Detection of Bacterial α-l-Fucosidases with an Ortho-Quinone Methide-Based Probe and Mapping of the Probe-Protein Adducts

Fucosidases are associated with several pathological conditions and play an important role in the health of the human gut. For example, fucosidases have been shown to be indicators and/or involved in hepatocellular carcinoma, breast cancer, and helicobacter pylori infections. A prerequisite for the detection and profiling of fucosidases is the formation of a specific covalent linkage between the enzyme of interest and the activity-based probe (ABP). The most commonly used fucosidase ABPs are limited to only one of the classes of fucosidases, the retaining fucosidases. New approaches are needed that allow for the detection of the second class of fucosidases, the inverting type. Here, we report an ortho-quinone methide-based probe with an azide mini-tag that selectively labels both retaining and inverting bacterial α-l-fucosidases. Mass spectrometry-based intact protein and sequence analysis of a probe-labeled bacterial fucosidase revealed almost exclusive single labeling at two specific tryptophan residues outside of the active site. Furthermore, the probe could detect and image extracellular fucosidase activity on the surface of live bacteria.


Introduction
α-L-Fucosidases are enzymes capable of catalyzing the hydrolytic removal of terminal L-fucose residues from glycoconjugates. It is known that an abnormal increase in α-Lfucosidase activity in humans is associated with several pathological conditions [1][2][3][4]. Furthermore, bacterial α-L-fucosidases are key enzymes for the degradation and metabolism of intestinal mucin O-glycans by gut microbes. This crucial family of enzymes thereby contributes to the composition of the gut microbiota and influences our health and disease [5]. We and others recently reported that fucosidases of two Bacteroides species from the human gut microbiota induced upregulation of growth and invasive properties of pathogenic Campylobacter jejuni strains [6,7]. A growing number of studies are implicating bacterial fucosidases in the host-microbe interplay in the intestine, microbiota cross-feeding, and colonization resistance [3,5,[8][9][10]. strains [6,7]. A growing number of studies are implicating bacterial fucosidases in the host-microbe interplay in the intestine, microbiota cross-feeding, and coloniza tion resistance [3,5,[8][9][10].
α-L-Fucosidases employ one of two mechanisms to catalyze the hydrolysis o terminal α-fucosidic linkages, and these result in either overall retention or inver sion of the anomeric configuration. Fucosidases employing a double displacemen (Koshland) retaining mechanism belong to the glycoside hydrolase family 29 (GH29), whereas those employing an inverting mechanism belong to the glycoside hydrolase family 95 (GH95) [11,12]. Given their importance and the vast amounts of fucosidases in the human gut microbiome, several attempts have been made to develop ABPs to elucidate their individual functions [13][14][15][16]. However, the major ity of ABPs developed against fucosidases exploit the double replacement retaining mechanism of GH29 fucosidases and are thus incapable of targeting GH95 fuco sidases [13][14][15][16], thereby highlighting the need to expand the molecular toolbox to study all known and still to be discovered bacterial α-L-fucosidases.
Quinone methide-based probes are a type of ABP that would allow the labeling of both GH29 and GH95 fucosidases [17,18]. Upon α-L-fucosidase cleavage of the glycosidic bond in such a probe, the liberated ortho-fluoromethyl-phenolate agly cone undergoes rapid 1,4-elimination through expulsion of a fluoride ion to gener ate a highly electrophilic ortho-quinone methide. This reactive intermediate quickly alkylates a nearby nucleophile, which results in the formation of a covalently probe labeled enzyme adduct (Scheme 1A). This type of probe has already been shown to be capable of labeling β-galactosidases, neuraminidases of V. cholera, and human GH29 α-L-fucosidase enzymes [4,17,19,20]. One of the challenges in the application of quinone methide-based probes is the possibility for off-target labeling of nucle ophiles in nearby biomolecules due to diffusion of the ortho-quinone methide with its relatively long lifetime. However, this property also makes these probes an in teresting candidate for imaging of enzyme activity as the probes do not react with the active site, which would lead to inactivation of the enzyme. Instead, the possi bility for repeated hydrolysis of new copies of the probe could lead to the accumu lation of multiple adducts near the active site. This strategy with multiple adducts has previously been applied for signal amplification in imaging studies. [21,22] A quinone methide-based probe to detect fucosidase activity has previously been developed by Hsu et al. [19]. However, the application of this probe is limited A quinone methide-based probe to detect fucosidase activity has previously been developed by Hsu et al. [19]. However, the application of this probe is limited because of the bulky hydrophobic reporter group that might prevent the probe from accessing the active site of some α-L-fucosidases in gut microbiota [23]. Instead, the installation of a mini-tag, such as an azide or an alkyne, could overcome this limitation and make the probe more flexible in the detection of labeling. Thus, to enable effective in vivo imaging of both GH29 and GH95 bacterial fucosidases, we developed a more compact and flexible version of the Hsu et al. probe by the incorporation of an azide mini-tag [19]. Conjugation of a reporter group to the probe following labeling of the enzyme target was accomplished by the azide mini-tag reacting via Huisgen's 1,3-dipolar cycloaddition with a variety of alkyne-bearing reporter groups (Scheme 1B) [19].
