Artificial Fluorescent Glucosinolates (F-GSLs) Are Transported by the Glucosinolate Transporters GTR1/2/3

The glucosinolate transporters 1/2/3 (GTR1/2/3) from the Nitrate and Peptide transporter Family (NPF) play an essential role in the transport, accumulation, and distribution of the specialized plant metabolite glucosinolates. Due to representing both antinutritional and health-promoting compounds, there is increasing interest in characterizing GTRs from various plant species. We generated seven artificial glucosinolates (either aliphatic or benzenic) bearing different fluorophores (Fluorescein, BODIPY, Rhodamine, Dansylamide, and NBD) and investigated the ability of GTR1/2/3 from Arabidopsis thaliana to import the fluorescent glucosinolates (F-GSLs) into oocytes from Xenopus laevis. Five out of the seven F-GSLs synthesized were imported by at least one of the GTRs. GTR1 and GTR2 were able to import three F-GSLs actively above external concentration, while GTR3 imported only one actively. Competition assays indicate that the F-GSLs are transported by the same mechanism as non-tagged natural glucosinolates. The GTR-mediated F-GSL uptake is detected via a rapid and sensitive assay only requiring simple fluorescence measurements on a standard plate reader. This is highly useful in investigations of glucosinolate transport function and provides a critical prerequisite for elucidating the relationship between structure and function through high-throughput screening of GTR mutant libraries. The F-GSL themselves may also be suitable for future studies on glucosinolate transport in vivo.


Introduction
Glucosinolates (GSL) are amino acid-derived, sulfur-and nitrogen-containing thioglucosides with more than 130 identified structures in nature [1]. GSL are found mainly in the order Brassicales [2], wherein they function as important defense compounds against pathogens and herbivores [3]. From an agricultural perspective, there is a desire to control the accumulation patterns of glucosinolate in plants. For example, in both established and emerging oilseed crops such as Brassica napus (rape), Brassica juncea (mustard), and Camelina Sativa, the high accumulation of antinutritional glucosinolates in seeds must be eliminated to enable the usage of the otherwise protein-rich seed-meal in animal feed [4][5][6].
GSLs are translocated to the seeds, which are devoid of biosynthesis capability. Accordingly, the identification of the glucosinolate transporters (GTRs), GTR1/AtNPF2.10, GTR2/AtNPF2.11 GTR3/AtNPF2.9 in Arabidopsis thaliana [7,8] paved the way for targeting GTRs as a novel strategy for modulating the levels of GSL in Brassica crops in a

Synthesis of Seven Artificial Fluorescent Glucosinolates (F-GSLs)
Different strategies were applied to generate the aliphatic and benzylic fluorescent glucosinolates (F-GSLs). On the first hand, F-GSLs were generated based on simple copper(I)-mediated azide-alkyne click chemistry (CuAAC) linking different alkyne-bearing fluorophores to propyl GSL-A-N3, an azide containing artificial GSL [23]. Using this strategy, GSL-A-BODIPY, GSL-A-Rhodamine, and GSL-A-Dansylamide were synthesized as outlined in Scheme 1, bearing a fluorescently labeled sidechain mimicking an aliphatic ornithine residue. In a different method, the fluorescein fluorophore was condensed on the pentyl GSL-A-N3, producing the GSL-A-Fluorescein. Access to the benzylic GSLs followed similar procedures. We generated GSL-B-Fluorescein, GSL-B-DNS, and GSL-B-NBD from GSL-B-N3 [24] by sequence of Staudinger reduction in the presence of triphenylphospine and subsequent reaction of the released GSL-B-NH2 with dansylchloride, NBD chloride, or FITC, as outlined in Scheme 2.

Synthesis of Seven Artificial Fluorescent Glucosinolates (F-GSLs)
Different strategies were applied to generate the aliphatic and benzylic fluorescent glucosinolates (F-GSLs). On the first hand, F-GSLs were generated based on simple copper(I)-mediated azide-alkyne click chemistry (CuAAC) linking different alkyne-bearing fluorophores to propyl GSL-A-N 3 , an azide containing artificial GSL [23]. Using this strategy, GSL-A-BODIPY, GSL-A-Rhodamine, and GSL-A-Dansylamide were synthesized as outlined in Scheme 1, bearing a fluorescently labeled sidechain mimicking an aliphatic ornithine residue. In a different method, the fluorescein fluorophore was condensed on the pentyl GSL-A-N 3 , producing the GSL-A-Fluorescein. Access to the benzylic GSLs followed similar procedures. We generated GSL-B-Fluorescein, GSL-B-DNS, and GSL-B-NBD from GSL-B-N 3 [24] by sequence of Staudinger reduction in the presence of triphenylphospine and subsequent reaction of the released GSL-B-NH 2 with dansylchloride, NBD chloride, or FITC, as outlined in Scheme 2. In all the F-GSLs synthesized in this study, the fluorophores are attached to the side chains of core structure mimicking two of the more important types of GSLs, namely the L-methionine and L-tyrosine-derived ones with aliphatic and benzylic side chains, respectively. To some extent, the natural GSLs such as Sinalbin (4-Hydroxybenzyl GSL),

