Synthesis, trehalase hydrolytic resistance and inhibition properties of 4- and 6-substituted trehalose derivatives

Abstract Although trehalose has recently gained interest because of its pharmaceutical potential, its clinical use is hampered due to its low bioavailability. Hence, hydrolysis-resistant trehalose analogues retaining biological activity could be of interest. In this study, 34 4- and 6-O-substituted trehalose derivatives were synthesised using an ether- or carbamate-type linkage. Their hydrolysis susceptibility and inhibitory properties were determined against two trehalases, i.e. porcine kidney and Mycobacterium smegmatis. With the exception of three weakly hydrolysable 6-O-alkyl derivatives, the compounds generally showed to be completely resistant. Moreover, a number of derivatives was shown to be an inhibitor of one or both of these trehalases. For the strongest inhibitors of porcine kidney trehalase IC50 values of around 10 mM could be determined, whereas several compounds displayed sub-mM IC50 against M. smegmatis trehalase. Dockings studies were performed to explain the observed influence of the substitution pattern on the inhibitory activity towards porcine kidney trehalase.


Introduction
Carbohydrates are the most diverse and abundant biomolecules on earth, displaying a wide variety of functions 1 . During the last decades, interest arose in trehalose 1 (Figure 1), a disaccharide consisting of two D-glucose units linked via an a-1,1-a-bond, resulting in a molecule with unique stabilising properties 2,3 . The stability of this non-reducing glucobiose is reflected by several interesting physicochemical properties like broad pH stability 4 , high glass transition and melting temperatures 5,6 , low hydrolysis rate 5 . During heating and processing of food products acrylamide formation can be suppressed by trehalose as it interacts with glucose and thus reduces the glucose-Asn reaction leading to the toxic compound 4,6 . In nature, trehalose fulfils a biological role in various organisms including bacteria, yeast, fungi, insects, invertebrates and plants, although it is not found in mammals 6,7 . It can serve as energy and carbon source 6,7 , signalling molecule 6,8,9 and cell wall building block 6,10 . Furthermore, it can protect organisms and compounds under stress conditions like desiccation 7,11,12 , dehydration 7,13-15 , heat 7,13 , cold 7 and oxidation 7,14 .
Since the establishment of the Hayashibara process, the availability of trehalose has drastically increased, leading to its widespread use in the food, cosmetic and pharmaceutical industry 6 . Moreover, due to its safeguarding capabilities, trehalose has received attention as a chemical chaperone 16,17 and autophagy inducer 16 in the treatment of neurodegenerative diseases like Alzheimer's 18,19 , Huntington's 20 and Parkinson's 16,21,22 . Unfortunately, the therapeutic use of trehalose is hampered due to its low bioavailability. Indeed, this disaccharide is rapidly degraded into glucose by the trehalase (EC 3.2.1.28) present in our small intestine 23 . Trehalose analogues and derivatives that show resistance against this human intestinal enzyme could, therefore, be of great interest for pharmaceutical formulations and for drug discovery due to their increased residence time in the human body 3,23 . Recently, a trehalose analogue substituted on the 4-hydroxy group, lentztrehalose A 2 (Figure 1), was isolated from the actinomycete Lentzea sp. and was shown to be only weakly hydrolysed by porcine kidney trehalase 24 . Furthermore, it also exhibited antitumor activity 24 . Subsequently, Wada and co-workers isolated two additional natural analogues, lentztrehalose B 3 and C 4 (Figure 1), which were shown to be possible inducers of autophagy 25 . Importantly, these lentztrehaloses were found not to be digested in a range of microbes and cancer cell lines, and their in vivo bioavailability and stability was confirmed after oral administration to mice 26 .
In this study, an exploration of 34 trehalose derivatives (including 33 new compounds next to lentztrehalose A 2) was established. Inspired by the lentztrehaloses A-C 2-4, a series of 4-Osubstituted trehalose derivatives was synthesised using an etheror carbamate-type linkage. In addition, the analogous 6-O-substituted series was explored while some double substituted derivatives were also investigated. Their biological relevance was assessed by measuring their hydrolysis (as substrate) and binding (as inhibitor) with two relevant trehalases, i.e. porcine kidney trehalase (91% similarity with human trehalase) and Mycobacterium smegmatis trehalase (93% similarity with M. tuberculosis trehalase). These enzymes are classified in family GH37 and GH15, respectively, of the CAZy-classification. Although they adopt a similar overall fold (i.e. an (a/a) 6 -barrel) and follow the same general reaction mechanism (i.e. inversion of the anomeric configuration through single displacement), they are otherwise unrelated (about 20% similarity). Computational studies were performed to clarify the structural determinants behind the activities and selectivity profiles of the studied trehalose derivatives towards porcine kidney trehalase.