Herein, we report the synthesis of a quinone methide-based ABP that can directly confer the activity of α-L-fucosidases. AH062 shows effective and selective labeling of both recombinant GH29 and GH95 fucosidases. We used Bacteroides fragilis as a model bacterium to investigate imaging properties of the fucosidase-activated AH062 and show specific extracellular fucosidase activity. Furthermore, we used the recombinantly expressed fucosidases from B. fragilis and Thermotoga maritima to investigate, by MS analysis, which are the preferred amino acid residues modified by AH062.

Results and Discussion
We synthesized the self-immobilizing fucosidase probe AH062 in eight steps by adapting the synthetic route previously published by Hsu et al. (Scheme 2). To determine the labeling efficiency of AH062, we incubated a recombinant B. fragilis fucosidase (BfFucH, GH29 fucosidase) at various concentrations (4-0.25 µg corresponding to 60-960 nM) with AH062 (350 µM) for 2 h at 37 • C. The resulting mixture was analyzed by SDS-PAGE followed by Western blotting. Detection was achieved by a Cu(I)-catalyzed click reaction on the blot surface with alkyne-biotin followed by incubation with anti-biotin HRP-linked antibody and detection using an enhanced chemiluminescence (ECL) substrate. The probe could detect the BfFucH enzyme at concentrations as low as 0.25 µg (60 nM) ( Figure 1A).
we developed a more compact and flexible version of the Hsu et al. probe by the incorporation of an azide mini-tag [19]. Conjugation of a reporter group to the probe following labeling of the enzyme target was accomplished by the azide minitag reacting via Huisgen's 1,3-dipolar cycloaddition with a variety of alkyne-bearing reporter groups (Scheme 1B) [19].
Herein, we report the synthesis of a quinone methide-based ABP that can directly confer the activity of α-L-fucosidases. AH062 shows effective and selective labeling of both recombinant GH29 and GH95 fucosidases. We used Bacteroides fragilis as a model bacterium to investigate imaging properties of the fucosidase-activated AH062 and show specific extracellular fucosidase activity. Furthermore, we used the recombinantly expressed fucosidases from B. fragilis and Thermotoga maritima to investigate, by MS analysis, which are the preferred amino acid residues modified by AH062.

Results and Discussion
We synthesized the self-immobilizing fucosidase probe AH062 in eight steps by adapting the synthetic route previously published by Hsu et al. (Scheme 2). To determine the labeling efficiency of AH062, we incubated a recombinant B. fragilis fucosidase (BfFucH, GH29 fucosidase) at various concentrations (4-0.25 µ g corresponding to 60-960 nM) with AH062 (350 µ M) for 2 h at 37 °C. The resulting mixture was analyzed by SDS-PAGE followed by Western blotting. Detection was achieved by a Cu(I)-catalyzed click reaction on the blot surface with alkyne-biotin followed by incubation with anti-biotin HRP-linked antibody and detection using an enhanced chemiluminescence (ECL) substrate. The probe could detect the BfFucH enzyme at concentrations as low as 0.25 µ g (60 nM) ( Figure 1A). Various concentrations of BfFucH were incubated with 350 μM of AH062 in PBS for 2 h at room temperature. Silver staining was included as a loading control. (B) Selective labeling of BfFucH (50 kDa) with AH062 in the presence of decreasing concentrations of TF (80 kDa). Silver staining was included as a loading control. (C) Different concentrations of AfcA were incubated with 350 μM of AH062 in PBS for 2 h at room temperature. Band intensity measurements were carried out by integrating pixel densities in the band of interest and subtracting the contribution of the background. PageBlue was included as a loading control. CL, chemiluminescence; SS, silver staining; PB, PageBlue.