Transport of Fluorescent Glucosinolates Depends on the Fluorophore Attached
Although GTRs are promiscuous towards the side chain [7,25], it is not given that they would accept artificial GSLs to which an entire fluorophore molecule had been added to the side chain. To investigate whether any of the seven F-GSL are transported by the GTRs, we performed transport assays, wherein GTR1, GTR2, and GTR3 expressed in Xenopus laevis oocytes were exposed to 50 µM external concentration of each individual F-GSL at pH 5 for 1 h. To make sure the F-GSL are stable and remain intact during the assays, we first conducted three independent experiments where oocyte homogenates were analyzed by LCMS (Figure 2).
For GSL-A-Rhodamine, no uptake was detected for any GTR. For GSL-A-BODIPY, a relatively high background level was detected in mock oocytes. In comparison, for the other five F-GSLs, uptake was detected by at least one of the three GTRs, and they were either not detected in mock oocytes or were detected at very low levels (GSL-A-Fluorescein and GSL-B-Fluorescein) ( Figure 2). These five F-GSL were deemed suitable for transport assays and were included in the experiment where uptake was detected by fluorescence measurements on oocyte homogenates (see below). GTR1/2/3 are all characterized as active GSL transporters, which means that they can accumulate their GSL substrates against a concentration gradient [7,8]. Analyses of the LCMS-based transport assays show that GTR1 and GTR2 could transport GSL-B-NBD, GSL-B-DNS, and GSL-A-Dansylamide against the concentration gradient. GTR1 and GTR2 appear both to import GSL-B-NBD and GSL-B-DNS to similar levels, whereas GTR1 was several folds more efficient at importing GSL-A-Dansylamide compared to GTR2 ( Figure 2). In comparison, GTR2 was more efficient at importing GSL-B-Fluorescein compared to GTR1; however, it did not accumulate GSL-B-Fluorescein against the concentration gradient, making GSL-B-Fluorescein inferior to the GSL-B-NBD, GSL-B-DNS and GSL-A-Dansylamide substrates. Of these four F-GSL, GTR3 was only able to import GSL-B-NBD. The GTR3-mediated import of GSL-B-NBD was active, albeit at three-fold lower levels compared to GTR1 and GTR2. Lastly, we saw a reciprocal uptake pattern for GSL-A-Fluorescein, which was imported by GTR3 and not by GTR1 or GTR2 ( Figure 2).
After demonstrating GTR-mediated uptake of the five F-GSL, namely GSL-A-Fluorescein, GSL-B-Fluorescein, GSL-B-NBD, GSL-B-DNS, and GSL-A-Dansylamide, we investigated whether their uptake could be detected using fluorescence measurements of oocyte homogenates on a fluorescence plate reader. Similar to the LCMS-based assays, GTR-expressing oocytes were exposed to a 50 µM concentration of each of the five F-GSL at pH 5 for 1 h. As mentioned above, the same experiment was conducted three different times using different oocyte batches. With two exceptions, the results obtained by the fluorescence measurements matched the results obtained by LCMS-based detection ( Figure 3). The two differences were both seen in assays using fluorescein-linked GSL. In contrast, to the first experiment, GSL-B-Fluorescein levels were now only significantly different from mock in GTR2-expressing oocytes (Figure 3), whereas significant uptake was detected in both GTR1-and GTR2-expressing oocytes in the LCMS-based experiment ( Figure 2). For GSL-A-Fluorescein, significant uptake was now detected in both GTR3 and GTR2-expressing oocytes, whereas previously, we only saw an uptake in GTR3-expressing oocytes. For both of these aliphatic and benzylic GSLs bearing fluorescein, the absolute level of uptake was very low and was bordering detection limits for the GSL-A-Fluorescein. Problematically, it was seen that homogenates from oocytes that had not been exposed to any F-GSL gave a strong fluorescent signal when excited by the wavelength used to excite fluorescein. This signal was subtracted from the measurements on samples wherein GTR-expressing oocytes had been exposed to GSLs bearing fluorescein. Although uptake was detected for both aliphatic and benzylic GSL-Fluoresceins, the subtraction of a high background signal introduces an element of uncertainty that can explain the slight differences in uptake between the two experiments. Together, these data indicate that the fluorescein-linked GSLs are not as useful as substrates.
used to excite fluorescein. This signal was subtracted from the measurements on samples wherein GTR-expressing oocytes had been exposed to GSLs bearing fluorescein. Although uptake was detected for both aliphatic and benzylic GSL-Fluoresceins, the subtraction of a high background signal introduces an element of uncertainty that can explain the slight differences in uptake between the two experiments. Together, these data indicate that the fluorescein-linked GSLs are not as useful as substrates. Figure 2. Uptake of the 7 fluorescent glucosinolates, analyzed by LCMS. Xenopus laevis oocytes expressing GTR1, GTR2, GTR3, or water-injected Mock were used in import assay. Numbers above X-axis refer to n number of single oocytes analyzed over 3 batches; the points show the signal from individual samples, and different shapes and colors represent different oocyte batches. Assay conditions: 1 h assay, pH 5, 50 µM intended concentration of the F-GSL (Dotted horizontal lines represent the average measured concentration of F-GSL in the external media in the 3 assays, color Figure 2. Uptake of the 7 fluorescent glucosinolates, analyzed by LCMS. Xenopus laevis oocytes expressing GTR1, GTR2, GTR3, or water-injected Mock were used in import assay. Numbers above X-axis refer to n number of single oocytes analyzed over 3 batches; the points show the signal from individual samples, and different shapes and colors represent different oocyte batches. Assay conditions: 1 h assay, pH 5, 50 µM intended concentration of the F-GSL (Dotted horizontal lines represent the average measured concentration of F-GSL in the external media in the 3 assays, color correlates to different batches). GSL-B-Fluorescein lacks the assay concentration line since the concentration is 6-10 times higher than the uptake (i.e., not visible in the plot). Statistics: Letters indicate statistical significance comparing the transporters (one-way ANOVA (model: GSL~ID), Tukey post hoc test, p < 0.05). ANOVA table in Supplementary.  Figure 2, analyzed on plate reader. Xenopus laevis oocytes expressing GTR1, GTR2, GTR3, or water-injected Mock were used. Numbers above X-axis refer to n number of single oocytes analyzed over 3 batches; the points show the signal from individual samples, and different shapes and colors represent different oocyte batches. Assay conditions: 1 h assay, pH 5, 50 µM intended external concentration of the glucosinolates (Dotted horizontal lines represent the average measured concentration of F-GSL in the external media in the 3 assays, color correlates to different batches). GSL-B-Fluorescein lacks the media line since the media are 6-10 times higher than the uptake (i.e., not visible in the plot). Statistics: Letters indicate statistical significance comparing the transporters (one-way ANOVA (model: GSL~ID), Tukey post hoc test, P < 0.05). ANOVA table in Supplementary.