Chemistry
General remarks All reactions, unless otherwise stated, were carried out under argon atmosphere in dry solvents. Dichloromethane and triethylamine were freshly distilled from CaH 2 . Toluene was freshly distilled from Na. Tetrahydrofuran was freshly distilled from Na/ benzophenone. Other solvents and reagents were obtained from commercial sources and were used as received without further purification. Flash chromatography was carried out with Rocc silicagel 60 Å, 40À63 mm. Precoated silica gel plates (Macherey-Nagel SIL G-25 UV 254 ) were used for TLC employing UV-absorption at 254 nm and Mo 7 O 24 /Ce(SO 4 ) 2 /aq.H 2 SO 4 staining for visualisation. Electrospray mass spectra were recorded on an Agilent 1100 series single quadrupole MS detector type VL with an APCI source and an API-ES source, provided with a Phenomenex Luna C18 (2), 5 mm 250 mm Â 4.60 mm column. High resolution mass spectrometry (HRMS) was performed on an Agilent 1100 series connected to a 6220 A TOF-MS detector equipped with an APCI-ESI multimode source. 1 H-NMR and 13 C-NMR spectra (see Supplementary information) were recorded on a Bruker Avance 300, Bruker Avance 400 or a Bruker AM 500 spectrometer as indicated with chemical shifts reported in parts per million, referenced to the residual solvent signals (CDCl 3 13 C-NMR Spectra recorded in D 2 O were referenced to the signal (30.89 ppm) of acetone (1 drop added). Coupling constants, J, are reported in hertz (Hz). Infra-red spectra were recorded on a Perkin-Elmer 1000 FT-IR infra-red spectrometer (horizontal attenuated total reflection (HATR)). Optical rotation was measured on a Perkin Elmer 241 Polarimeter.

General procedure A: O-alkylation
To a solution of an OH-containing starting material in anhydrous DMF (0.1 M concentration) was added NaH (60% in mineral oil, 2.5 eq for each hydroxyl group). After stirring for 15 min, the alkyl halogenide (4 eq for each hydroxyl group) was added and the reaction mixture was stirred overnight. Methanol (5 ml per mmol starting material) was added dropwise (caution: gas formation) and the mixture was transferred to a separation funnel using EtOAc (50 ml per mmol starting material). The organic layer is washed with brine (5Â, 50 ml per mmol starting material) and concentrated under reduced pressure. The residue was purified by column chromatography.
General procedure B: carbamate formation To a solution of an OH-containing starting material in CH 2 Cl 2 (0.1 M concentration) was added DMAP (0.2 eq for each hydroxyl group) and the isocyanate (3 eq for each hydroxyl group). The reaction mixture was stirred at room temperature until the consumption of starting material (TLC monitoring, 24-120 h). The mixture was concentrated under reduced pressure and the residue was purified by column chromatography.
General procedure C: hydrogenolysis To a solution of a benzyl/benzylidene-protected starting material in EtOAc/MeOH (1/1, 0.05 M concentration), Pd/C (10% w/w Pd, 0.05 eq) was added. A H 2 balloon was placed and the reaction mixture was stirred overnight after which it was filtered over celite. The filtrate was concentrated under reduced pressure and the residue was purified by column chromatography.
2,3,4,2',3'-Penta-O-benzyl-4',6'-O-benzylidene-a,a-D-trehalose (7) To a cooled (À18 C) solution of 6 29 (5.576 g, 6.35 mmol, 1 eq) in toluene (56 ml), DIBAL-H (supplied as 1.0 M solution in toluene, 31.9 ml, 31.9 mmol, 5 eq) was added dropwise. The cooling bath was removed and the reaction mixture was stirred at room temperature for 90 min. The solution was cooled to 0 C and the reaction was quenched by dropwise addition of MeOH (11.5 ml) and aqueous KOH (10% w/v, 3.75 ml). The resulting suspension was transferred to a separation funnel using CH 2 Cl 2 (400 ml) and H 2 O (400 ml). The organic layer was separated and the aqueous phase was extracted using CH 2 Cl 2 (3 Â 400 ml). The combined organic layers were dried on MgSO 4 , the drying agent was filtered and the filtrate was concentred under reduced pressure. The residue was purified by flash column chromatography (gradient elution: hexane/EtOAc 8/2 to 1/1) to obtain the title compound 7 as a white foam with a yield of 80% (4.443 g; 5.05 mmol). Spectral data were in agreement with literature 27 .
To a solution of crude 13 in anhydrous DMF (12 ml), NaH (60% in mineral oil, 700 mg, 17.4 mmol) was added slowly (gas formation). After stirring for 15 min, the resulting suspension was cooled to 0 C and BnBr (1.67 ml, 13.9 mmol) was added dropwise, followed by TBAI (129 mg, 0.348 mmol). The ice bath was removed and the reaction mixture was stirred overnight. After cooling to 0 C, MeOH (5 ml) was added (gas formation is observed) and the mixture is stirred for 10 min after which it is diluted with ethyl acetate (40 ml). The organic layer was washed with brine (5 Â 40 ml), and dried on MgSO 4 . The drying agent was filtered and the filtrate was concentrated under reduced pressure. The crude product was purified by flash column chromatography (gradient elution: hexane/EtOAc 95/5 to 8/2) giving 11 as a white solid with a yield of 51% over two steps (725 mg, 0.747 mmol).
General procedure A was applied (use of 1-bromodecane).
Purification via column chromatography (gradient elution: hexane/ EtOAc 9/1 to 8/2) afforded the partially purified title compound 16-g as a white solid (320 mg), which was used as such in the next step.