AH062 was also capable of distinguishing the active from the inhibited enzyme (prepared by preincubation with the competitive inhibitor L-fuconojirimycin (FNJ)) [16]. Generally, a concern with quinone-methide based probes is their labeling selectivity, due to diffusion of the ortho-quinone methide trapping unit away from the targeted enzyme. To determine the labeling selectivity, we examined the ability of AH062 to specifically target fucosidases in the presence of a second protein. BfFucH samples were spiked with an equimolar concentration of transferrin (TF) and incubated with the probe. TF is a glycoprotein of 79.5 kDa and can therefore be clearly separated on gel from the ~50 kDa BfFucH. AH062 showed clear signals for BfFucH, but no chemiluminescent signal was detected for TF ( Figure 1B). These results suggest that AH062 is specifically reactive in or near the active site of the fucosidase. Furthermore, treatment of TF with AH062 in the absence of active BfFucH did not give any detectable signal.
To assess the ability of AH062 to visualize catalytically active inverting GH95 fucosidases, we expressed the known GH95 fucosidase AfcA from Bifidobacterium bifidum. AH062 was incubated with five different concentrations of AfcA (8-3.7 µ g), and the probe detected the enzyme in a concentration-dependent manner ( Figure  1C). AH062 labeled as little as 4.8 µ g of AfcA ( Figure 1C). To our knowledge, this is the first example of an ABP capable of labeling of an inverting fucosidase (GH95). We next performed mass spectrometric studies to determine the attachment sites of the novel quinone methide probe onto the fucosidase enzyme. After incubation of AH062 for 2 h at 37 °C with BfFucH, intact LC-MS profiles from labeled and untreated samples were compared. The spectrum of labeled proteins showed that approximately 6.1% of the fucosidases had a mass shift of 350 Da that fit with a single probe -protein adduct. Additionally, < 0.5% of the fucosidases had a mass shift that corresponded to two probe adducts on BfFucH ( Figure S1).
To identify the specific amino acids on the fucosidase surface that were being labeled by the probe, both labeled and unlabeled samples were digested with pepsin at a low pH followed by LC-MS/MS analysis. Unfortunately, the resulting AH062 was also capable of distinguishing the active from the inhibited enzyme (prepared by preincubation with the competitive inhibitor L-fuconojirimycin (FNJ)) [16]. Generally, a concern with quinone-methide based probes is their labeling selectivity, due to diffusion of the ortho-quinone methide trapping unit away from the targeted enzyme.
To determine the labeling selectivity, we examined the ability of AH062 to specifically target fucosidases in the presence of a second protein. BfFucH samples were spiked with an equimolar concentration of transferrin (TF) and incubated with the probe. TF is a glycoprotein of 79.5 kDa and can therefore be clearly separated on gel from the~50 kDa BfFucH. AH062 showed clear signals for BfFucH, but no chemiluminescent signal was detected for TF ( Figure 1B). These results suggest that AH062 is specifically reactive in or near the active site of the fucosidase. Furthermore, treatment of TF with AH062 in the absence of active BfFucH did not give any detectable signal.