Natural Glucosinolate Glucoerucin Is in Competition with the Fluorescent Glucosinolates
Lastly, we sought to investigate whether the artificial F-GSLs are likely transported via the same mechanism as natural GSLs. Ideally, an estimation of Km of each GTR towards each of the three F-GSLs is needed. However, the limited amount of F-GSL available did not permit such investigations using electrophysiological measurements, which require high amounts of substrate. Additionally, although possible, Km estimation using LCMS or fluorescence-based measurements remains imprecise as both represent cumulative assays where one can only approximate initial transport rates. As an alternative, we challenged the uptake of each of the three F-GSLs with an equimolar concentration of Glucoerucin (4-methylthiobutyl glucosinolate, 4MTB, Figure 1). This is a natural GSL towards which GTR1 and GTR2 exhibit an approximate Km of ~25 µM in oocytes [7], whereas GTR3 only transports 4MTB passively and with low affinity [8].

Natural Glucosinolate Glucoerucin Is in Competition with the Fluorescent Glucosinolates
Lastly, we sought to investigate whether the artificial F-GSLs are likely transported via the same mechanism as natural GSLs. Ideally, an estimation of Km of each GTR towards each of the three F-GSLs is needed. However, the limited amount of F-GSL available did not permit such investigations using electrophysiological measurements, which require high amounts of substrate. Additionally, although possible, Km estimation using LCMS or fluorescence-based measurements remains imprecise as both represent cumulative assays where one can only approximate initial transport rates. As an alternative, we challenged the uptake of each of the three F-GSLs with an equimolar concentration of Glucoerucin (4-methylthiobutyl glucosinolate, 4MTB, Figure 1). This is a natural GSL towards which GTR1 and GTR2 exhibit an approximate Km of~25 µM in oocytes [7], whereas GTR3 only transports 4MTB passively and with low affinity [8].
Similar to the previous experiments, GTR-expressing oocytes were incubated at pH 5 for 1 h. The external buffer now contained either 50 µM 4MTB alone, 50 µM of either of the three F-GSLs alone, or an intended equimolar concentration of 4MTB and each of the respective F-GSLs individually (i.e., 4MTB + GSL-B-NBD, 4MTB + GSL-B-DNS or 4MTB + GSL-A-Dansylamide).
The homogenate of the oocytes was divided into two parts. One part was analyzed on the plate reader, and the other half was analyzed by LCMS. The data interpretations presented in the following are corroborated by both detection methods.
As expected, when exposed to 4MTB alone, GTR1 and GTR2 expressing oocytes imported 4MTB actively and to similar levels. However, when 4MTB was mixed with either of the three F-GSL, the level of 4MTB uptake was reduced by approximately 50%. GTR1 and GTR2 also imported each individual F-GSL actively, and this uptake level was reduced by approx. 60% when mixed with an equimolar concentration of 4MTB (Figure 4). Similar to the previous experiments, GTR-expressing oocytes were incubated at pH 5 for 1 h. The external buffer now contained either 50 µM 4MTB alone, 50 µM of either of the three F-GSLs alone, or an intended equimolar concentration of 4MTB and each of the respective F-GSLs individually (i.e., 4MTB + GSL-B-NBD, 4MTB + GSL-B-DNS or 4MTB + GSL-A-Dansylamide).
The homogenate of the oocytes was divided into two parts. One part was analyzed on the plate reader, and the other half was analyzed by LCMS. The data interpretations presented in the following are corroborated by both detection methods.
As expected, when exposed to 4MTB alone, GTR1 and GTR2 expressing oocytes imported 4MTB actively and to similar levels. However, when 4MTB was mixed with either of the three F-GSL, the level of 4MTB uptake was reduced by approximately 50%. GTR1 and GTR2 also imported each individual F-GSL actively, and this uptake level was reduced by approx. 60% when mixed with an equimolar concentration of 4MTB (Figure 4).  . Competition assays. Xenopus laevis oocytes expressing GTR1, GTR2, GTR3, or waterinjected Mock were used. Numbers above X-axis refer to n number of single oocytes analyzed over 1-2 batches. Upper: 4MTB uptake analyzed by LCMS; 1 h assay, pH 5, 50 µM intended concentration of the 4MTB alone or in combination with one of the F-GSLs also at 50 µM. (Dotted horizontal lines represent the average measured concentration of 4MTB in the external media averaged over each assay. In order of assay: 100 pmol/µL, 161 pmol/µL, 86 pmol/µL, and 87 pmol/µL). Lower: F-GSL uptake analyzed on plate reader; 1 h assay, pH 5, 50 µM intended concentration of the F-GSL alone or in combination with 4MTB at 50 µM. (Dotted horizontal lines represent the average measured concentration of GSL in the external media averaged over each assay.) Statistics: one-way ANOVA within gene groups (model: GSL~Assay), using Dunnett post hoc test (single assay as reference, against the equimolar assays), p-value: *** < 0.001, ** < 0.01, and * <0.05.
These results indicate that the natural and F-GSL are likely competing for the same substrate binding sites within the GTRs and that the fluorophores attached do not render the F-GSL as inferior substrates. For GTR3, we only analyzed the GSL-B-NBD, since GTR3 does not transport the other F-GSL (Figures 2 and 3). As seen previously, GTR3 only imported 4MTB up to-but not above the concentration in the external medium. This uptake of 4MTB was significantly lowered when GSL-B-NBD was added in equimolar concentrations and vice versa ( Figure 4).