Enzyme activity assays
Trehalose and each derivative (20 mM) were incubated with 0.01 mg ml À1 trehalase from porcine kidney (Sigma Aldrich) or Mycobacterium smegmatis (NZYtech) in 100 mM of sodium phosphate buffer (pH 7) at 37 C for 24 h. Mixtures with Mycobacterium smegmatis trehalase also contained 6 mM of MgSO 4 , as the enzyme requires Mg 2þ for activity 32 . Samples were taken at regular time intervals and diluted with ultrapure water for analysis with high performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD, Dionex ICS-3000).
In order to evaluate their relative inhibition percentage, each trehalose derivative (20 mM) was incubated with 0.01 mg ml À1 of above-mentioned trehalases with 3 or 20 mM of trehalose for porcine kidney and Mycobacterium smegmatis trehalases, respectively, in 100 mM of sodium phosphate buffer (pH 7) at 37 C for 14 min. Again, 6 mM of MgSO 4 was added to the reaction mixture with Mycobacterium smegmatis trehalase. Samples were taken every 2 min and diluted with ultrapure water for analysis with HPAEC-PAD. The residual activity of the enzyme towards trehalose in each mixture was determined by comparing the measured activity to the activity determined in the absence of trehalose derivative.
For selected compounds 2, 17-a, 19-d, 19-e, 19-f, 21-a, 23-f, and 31, IC 50 values were determined by incubating different concentrations of the trehalose derivative (0.01-50 mM) with the selected trehalases under the same conditions as mentioned above. Samples were taken every 2 min and diluted for analysis with HPAEC-PAD. The residual activity of the enzyme towards trehalose in each mixture was determined compared to a sample without trehalose derivative. Plots and curve fits were obtained via Sigmaplot 13. All reactions were performed in triplicate, values are represented as the mean and standard deviations were in the range of 15% of the reported value.

Molecular modelling
All manipulations were performed with the molecular modelling program YASARA and the YASARA/WHATIF twinset and figures were created with PyMOL 2.0. The ligand free and inhibitor bound crystal structures of trehalase from Enterobacter cloacae (GH37 family) (PDB code 5Z6H and 5Z66, respectively) 33 were used as template for docking. Trehalose derivative structures were created with YASARA Structure and subsequently minimised with the AMBER03 force field. The grid box for docking had a dimension of 25 Â 25 Â 25 Å and comprised the entire catalytic cavity. Docking was performed with AutoDock VINA using default parameters and ligands were allowed to rotate freely during the simulation. The conformer with an appropriate glucose in both the À1 and þ1 subsite was selected as the binding mode for further analysis.