To assess the ability of AH062 to visualize catalytically active inverting GH95 fucosidases, we expressed the known GH95 fucosidase AfcA from Bifidobacterium bifidum. AH062 was incubated with five different concentrations of AfcA (8-3.7 µg), and the probe detected the enzyme in a concentration-dependent manner ( Figure 1C). AH062 labeled as little as 4.8 µg of AfcA ( Figure 1C). To our knowledge, this is the first example of an ABP capable of labeling of an inverting fucosidase (GH95). We next performed mass spectrometric studies to determine the attachment sites of the novel quinone methide probe onto the fucosidase enzyme. After incubation of AH062 for 2 h at 37 • C with BfFucH, intact LC-MS profiles from labeled and untreated samples were compared. The spectrum of labeled proteins showed that approximately 6.1% of the fucosidases had a mass shift of 350 Da that fit with a single probe-protein adduct. Additionally, <0.5% of the fucosidases had a mass shift that corresponded to two probe adducts on BfFucH ( Figure S1).
To identify the specific amino acids on the fucosidase surface that were being labeled by the probe, both labeled and unlabeled samples were digested with pepsin at a low pH followed by LC-MS/MS analysis. Unfortunately, the resulting BfFucH samples did not yield good coverage for the adduct-labeled peptide fragments. Therefore, we repeated this analysis with the well-characterized bacterial GH29 TmFuc fucosidase for bottom-up analysis. We were now able to identify two unique peptide fragments that were labeled by the quinone methide from AH062.
To determine the exact site of the modification, tandem mass analysis was performed, whereby y n and b n were assigned based on the annotation proposed by Roepstorff and Fohlman (Figures 2A and S2). These data revealed that the quinone methide generated from AH062 had labeled either tryptophan W80 or W105. In Figure 2B, the modification sites are mapped on the known crystal structure of TmFuc. The modifications on W80 and W105 are positioned outside the fucosidase active site, separated, respectively, by 27 Å and 28 Å from the active-site nucleophile D224. Although quinone methides have been reported to react with various nucleophiles, the one generated from AH062 shows a high preference for the surface tryptophans in TmFuc. The previous literature has shown a high selectivity of a very similar quinone methide for tryptophan residues in proteins [24]. A closer look at the tryptophan units of TmFuc showed that W80 and W105 are two tryptophans located further away from the active site compared to various other tryptophan residues ( Figure S4), potentially indicating a slow generation of the quinone methide in AH062 that allows it to diffuse away from the active site. Additionally, sequence alignment with Clustal Omega on all three fucosidases used in this study showed that W80 is not conserved (Supplementary, Section S4). However, all fucosidases shown in the alignment do possess an aromatic amino acid residue at the W105 position. BfFucH samples did not yield good coverage for the adduct-labeled peptide fragments. Therefore, we repeated this analysis with the well-characterized bacterial GH29 TmFuc fucosidase for bottom-up analysis. We were now able to identify two unique peptide fragments that were labeled by the quinone methide from AH062.
To determine the exact site of the modification, tandem mass analysis was performed, whereby yn and bn were assigned based on the annotation proposed by Roepstorff and Fohlman (Figure 2A and Figure S2). These data revealed that the quinone methide generated from AH062 had labeled either tryptophan W80 or W105. In Figure  2B, the modification sites are mapped on the known crystal structure of TmFuc. The modifications on W80 and W105 are positioned outside the fucosidase active site, separated, respectively, by 27 Å and 28 Å from the active-site nucleophile D224. Although quinone methides have been reported to react with various nucleophiles, the one generated from AH062 shows a high preference for the surface tryptophans in TmFuc. The previous literature has shown a high selectivity of a very similar quinone methide for tryptophan residues in proteins [24]. A closer look at the tryptophan units of TmFuc showed that W80 and W105 are two tryptophans located further away from the active site compared to various other tryptophan residues ( Figure S4), potentially indicating a slow generation of the quinone methide in AH062 that allows it to diffuse away from the active site. Additionally, sequence alignment with Clustal Omega on all three fucosidases used in this study showed that W80 is not conserved (Supplementary, Section S4). However, all fucosidases shown in the alignment do possess an aromatic amino acid residue at the W105 position.

Figure 2. (A)
Tandem MS spectrum of a pepsin-generated peptide of TmFuc, revealing its labeling at W105. The annotated y and b ion series are shown above and below the peptide sequence, respectively. The y5 and b6 ions carry the mass increment of the probe (+350.16 Da), indicating the modification to be on the W105 residue. (B) Structure of TmFuc with its active-site catalytic acid D224 annotated in red, and the two residues modified by the quinone methide probe AH062 annotated in green (PDB code 1ODU).