Discussion
Since their discovery, GTRs have been functionally expressed and characterized in several heterologous host organisms via different assays. Most studies have used Xenopus oocytes as expression hosts and have characterized transport activity directly by measuring imported glucosinolate levels via LCMS or indirectly by measuring the currents elicited by TEVC electrophysiology. LCMS has also been used to measure GTR-mediated glucosinolate import in other heterologous hosts, including yeast [25], cotton cells [26], and insect cells (unpublished data) [27]. In one example, glucosinolate import was detected by monitoring the influx of radiolabeled glucosinolates that were generated by feeding radiolabeled tyrosine to CYP79A1 overexpressing Arabidopsis plants [7,28].
A clear advantage of the fluorescence-based GSL uptake assays presented here is that they are rapid and that detection can be accomplished by fluorescence microscopy, fluorescence plate reading, or flow cytometry. Additionally, single-cell fluorescence-based transport assays allow GTR-mutant library screening using fluorescence-activated cell sorting (FACS). For these purposes, the azide-glucosinolate precursors are produced in abundance and can be used to generate more of the three best-performing F-GSL described in this study GSL-B-NBD, GSL-B-DNS, and GSL-A-Dansylamide. In addition, they can be used to generate other F-GSLs with different fluorophores.
Oocyte homogenates did not emit fluorescence at the emission wavelengths of Dansylamide and NBD (Excitation at 335 nm and 465 nm, Emission at 518 nm and 560 nm, respectively), which resulted in very low background signal in mock oocytes. In comparison, the homogenates emitted a strong signal for the settings for fluorescein (Excitation at 492 nm and Emission at 524 nm), which renders the Fluorescein-linked GSLs less suitable for transport assays using Xenopus oocytes. Thus, it is advisable to investigate the spectral properties of heterologous expression hosts prior to selecting which F-GSL to use.
So far, GTR1 and GTR2 have been shown to transport glucosinolates with near similar preference irrespective of the structure of the glucosinolate side chain [7,25]. In comparison, GTR3 displays a strong preference for tryptophan-derived glucosinolates [8]. The ability of all GTRs to transport glucosinolates with fluorophores attached to the side chain shows that the GTR promiscuity towards the side chain extends beyond natural glucosinolates.
Rhodamine is by far the largest of the five tested fluorophores. Hence, the lack of transport by any of the GTRs likely indicates that we exceeded the structural constraints of the substrate binding pocket. However, we cannot exclude that the transporters may exhibit different kinetics towards the different F-GSLs. Therefore, it is possible that the GSL-A-Rhodamine may be transported by the GTRs if applied at higher concentrations. BODIPY is a lipid-binding fluorophore, which is typically used to quantify levels of neutral lipids in biological samples [29,30]. The high background levels seen in the transport assays likely reflect the binding of the fluorophore to the oocyte membrane, rendering GSL-A-BODIPY unsuitable for transport assays. To explore whether GSL linked to BODIPY can be used as substrates by GTRs, it would be necessary to use a modified non-lipid binding BODIPY as a fluorophore [31].
We did not measure the amount of GTR protein expressed in this study. Differences in GTR protein expression could therefore, in principle, explain the variation in uptake of different F-GSLs. However, we have shown in earlier studies that the oocytes generally express the different GTRs to very similar levels [8]. Accordingly, it is more likely that the different uptake levels reflect variations in substrate preference.
Kinetic values such as Km were not determined in this study. This is mainly due to a lack of sufficient amounts of F-GSL to carry out these investigations. In addition, we noticed that for the Fluorescein linked F-GSL, increasing external concentrations (beyond 100 µM) increased background levels in mock oocytes to the extent that masked uptake by GTR. However, the ability of GTR1 and GTR2 to accumulate the Dansylamide and NBD-linked F-GSL' actively (above external media level) combined with the reduction of 4MTB uptake in equimolar competition assays indicates that these F-GSL are transported by the same mechanism as natural GSL and with similar affinity. This is further supported by the lower uptake by GTR3 of these F-GSL, which mimics the inherent difference in substrate preference of GTR1/GTR2 versus that of GTR3 [8]. Accordingly, the F-GSL provides the prerequisites for conducting high-throughput screening of mutant libraries in studies aimed at unraveling the structural basis underlying substrate specificity in the GTRs. In this context, it is noteworthy that GTR1 and GTR2 have so far been shown to have almost completely overlapping substrate preference [7,8,25]. Here, the similar uptake of GSL-B-DNS and GSL-B-NBD but varying uptake of GSL-A-Dansylamide by GTR1 and GTR2, respectively, indicates that they harbor distinct substrate preferences. Whether this also extends to natural GSL remains to be investigated.