Synthesis of strategic intermediates
Generally, synthesis of all final analogues was achieved from a central set of partially protected strategic intermediates 7-10, 12 and 15 which could be obtained from trehalose 1 (Scheme 1). After protection of both 4-and 6-hydroxyl groups in trehalose as benzylidene acetal, the remaining OH functional groups were converted to the corresponding benzyl ethers, leading to 6 27 . In the following crucial desymmetrisation step, regioselective reductive opening of one of the acetal moieties to either the 4-or 6-monoalcohol can be achieved using diisobutylaluminium hydride (DIBAL-H) in an appropriate solvent, according to existing literature 27 . Based on reported conditions, selective opening towards strategic intermediate 7 containing a free 6-OH group was achieved through reaction in toluene. The alternative 4-OH isomer 8 could be obtained via DIBAL-H reduction in dichloromethane: in a typical experiment, starting material still remained at moment of quenching (10% recovered), while in addition to the desired compound 8 (55%), regio-isomer 7 (19%) and product 9 (9%, resulting from overreduction) were also isolated through tedious chromatographic separation. Compound 9 was found to be useful as an additional strategic intermediate to access 4/4 0 -bis-substituted derivatives; the corresponding 6/6 0 -bis-OH intermediate 10 was obtained after a second DIBAL-H treatment of 7 in toluene. To avoid the above-mentioned practical issues concerning synthesis of 8, we decided to additionally synthesise alternative strategic intermediate 12, also allowing the subsequent synthesis of 4-substituted derivatives. This was achieved via sequential O-benzylation of 7 and opening of the resulting mono-benzylidene acetal 11 using Me 3 N-BH 3 /AlCl 3 34 . Under the latter conditions, the desired 4-OH isomer 12 could be isolated in 49% yield, while the fraction of corresponding 6-OH regio-isomer amounted to 38% (compound not shown). Alternatively, intermediate 11 could be obtained via the mono-benzylidene acetal-protected 13. Although this was expected to involve a shorter reaction route from trehalose, an extra acetylation/deacetylation sequence via 14 was found to be necessary to facilitate intermediate purification, leading to an equal number of reaction steps but significantly lower overall yield. Finally, acid hydrolysis of the acetal moiety in 11 was used to obtain 4/6-bis-OH strategic intermediate 15. With these strategic intermediates in hand, different sets of 4-or 6-monosubstituted, and 4/4 0 -, 6/6 0 -and 4/6-double substituted analogues were accessible, via derivatisation of the remaining unprotected alcohol moieties.

Derivatisation of strategic intermediates to final compounds
First, using strategic intermediate 7, two series of 6-O-monosubstituted derivatives were prepared (Scheme 2). On one hand, O-alkylation using alk(en)yl halogenides with increasing chain length under standard conditions delivered intermediates 16-a-g, which were overall deprotected using catalytic hydrogenolysis to obtain the intended 6-O-alkylated trehalose derivatives 17-a-g. On the other hand, a series of 6-O-carbamoyl derivatives 19-a-g was prepared via treatment of 7 with a set of isocyanates in presence of DMAP, followed by hydrogenolysis.
Next, two analogous series of 4-O-substituted trehalose derivatives were envisaged (Scheme 3). Indeed, alkylation of 8 using the same set of alk(en)yl halogenides mentioned above gave the expected ethers 20-a-g which were deprotected to the final derivatives 21-a-g. Additionally, lentztrehalose A 2 was synthesised following reported conditions, via Sharpless asymmetric dihydroxylation of 4-O-prenylated intermediate 20-f and subsequent hydrogenolysis of 20-h 30 . From 1 H-NMR analysis, a diastereomeric excess of 73% 20-h could be determined ( Figure S1 in Supplementary Information). Both diastereomers could not be separated chromatographically and the resulting lentztrehalose A 2 was therefore used as such for trehalase evaluation. In analogy to the 6-O-series 19-a-g, we also prepared the corresponding 4-O-carbamates 23-a-g in two steps from strategic intermediate 12.

Biological evaluation
Resistance to trehalase hydrolytic activity The hydrolytic resistance of each trehalose derivative was evaluated by incubating 20 mM of substrate with 0.01 mg ml À1 of enzyme at pH 7 and 37 C for 24 h. No degradation was observed for the majority of derivatives, whereas trehalose was already broken down completely after just 12 h (data not shown). The only exceptions showing significant hydrolysis were derivatives 17-a (7 and 8% relative hydrolysis by the trehalase from porcine kidney and M. smegmatis, respectively), 17-b (1 and 6%, respectively) and 17-c (1% and 1%, respectively), which all contain a small alkyl group (i.e. methyl, ethyl and n-propyl, respectively) on the 6-O-position.