To further assess the utility of this probe for imaging bacterial fucosida se activity, we used AH062 to stain fucosidases associated with the cell wall of B. fragilis. Anaerobic overnight cultures of B. fragilis were incubated for 2h at 37 °C with AH062. As a negative control, bacterial samples were used that were preincubated with 100 µ M of the competitive fucosidase inhibitor FNJ, followed by incubation with the probe. To visualize the probe, the bacteria were clicked by a copper(I)catalyzed alkyne-azide cycloaddition (CuAAC) to an Alexa488 dye, and the bacterial cell membranes were stained with CellTrace TM Yellow (CTY) to facilitate locating the bacteria ( Figure 3A). Incubation with AH062 resulted in highly specific fluorescence labeling that was not observed after preincubation with FNJ. The AH062 Figure 2. (A) Tandem MS spectrum of a pepsin-generated peptide of TmFuc, revealing its labeling at W105. The annotated y and b ion series are shown above and below the peptide sequence, respectively. The y5 and b6 ions carry the mass increment of the probe (+350.16 Da), indicating the modification to be on the W105 residue. (B) Structure of TmFuc with its active-site catalytic acid D224 annotated in red, and the two residues modified by the quinone methide probe AH062 annotated in green (PDB code 1ODU).
To further assess the utility of this probe for imaging bacterial fucosidase activity, we used AH062 to stain fucosidases associated with the cell wall of B. fragilis. Anaerobic overnight cultures of B. fragilis were incubated for 2h at 37 • C with AH062. As a negative control, bacterial samples were used that were preincubated with 100 µM of the competitive fucosidase inhibitor FNJ, followed by incubation with the probe. To visualize the probe, the bacteria were clicked by a copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC) to an Alexa488 dye, and the bacterial cell membranes were stained with CellTrace TM Yellow (CTY) to facilitate locating the bacteria ( Figure 3A). Incubation with AH062 resulted in highly specific fluorescence labeling that was not observed after preincubation with FNJ. The AH062 pattern of labeling showed a similar localization on the outer membrane compared to the CTY cell membrane staining. By applying high-resolution structured illumination microscopy (SIM), we confirmed that AH062 stained the B. fragilis cell wall or cell membrane ( Figure 3B). We next performed FACS analysis of AH062-labeled bacteria and observed a shift in the whole peak, indicating that most of the bacteria were fluorescently labeled by covalent attachment of the activated quinone methide. Again, labeling with AH062 could be blocked by preincubation with FNJ, showing the selectivity of the probe ( Figure 4). pattern of labeling showed a similar localization on the outer membrane compared to the CTY cell membrane staining. By applying high-resolution structured illumination microscopy (SIM), we confirmed that AH062 stained the B. fragilis cell wall or cell membrane ( Figure 3B). We next performed FACS analysis of AH062-labeled bacteria and observed a shift in the whole peak, indicating that most of the bacteria were fluorescently labeled by covalent attachment of the activated quinone methide. Again, labeling with AH062 could be blocked by preincubation with FNJ, showing the selectivity of the probe (Figure 4).   to the CTY cell membrane staining. By applying high-resolution structured illumination microscopy (SIM), we confirmed that AH062 stained the B. fragilis cell wall or cell membrane ( Figure 3B). We next performed FACS analysis of AH062-labeled bacteria and observed a shift in the whole peak, indicating that most of the bacteria were fluorescently labeled by covalent attachment of the activated quinone methide. Again, labeling with AH062 could be blocked by preincubation with FNJ, showing the selectivity of the probe (Figure 4).