In addition, owing to their presence in Arabidopsis thaliana, the GTRs have emerged as a model system for studying the transport of specialized metabolites in planta. In this context, in vivo feeding of F-GSL could be utilized to identify apoplastic barriers at cellular resolution. Similar investigations could be conducted in specialized insects such as flea beetles that have evolved distinct glucosinolate transporters that enable the beetle to take up GSL from host plants and use it for their own defense [27]. However, it is not known whether the F-GSL are accepted as substrates by the flea beetle glucosinolate transporters that belong to a different transporter family. puriss., extra dry (water ≤ 0.005%)). Moisture-sensitive reactions were performed under an argon atmosphere in dried glassware. Dry dichloromethane, diethyl ether, toluene, and tetrahydrofuran for moisture-sensitive reactions have been taken from an MB-SPS-800 (MBraun) solvent purifications system and stored under argon. All solvents used for workup and purification were of HPLC grade. Reactions were monitored by TLC, LCMS, or NMR. The solution of compounds in organic solvents was concentrated using rotary evaporators at a water bath temperature of max. 30 • C. Solvent residues were removed in a high vacuum at a pressure of appr. 10 −2 mbar. Unless otherwise noted, solvents were degassed either by a continuous Argon flow over a minimum of 15 min or using the Freeze-Pump-Thaw technique [32]. Flash chromatography [33] was conducted using appropriate glass columns filled with silicagel (Merck Millipore, Geduran ® Si60, 1.11567.9025, 40-63 µm) or using the Biotage Select ® chromatography system with a DAD detector and cartridges packed with silicagel (Merck Millipore, Geduran ® Si60, 1.11567.9025, 40-63 µm) using a Cartridger ® C-670 from the company BüchiLabortechnik AG, Flawil, Switzerland. Preparative reversed-phase high-pressure liquid chromatography (prep. HPLC RP) was performed on either a Hypersil GOLD C18 RP-column (Part No. 25005-259270), 5 µm, 250 mm × 21.2 mm (10 mL/min) or a Hypersil GOLD C18 RP-column (Part No. 25005-259070A), 5 µm, 250 mm × 10.0 mm (5 mL/min) each equipped with a guard column of the same material using a Thermo Fisher Scientific (Waltham, MA, USA) Dionex Ultimate 3000 HPLC system. Eluents, gradients, and additives are given in parentheses. As eluents, HPLC grade acetonitrile and water (VWR Chemicals, Darmstadt, Germany, HPLC grade) with or without 0.1% of TFA (Carl Roth" Karlsruhe, Germany, 6957. NMR spectra were recorded on Bruker AV-300, AVIII400, and AVIIIHD500 with a cryoprobe system at 293.15 K. 1 H NMR spectra were recorded at 300 MHz, 400 MHz, and 500 MHz. 13 C NMR spectra were recorded at 76 MHz, 100 MHz, and 126 MHz. Chemical shifts are reported in ppm relative to the solvent signal. Multiplicity is indicated as follows: s (singlet); bs (broad singlet); d (doublet); t (triplet); q (quartet); m (multiplet); dd (doublet of doublets), etc. For the processing of the raw data, the software MestReNova (Version 9.0.1-13254) from MestreLab Research S.L. was utilized. IR spectra were recorded on a Bruker Tensor 27 IR spectrometer with the ATR technique. Only the wave numbers of observed absorption peaks are given. Low-resolution mass spectrometry (LRMS) data were recorded using an LC-MS system consisting of an Accela HPLC (Thermo Scientific) equipped with an Accela photodiode array (PDA) Detector, Accela autosampler, and Accela 1250 pump which was coupled to an LTQ XL mass spectrometer (Thermo Scientific) for HPLC/HESI-MS analyses. Heated electrospray ionization was used with an enhanced scan range of 120 to 2000 amu. Gradient HPLC solvent programs consisted of LCMSgrade H 2 O, CH 3 CN, and 2% formic acid in H 2 O. An Agilent Zorbax Eclipse Plus C18 (3.5 µm, 2.1 × 150 mm) column was used, which was kept at 30 • C. The PDA detector was set to a scanning range from 190 to 600 nm with 1 nm wavelength steps. High-resolution mass spectrometry (HRMS) data were recorded on a Finnigan MAT 95 (EI, 70 eV) mass spectrometer and a Finnigan MAT 95 XL (ESI) mass spectrometer. UV-Vis spectroscopy data were recorded on a Cary 100 Bio (Varian). Fluorescence Emission Spectroscopy data were recorded on a Cary Eclipse (Varian).

