. C) Comparison of deconvoluted full protein mass spectra of AzoR from E. coli before (top) and after incubation with ps GSL _PEG (Azo)–N ₃ , FMN (20 μM), and NADH (5 μM) at 37 °C for 2 h (bottom). The incubation leads to oxidation of AzoR (+16 Da) and to its modification with ITC 15 (+336 Da). Full spectra are shown in Supporting Information.

Additionally, fluorescent small molecules, presumably corresponding to the fluorescent ITC 20 (for structures, see Figure 4,B) or its hydrolysis products, are visible in the flow‐through (Figure 5A). The untreated azoreductase had a deconvoluted neutral molecular mass of 23819.9 Da according to an intact protein electron spray ionization–MS experiment, that matches perfectly with the predicted mass (Figure S6, Supporting Information).

When ps GSL _PEG (Azo)–N ₃ was incubated with AzoR, FMN, and NADH at 37 °C for 2 h and analyzed by high‐resolution mass spectrometry, we observed a partial oxidation of the protein (+ 16 Da) as well as an additional peak series that corresponded to a protein modification by +336 Da. This mass shift reflects the covalent modification with ITC 15 (Figure S7, Supporting Information). If such covalent labeling is inhibiting, the converting enzyme has not been tested.

Beyond the labeling of the converting enzyme AzoR, we could demonstrate that bovine serum albumin (BSA, 69.3 kDa, Uniprot P02769) added to the reaction mixture was labeled efficiently with ps GSL _PEG (Azo)–BODIPY in the presence of AzoR suppressing the labeling of AzoR as shown in Figure 5B.

Next, we investigated whether the observed fluorescence turn‐on for ps GSL _PEG (Azo)–BODIPY in the presence of AzoR would similarly occur in the presence of E. coli (Figure  6 ).

Figure 6.

A) Fluorescence turn‐on of ps GSL _PEG (Azo)‐BODIPY (@100 μM) after incubation in the presence of different E. coli strains at 37 °C. B) Boxplot of initial and final normalized fluorescence intensity after 22 h. Fluorescence intensity is normalized by subtraction of no cell background. Colored wide bands indicate standard deviations. None: no bacteria present, ASKA: BW25113/pCA24N‐azoR (camR),^{[ 60 ]} BW: E. coli K‐12 BW25113, Keio: BW25113 Δ azoR:kanR.^{[ 54 ]}

Therefore, three E. coli strains were employed. E. coli K‐12 BW25113 was chosen as reference strain containing the azoR gene on its chromosome. The Keio mutant, has an in‐frame, single‐gene knock out of azoR in E. coli K‐12 BW25113 where the azoR gene is replaced with a kanamycin resistance cassette.^{[ 54 ]} Finally, we also increased azoR expression in E. coli K‐12 BW25113 by introducing a plasmid containing azoR under an Isopropyl β‐D‐thiogalactoside‐inducible promoter (ASKA strain).^{[ 55 ]}

All three strains were then incubated with ps GSL _PEG (Azo)–BODIPY at 23 °C and fluorescence at 535 nm after excitation at 485 nm was measured every 90 min over the course of 22 h (Figure 6). For both the reference strain E. coli K‐12 BW25113 and the azoR‐knock‐out mutant Keio, only little fluorescence intensity was measured compared to the negative control in absence of bacterial cells. We assume that the similar fluorescence levels for both reflect the usually low basal expression levels of AzoR, which is mostly expressed under various stress conditions.^{[ 45} , ⁴⁶ , ⁴⁷ , ^{48 ]}

Beyond that it needs to be considered that bacteria produce hydrogen sulfide (H₂S) as a metabolic end product, which can act as a redox partner with azo dyes.^{[ 55 ]} In E. coli, the azoR gene is partially dispensable for azo dye depletion due to the redundant ability of H₂S to degrade azo dyes,^{[ 56 ]} which explains the background fluorescence in the Azor‐deletion Keio mutant.

In contrast, for the AKSA mutant with plasmid‐induced AzoR overexpression showed a significant, continuous increase in fluorescence overtime was observed confirming the conversion of ps GSL _PEG (Azo)–BODIPY into its corresponding fluorescent ITC 20 in the presence of these cells.

Although the decolorizing reduction of heavily sulfonated azo dyes by different bacteria including E. coli in whole cell assays have been reported,^{[ 48} , ^{53 ]} we assume that the reduction of the O‐sulfonate bearing probe ps GSL _PEG (Azo)–BODIPY is taking place extracellularly as recent evidence has been found that AzoR is secreted by E. coli during the exponential growth period in cultivation.^{[ 57 ]}

3. Conclusion

We have successfully expanded the concept of psGSLs toward azoreductase‐responsive psGSLs and demonstrated the release of ITCs from ps GSL _PEG (Azo)–N ₃ and ps GSL _PEG (Azo)–BODIPY in the presence of azoreductase AzoR from E. coli. The probes show a high chemical stability, similar to the nitroreductase‐responsive psGSLs reported earlier by us.^{[ 36 ]} The released ITCs show a lysine‐selective modification of proteins,^{[ 36 ]} and the probes were demonstrated to be effective in labelling of BSA. Furthermore, we demonstrated that ps GSL _PEG (Azo)–BODIPY is a turn‐on fluorophore, capable of indicating azoreductase activity in vitro and in cell culture of E. coli. Currently, further investigations to expand the concept of psGSLs toward a platform technology for bio‐responsive protein labeling and ITC‐based covalent inhibitors of proteins are ongoing.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

Contribution are given with CRediT definition according to Brand et al.^{[ 58 ]} Philipp Klahn: conceptualization; Claire. C. Jimidar, Charity S. G. Ganskow, Mervic D. Kagho, Aishi Chakrabarti, Neslihan Beyazit, Lorenz Wiese, Margarita Seeger, Silvana Smits, Stella Nyström, and Ulrike Beutling: methodology; Software: ‐ ; Charity S. G. Ganskow, Ulrike Beutling, Silvana Smits, and Claire. C. Jimidar: validation; Ulrike Beutling: formal analysis; Claire. C. Jimidar, Charity S. G. Ganskow, Mervic D. Kagho, Aishi Chakrabarti, Neslihan Beyazit, Lorenz Wiese, Margarita Seeger, Stella Nyström, Silvana Smits, and Ulrike Beutling: investigation; Philipp Klahn, Mark Brönstrup, Anne Farewell, and Anett Schallmey: resources; Philipp Klahn: data curation; Charity S. G. Ganskow, Claire. C. Jimidar, Aishi Chakrabarti, Mervic D. Kagho, and Philipp Klahn: writing—original draft; all authors: writing—review and editing; Philipp Klahn, Aishi Chakrabarti, Charity S. G. Ganskow, Mervic D. Kagho, Silvana Smits, and Margarita Seeger: visualization; Philipp Klahn, Mark Brönstrup, Anne Farewell, Neslihan Beyazit, Silvana Smits, and Anett Schallmey: supervision; Philipp Klahn: project administration; and Philipp Klahn, Mark Brönstrup, Anne Farewell, and Anett Schallmey: funding acquisition. A. Chakrabarti, Charity S. G. Ganskow, Mervic D. Kagho, Claire. C. Jimidar, and Lorenz Wiese contributed equally to this work.

Supporting information

Supplementary Material

Data Availability Statement

Nuclear magnetic resonance raw data of all synthesized compounds have been deposited to the NMRXiv (https://nmrxiv.org) with data set identifier 10.57992/nmrxiv.p79. Other primary data of this work are available on requests from the authors.