General Methods and Materials
All reagents and starting materials were obtained from commercial suppliers and were used without further purification. MeCN and CH 2 Cl 2 and were dried and purified using an MBraun MB SPS 800 prior to use. Pyridine and DMSO were dried for 24 h over pre-activated (5 min,~300 • C) 4Å molecular sieves prior to use. Glassware for anhydrous reactions was flame-dried and cooled under a nitrogen atmosphere immediately prior to use. Analytical TLC was performed on glass-backed TLC plates pre-coated with silica gel (60G, F 254 ). Compound 10 was visualized by staining with molybdenum, ninhydrin, and 10% PPh 3 in CH 2 Cl 2 followed by ninhydrin. All other compounds were visualized under UV light and/or by staining with molybdenum and 5% sulfuric acid in EtOH. Column chromatography was performed with silica gel (230-400 mesh). 1 H NMR, 13 C NMR, and 19 F NMR spectra were recorded on either a Bruker Avance 600 MHz NMR spectrometer or 400 MHz spectrometer in the designated deuterated solvents (CDCl 3 or CD 3 OD). 1 H and 13 C NMR peak assignments were established based on 1 H-1 H COSY and 1 H-13 C HSQC experiments and, where possible, compared to previously reported data [19]. Chemical shifts (δ) are listed in ppm downfield from TMS using TMS as an internal reference. Coupling constants are reported in Hz. The following abbreviations were used to explain the multiplicities: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, b = broad. Expression and purification of the recombinant proteins BfFucH and AfcA were based on previous publications [11,25]. Recombinant TmFuc and tissue-derived transferrin (TF), both of sufficient quality, were commercially available via, respectively, Megazyme (9037-65-4) and Calbiochem (616397).

General Methods and Materials
All reagents and starting materials were obtained from commercial suppliers and were used without further purification. MeCN and CH2Cl2 and were dried and purified using an MBraun MB SPS 800 prior to use. Pyridine and DMSO were dried for 24h over pre-activated (5 min, ~300 °C) 4Å molecular sieves prior to use. Glassware for anhydrous reactions was flame-dried and cooled under a nitrogen atmosphere immediately prior to use. Analytical TLC was performed on glass-backed TLC plates pre-coated with silica gel (60G, F254). Compound 10 was visualized by staining with molybdenum, ninhydrin, and 10% PPh3 in CH2Cl2 followed by ninhydrin. All other compounds were visualized under UV light and/or by staining with molybdenum and 5% sulfuric acid in EtOH. Column chromatography was performed with silica gel (230-400 mesh). 1 H NMR, 13 C NMR, and 19 F NMR spectra were recorded on either a Bruker Avance 600 MHz NMR spectrometer or 400 MHz spectrometer in the designated deuterated solvents (CDCl3 or CD3OD). 1 H and 13 C NMR peak assignments were established based on 1 H-1 H COSY and 1 H- 13 C HSQC experiments and, where possible, compared to previously reported data [19]. Chemical shifts (δ) are listed in ppm downfield from TMS using TMS as an internal reference. Coupling constants are reported in Hz. The following abbreviations were used to explain the multiplicities: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, b = broad. Expression and purification of the recombinant proteins BfFucH and AfcA were based on previous publications [11,25]. Recombinant TmFuc and tissue-derived transferrin (TF), both of sufficient quality, were commercially available via, respectively, Megazyme (9037-65-4) and Calbiochem (616397). After completion of the reaction, the mixture was concentrated under reduced pressure. The residue was dissolved in EtOAc (50 mL), and the resulting solution was washed with 5% citric acid (3 x 25 mL), 5% NaHCO3 (3 × 25 mL), H2O (3 × 25 mL), and brine (25 mL). The organic phase was dried over anhydrous Na2SO4, filtered, and concentrated under reduced pressure which yielded a dark orange powder. Purification by silica gel column chromatography (eluent: 70% PetEt/EtOAc) provided the desired glycosylated product 5 L-Fucose (1 g, 6.1 mmol) was dissolved in pyridine (6 mL) and Ac 2 O (6 mL), and the reaction mixture was stirred for 5 h at room temperature. Solvents were removed under reduced pressure, and the residue obtained was dissolved in 20 mL DCM. The solution was washed with 1M HCl (3 × 20 mL), H 2 O (3 × 20 mL), and brine (20 mL), dried over anhydrous Na 2 SO 4 , filtered, and concentrated under reduced pressure. Purification of per-acetylated L-fucose by silica gel column chromatography (eluent: 80% PetEt/EtOAc) provided the desired product in 99% yield as a colorless oil. R f = 0.45 (PetEt/EtOAc = 2/1).