Synthesis of the GSL-A-N 3 and GSL-A-BODIPY
The synthesis of GSL-A-BODIPY and GSL-N 3 was performed as outlined in Scheme 3. The synthesis of GSL-A-BODIPY and GSL-N3 was performed as outlined in Scheme 3.

Synthesis of GSL-A-Dansylamide
The synthesis of GSL-A-Dansylamide was performed as outlined in Scheme 4  The synthesis of GSL-A-BODIPY and GSL-N3 was performed as outlined in Scheme 3.

Synthesis of GSL-A-Dansylamide
The synthesis of GSL-A-Dansylamide was performed as outlined in Scheme 4

Synthesis of GSL-A-Rhodamine
The synthesis of GSL-A-Rhodamine was performed as outlined in Scheme 5.

Synthesis of GSL-A-Rhodamine
The synthesis of GSL-A-Rhodamine was performed as outlined in Scheme 5. CH2Cl2, λEx = 415 nm) λEm = 505 nm. The analytical data were in good accordance with prior published ones [34]. GSL-A-Dansylamide.

Synthesis of GSL-A-Rhodamine
The synthesis of GSL-A-Rhodamine was performed as outlined in Scheme 5.  CH2Cl2, λEx = 415 nm) λEm = 505 nm. The analytical data were in good accordance with prior published ones [34]. GSL-A-Dansylamide.

General Methods for GLS-A-Fluorescein, GSL-B-DNS, GSL-B-NBD and GSL-B-Fluorescein
Commercial solvents and reagents were purchased from SigmaAldrich, Acros, Alfa-Aesar, TCI, Carbosynth, and Fluorochem suppliers and were used without further purification unless otherwise stated. Anhydrous solvents were dried by standard methods: DCM was distilled over P 2 O 5 , THF and acetonitrile were purified with a dry station GT S100 immediately prior to use; dried methanol from ACROS ORGANICS, N,N-dimethylformamide and dioxane were dried over molecular sieves; pyridine and triethylamine were dried over potassium hydroxide. Molecular sieves were activated by heating overnight in an Erscem oven at 500 • C. For the anhydrous reactions, all glassware was dried in an oven overnight (100 • C), then removed and cooled down to rt under argon flow. Argon flow was dried with a solution of conc. H 2 SO 4 solution and CaCl 2 . Flash silica column chromatography was performed on silica gel 60 N (spherical neutral, 40-63 µm, Merck, Darmstadt, Germany) and column chromatography on the C-18 reverse phase was performed using a Reveleris ® flash chromatography system. The reactions were monitored by thin layer chromatography (TLC) on silica gel 60F254 precoated aluminum plates. Compounds were visualized under UV light (λ = 254 nm or λ = 365 nm) and by charring with a 10% H 2 SO 4 ethanolic solution. Infrared spectra (ATR) were recorded on a PerkinElmer PARAGON 1000 PC instrument, and values were reported in cm −1 . Mass Spectra were performed on a PerkinElmer Sciex API 300 mass spectrometer (low resolution) and a Bruker MaXis Q-Tof (High resolution) from the "Federation de Recherche" ICOA/CBM FR2708 platform in the electrospray ionization (ESI) mode. Optical rotation was measured at 20 • C into sodium light using a Jasco P-2000 polarimeter with a quartz tank with a path length of 1 cm; values are given in deg.dm −1 .g −1 mL −1 with concentrations reported in g/100 mL. 1 H NMR and 13 C NMR were recorded with Bruker Avance II 250 MHz or an Avance III HD Nanobay 400 MHz spectrometer. CD 3 OD, D 2 O, and mainly CDCl 3 , with tetramethylsilane as an internal reference, were used as deuterated solvents. Acetone was added to D 2 O NMR samples as an internal reference for carbon NMR. Assignments were based on DEPT 135 sequence, homo-and heteronuclear 2D correlations. Chemical shifts were reported in parts per million (ppm). Coupling constants (J) are reported, expressed in Hertz (Hz), and rounded to the nearest 0.5 Hz; splitting patterns are designated as b (broad), s (singlet), d (doublet), dd (doublet of doublet), ddd (doublet of doublet of doublet), t (triplet), dt (doublet of triplet), q (quartet), or m (multiplet). To simplify, the NMR attribution protecting group was abbreviated: acetyl group (OAc). The following solvents were abbreviated: DCM (dichloromethane), DMF (N,N-dimethylformamide), EA (ethyl acetate), MeOH (methanol), and PE (petroleum ether).