General Methods and Materials
All reagents and starting materials were obtained from commercial suppliers and were used without further purification. MeCN and CH2Cl2 and were dried and purified using an MBraun MB SPS 800 prior to use. Pyridine and DMSO were dried for 24h over pre-activated (5 min, ~300 °C) 4Å molecular sieves prior to use. Glassware for anhydrous reactions was flame-dried and cooled under a nitrogen atmosphere immediately prior to use. Analytical TLC was performed on glass-backed TLC plates pre-coated with silica gel (60G, F254). Compound 10 was visualized by staining with molybdenum, ninhydrin, and 10% PPh3 in CH2Cl2 followed by ninhydrin. All other compounds were visualized under UV light and/or by staining with molybdenum and 5% sulfuric acid in EtOH. Column chromatography was performed with silica gel (230-400 mesh). 1 H NMR, 13 C NMR, and 19 F NMR spectra were recorded on either a Bruker Avance 600 MHz NMR spectrometer or 400 MHz spectrometer in the designated deuterated solvents (CDCl3 or CD3OD). 1 H and 13 C NMR peak assignments were established based on 1 H-1 H COSY and 1 H- 13 C HSQC experiments and, where possible, compared to previously reported data [19]. Chemical shifts (δ) are listed in ppm downfield from TMS using TMS as an internal reference. Coupling constants are reported in Hz. The following abbreviations were used to explain the multiplicities: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, b = broad. Expression and purification of the recombinant proteins BfFucH and AfcA were based on previous publications [11,25]. Recombinant TmFuc and tissue-derived transferrin (TF), both of sufficient quality, were commercially available via, respectively, Megazyme (9037-65-4) and Calbiochem (616397).
3.2.6. (2S,3S,4R,5R,6S)-2-(4-(tert-butoxycarbonylamino)-2-(fluoromethyl)phenoxy)-6-methyltetrahydro-2H-pyran-3,4,5-triyl triacetate 9 A solution of diethylaminosulfur trifluoride (DAST, 79 µ L, 0.6 mmol) in anhydrous CH2Cl2 was added to an ice-cooled solution of compound 8 (204 mg, 0.40 mmol) in anhydrous CH2Cl2. The reaction mixture was allowed to warm up to room temperature and was stirred for 3.5 h under an argon atmosphere. The reaction was quenched by adding a small amount of silica and 10 drops of MeOH, followed by stirring for 10 min. Then, the reaction mixture was filtered over a fritted glass funnel and concentrated under reduced pressure. Purification by silica gel column chromatography (eluent: 80% PetEt/EtOAc) provided the desired product 9 (133 mg, 0.26 mmol) in 65% yield as a pale yellow oil. Rf The 1 H and 13 C spectra of this known compound were identical to previously reported spectra [19]. The 1 H spectrum of this known compound was identical to a previously reported spectrum [19].   The 1 H and 13 C spectra of this known compound were identical to previously reported spectra [19]. A solution of diethylaminosulfur trifluoride (DAST, 79 µL, 0.6 mmol) in anhydrous CH 2 Cl 2 was added to an ice-cooled solution of compound 8 (204 mg, 0.40 mmol) in anhydrous CH 2 Cl 2 . The reaction mixture was allowed to warm up to room temperature and was stirred for 3.5 h under an argon atmosphere. The reaction was quenched by adding a small amount of silica and 10 drops of MeOH, followed by stirring for 10 min. Then, the reaction mixture was filtered over a fritted glass funnel and concentrated under reduced pressure. Purification by silica gel column chromatography (eluent: 80% PetEt/EtOAc) provided the desired product 9 (133 mg, 0.26 mmol) in 65% yield as a pale yellow oil. R f = 0.78 (PetEt/EtOAc = 1/1). 1  The 1 H and 13 C spectra of this known compound were identical to previously reported spectra [19]. The 1 H and 13 C spectra of this known compound were identical to previously reported spectra [19].

Labeling Selectivity Assay
BfFucH (960 nM) was incubated with 350 µM AH062 in the presence of decreasing concentrations of TF (960-60 nM) for 2 h at room temperature in PBS. The enzyme mixtures were transferred, conjugated to alkyne-biotin, and visualized as described above.