Synthesis of GSL-A-Fluorescein
Synthesis of 6-azidohexanal oxime (Scheme 6, 2). Sodium azide (18.5 g, 0.285 mol, 2.7 eq) was added to a suspension of 6-chloro-hexan-1-ol 1 (14.4 g, 0.105 mol, 1 eq) in water (116 mL), and the reaction mixture was heated at reflux for 19 h. It was then cooled down to rt and extracted 3 times with ethyl acetate. The combined organic phases were dried over MgSO 4 , filtered and evaporated to give the crude azide product as a colorless oil.
Trichloroisocyanuric acid (TCCA) (24.7 g, 0.106 mol, 1.05 eq) was added to a solution of crude 6-azido-1-hexanol in anhydrous DCM (160 mL) at 0 • C. A solution of 2,2,6,6-Tetramethylpiperidine 1-oxyl (TEMPO) (158 mg, 1.01 mmol, 0.01 eq) in 3 mL of DCM was then added dropwise. The reaction mixture was stirred at 4 • C during 10 min, then at rt for 30 min, then filtered through a pad of Celite. The filtrate was washed once with an aqueous saturated solution of sodium carbonate, once with water, once with a 1 M HCl solution, and then once with brine. The organic phase was finally dried over MgSO 4 , filtered, and the solvent was evaporated under reduced pressure. The crude product was then purified by silica gel column chromatography (PE/EA: 100/0 to 0/100) to yield the pure aldehyde (4.4 g, 29%) as a colorless oil. Sodium azide (18.5 g, 0.285 mol, 2.7 eq) was added to a suspension of 6-chloro-hexan-1-ol 1 (14.4 g, 0.105 mol, 1 eq) in water (116 mL), and the reaction mixture was heated at reflux for 19 h. It was then cooled down to rt and extracted 3 times with ethyl acetate. The combined organic phases were dried over MgSO4, filtered and evaporated to give the crude azide product as a colorless oil.
Trichloroisocyanuric acid (TCCA) (24.7 g, 0.106 mol, 1.05 eq) was added to a solution of crude 6-azido-1-hexanol in anhydrous DCM (160 mL) at 0 °C. A solution of 2,2,6,6-Tetramethylpiperidine 1-oxyl (TEMPO) (158 mg, 1.01 mmol, 0.01 eq) in 3 mL of DCM was then added dropwise. The reaction mixture was stirred at 4 °C during 10 min, then at rt for 30 min, then filtered through a pad of Celite. The filtrate was washed once with an aqueous saturated solution of sodium carbonate, once with water, once with a 1 M HCl solution, and then once with brine. The organic phase was finally dried over MgSO4, filtered, and the solvent was evaporated under reduced pressure. The crude product was then purified by silica gel column chromatography (PE/EA: 100/0 to 0/100) to yield the pure aldehyde (4.4 g, 29%) as a colorless oil.
6-azidohexanal (4.4 g, 31.2 mmol, 1 eq) was dissolved in a mixture of H2O/MeOH 3/7 (42 mL), then hydroxylamine hydrochloride (2.7 g, 38.9 mmol, 1.25 eq) and potassium carbonate (2.15 g, 3.8 mmol, 0.50 eq) were added. The reaction mixture was stirred at room temperature for 4 h; then, the solvent was evaporated under reduced pressure. The crude residue was then taken up with ethyl acetate and washed twice with water and then once with brine; after drying over MgSO4, the solvent was evaporated under reduced pressure to give the desired 6-azidohexanal oxime 2 as a 55/45 mixture of E/Z isomers, (4.4 g, 90%) as a yellow solid. 6-azidohexanal (4.4 g, 31.2 mmol, 1 eq) was dissolved in a mixture of H 2 O/MeOH 3/7 (42 mL), then hydroxylamine hydrochloride (2.7 g, 38.9 mmol, 1.25 eq) and potassium carbonate (2.15 g, 3.8 mmol, 0.50 eq) were added. The reaction mixture was stirred at room temperature for 4 h; then, the solvent was evaporated under reduced pressure. The crude residue was then taken up with ethyl acetate and washed twice with water and then once with brine; after drying over MgSO 4 , the solvent was evaporated under reduced pressure to give the desired 6-azidohexanal oxime 2 as a 55/45 mixture of E/Z isomers, ( (Z)-S-(2,3,4,6-Tetra-O-acetyl-β-D-glucopyranosyl)-(5-azidopentano)thiohydroximate (Scheme 6, 3) A sodium hypochlorite solution (12.5% active chlorine) (20 mL, 36.6 mmol, 3 eq) was added to a vigorously stirred solution of crude 6-azidohexanal oxime 2 (4.4 g, 28.2 mmol, 1 eq) in anhydrous DCM (140 mL), the color of the solution changed from yellow to blue before returning to yellow. The solution was then stirred for 20 min at room temperature. The organic phase was separated from the aqueous one and slowly added to a solution of 2,3,4,6-tetra-O-acetyl-1-thio-β-D-glucopyranose (5.13 g, 14.1 mmol, 0.5 eq) in 140 mL of anhydrous DCM at −10 • C under argon atmosphere, then anhydrous triethylamine (11.8 mL, 84.5 mmol, 3 eq) was added dropwise, and the solution was allowed to warm up to room temperature. After stirring for 2 h at rt, the reaction was quenched by the addition of water, and the aqueous phase was extracted twice with DCM. The combined organic layers were then washed twice with a 0.5 M aqueous HCl solution, dried over MgSO 4 , filtered, and the solvent was evaporated under reduced pressure. The crude product was purified by silica gel column chromatography (EP/EA 100/0 to 0/100) to give the thiohydroximate 3 as a yellow solid ( The thiohydroximate 3 (3.5 g, 6.75 mmol, 1 eq) was dissolved in anhydrous DCM (84 mL), sulfur trioxide-pyridine complex (5.4 g, 33.7 mmol, 5 eq) was added, and the suspension was heated at reflux for 24 h. The reaction was then cooled down to 0 • C and quenched by the addition of a 0.5 M aqueous KHCO 3 solution (6.8 g, 67.5 mmol, 10 eq) and stirred for 45 min at room temperature. The solvent was then evaporated under reduced pressure, and the residue was purified using silica gel column chromatography (ethyl acetate/methanol: 9/1) to give the sulfated compound 4 as a resin (3.6 g g, 84%). Potassium methoxide (132 mg, 1.88 mmol, 0.4 eq) was added to a solution of the acetylated compound 4 (3 g, 4.7 mmol, 1 eq) in anhydrous methanol (60 mL). The reaction mixture was stirred at room temperature for 6h then the solvent was evaporated under reduced pressure. The crude product was purified using Reveleris ® column chromatography on C-18 reverse phase (H 2 O/MeOH: 100/0 to 0/100) to give the glucosinolate 5 as a white resin (1.9 g, 86%).

Data Analysis
All data analyses were performed using Rstudio. Concentrations from LCMS samples were calculated based on internal standard (Sinigrin, Prop-2-enyl GSL), and the response factor was generated based on standard curves. Concentrations from the plate reader were calculated based on standard curves slope-no fluorescent internal standard was used, hence no response factor. Outliers were removed automatically based on generated function in R that removes values outside of ±1.5*Inter quantile range. Statistical analysis and post hoc tests are stated in figure legends. All were performed using ANOVA function aov() in R (ANOVA tables in Supplementary), all but Figure 4 followed by TUKEY post hoc test. For Figure 4, the Dunnett Post hoc test was used.
LCMS data: Data from LCMS were calculated as described in [36], with a modification for the dilution factor. Briefly, the response factor for oocytes was determined by standard curves for the compound investigated and the internal standard: response factor (f) = slope internal standard/slope compound investigated. Concentration in samples was calculated: (Area of peak for compound/Area of peak for internal standard) * f * final internal standard concentration * dilution factor. Dilution factor: 156.25, final internal standard concentration: 0.5 µM.
Plate reader data: Concentration in samples was calculated: (Area of peak for compound/Slope standard curve) * dilution factor. Dilution factor: 55.

Conclusions
Seven artificial F-GSLs were generated by a combination of GSL precursors with five different fluorophores via two synthesis strategies. We demonstrate that the three F-GSLs, namely GSL-B-NBD, GSL-B-DNS, and GSL-A-Dansylamide, are transported actively by GTR1 and GTR2 and provide indications that they compete for a common binding site. Besides demonstrating their usefulness in heterologous GTR-based transport assays, the F-GSLs may be useful for unraveling the relationship between structure and function in the GTRs and possibly as tracers to monitor glucosinolate transport in vivo.