Mass Spectrometry
For the peptide-centric LC-MS-MS, 5 µg expressed fucosidases, either labeled or unlabeled, were denatured and reduced using 5 mM tris(2-carboxyethyl)phospine (TCEP) and incubated for 15 min at 60 • C. Samples were subsequently alkylated with 20 mM chloroacetamide (CAA), digested proteolytically using pepsin (Roche) in a 0.2% trifluoroacetic acid (TFA) buffer at pH < 2 (1:75 enzyme/protein ratio), and incubated overnight at 37 • C. The protein was subjected to several proteases including trypsin, chymotrypsin, and glu-C (data not shown). Pepsin proved to be the most suitable for producing peptides at appropriate lengths with the highest sequence coverage. Following digestion, solid-phase extraction was performed using Oasis microElution 96-well plates (Waters, Wexford, Ireland). In short, wells were conditioned with acetonitrile (ACN) and then equilibrated with 0.1% TFA. The supernatant from the digestion was loaded and washed twice with 0.1% TFA. Peptides were eluted using 60% ACN 0.1 % TFA. Peptide eluates were dried by low-pressure rotary evaporation and resuspended in 2% formic acid for LC-MS 2 analysis.
Per peptide sample, 50 ng was analyzed using a Thermo Scientific Ultimate HPLC nanoflow system, hyphenated to an Orbitrap Exploris 480 Mass Spectrometer (Thermo Fisher Scientific (Waltham, Massachusetts, MA, USA). Buffer A (0.1% formic acid (FA) in water) and buffer B (0.1% FA, 20% ACN in water) were used for peptide chromatography with the following gradient: 0 min 9% B; 2 min 13% B; 39 min 44% B; 42 min 99% B; 46 min 99% B; 47-55 min 9% B. Mass spectrometry analysis was performed in positive ion mode using electrospray ionization from a coated fused silica emitter at a 1900 V spray voltage. For MS 1 scans, the mass range was set from m/z 375 to 1600 with a resolution of 60,000. The AGC target was set to the MS standard with automated maximum injection times. The data-dependent MS 2 acquisition method initiated HCD fragmentation at 28% of normalized collision energy on the highest charge state and lowest m/z signals within a 1 s cycle time. The exclusion time was set to 10 s. The resolution for MS 2 acquisition was at 15,000 with a mass range from m/z 40 to 2500. Here, the AGC target was also set to standard with automated maximum injection times. For data analysis of LC-MS 2 , we PBS and taken up in 1 mL aquadest. Bacteria were diluted an additional 10× in aquadest and analyzed using a BD FACSVerse. Flow cytometry data were analyzed using FlowJo 10.

Conclusions
In summary, we report a novel fucosidase probe (AH062) that is capable of labeling both GH29 and GH95 bacterial fucosidases, thereby showing, for the first time, a probe that targets inverting fucosidases. The quinone methide generated by this probe mainly labels on specific tryptophan residues outside of the active site of the TmFuc fucosidase enzyme, with single protein-probe adducts dominating. The probe was successfully used for the detection of fucosidase-positive bacteria by high-resolution confocal microscopy and flow cytometry. For further application of this probe in complex co-culture systems, the likely occurrence of quinone methide labeling adducts on off-target proteins in co-cultures needs to be determined and minimized. In the future, we envision applying this novel probe for the labeling, identification, and characterization of important bacterial GH29 and GH95 fucosidases at the intestinal host-microbe interface.
Supplementary Materials: The following supporting information can be downloaded at, Section S1: ( Figure S1: Zero-charge deconvoluted mass spectra of the intact BfFucH analyzed by LC-MS, Figure S2: LC MS-MS spectra of pepsin generated peptides from AH062-labeled TmFuc indicative for the attachment of the probe to either W80 and W105, Figure S3: Raw Blots and gels that are presented as cropped images in the main manuscript as Figure 1, Figure S4: Structure of TmFuc.) Section S2. NMR spectra. Section S3. Protein sequences BfFuchH and TmFuc. Section S4 Clustal Omega alignment of BfFucH, TmFuc and AfcA.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.