Inhibitory activity towards trehalase
The inhibitory activity of the compounds was evaluated at a concentration of 20 mM, using trehalose as substrate at a concentration equal to the respective Michaelis-Menten constant (3 mM and 20 mM for porcine kidney and M. smegmatis trehalase, respectively). Despite their resistance to hydrolysis, several of these compounds were found to still bind quite tightly to these trehalases, causing enzyme inhibition (Table 1).
Regarding porcine kidney trehalase, the most interesting inhibitors are grouped within the series of 6-O-substituted derivatives, as their counterparts with 4-O-substitutions are all poor inhibitors ( 35% inhibition at 20 mM). Only two derivatives, namely 17-a and 19-f, exhibited complete inhibition. However, these compounds bear very different substituents in both size and chemical properties, i.e. methyl ether and phenethyl carbamate, respectively. Compared to 17-a, extension of the alkyl ether substituent resulted in a clear drop in inhibition. Similar to compound 19-f, the other phenyl group-containing carbamates, 19-d and 19-e, exhibited significant inhibitory effects (>80%). Compounds in the carbamate series devoid of a phenyl moiety, i.e. 19-a, 19-b, 19-c and 19-g, showed a clearly lowered inhibition, suggesting that the aromatic group forms a crucial interaction with the enzyme. Interestingly, compared to 17-a and 19-f, the additional introduction of either of their substituents on the 6 0 -O-position, i.e. leading to 25, 27 and 34, led to a lower rather than a higher affinity. Likewise, the introduction of an additional methyl substituent on the 4-O-position in 31 led to a similar lowered inhibition outcome.   with longer (n-octyl/n-decyl) hydrophobic chains at the 4-or 6-Opositions (17-g, 19-g, 21-g and 23-g), resulted in a lack of inhibitory effect towards M. smegmatis trehalase.
To further rank the derivatives that caused complete inhibition, the abovementioned experiments were repeated with a lower inhibitor concentration, i.e. 1 mM instead of 20 mM (Table 1). Additionally, eight trehalose derivatives were also subjected to half maximal inhibitory concentration (IC 50 ) experiments (Table 2). This confirmed that compounds 17-a and 19-f are the strongest inhibitors for porcine kidney trehalase, while derivatives 2, 17-a, 19-f and 31 displayed sub-mM IC 50 values against M. smegmatis trehalase. Interestingly, the IC 50 of the strongest inhibitors of porcine kidney trehalase is about 10 mM whereas substantially lower values (down to 0.35 mM) are found for the M. smegmatis trehalase.

Molecular docking studies
To clarify the displayed activities of the new derivatives, molecular docking studies were carried out. Unfortunately, no structural elucidation was possible for the enzyme from M. smegmatis as no crystal structure of a trehalase from family GH15 has yet been determined. However, several crystal structures of trehalases from family GH37 are described in literature, namely from Escherichia coli, Saccharomyces cerevisiae and Enterobacter cloacae 33,[35][36][37] . Interestingly, the latter has been crystallised in two different forms, i.e. with and without the inhibitor validoxylamine A 36, which resulted in structures that can be described as 'closed' and 'open', respectively. Indeed, the so-called lid loop (G 506 -G 519 ) and side loop (Y 147 -Y 159 ) were found to undergo a significant conformational change upon ligand binding (Figure 2) 33 . Docking simulations were performed with these structures as representative scenario for family GH37 trehalases, such as the one from porcine kidney. The bacterial enzyme shares about 30% sequence identity and 40% similarity with porcine and human trehalase, but all active site residues within a distance of 4 Å from validoxylamine A 36 are fully conserved (Supplementary Figure S2). It was hypothesised that a compound sensitive to hydrolysis should find a productive docking pose in both the open and closed conformation, whereas a hydrolysis-resistant inhibitor should only be accommodated by the former.
After removal of the cocrystallised validoxylamine A 36, trehalose 1 was docked in the active site pocket of the closed structure and compared with the reference compound 36 (Figure 2, bottom). A very good overlay was observed after superposition, with most of the crucial interactions being conserved. A similar result was obtained after docking of trehalose 1 and validoxylamine A 36 in the open structure (Figure 2, top). Subsequently, the binding of various trehalose derivatives was simulated (Figure 3; Table 3).
Compound 21-a (4-O-methyltrehalose) was picked as a 4-Osubstituted representative in order to find an explanation for their low inhibitory activity and high resistance to hydrolysis. In the open form, two docking poses were found, i.e. with the methyl substituent positioned in either subsite À1 or þ1. However, neither of these fits match with the one of trehalose, and very little interactions are formed between enzyme and ligand ( Figure 3). There obviously is little space around the 4-hydroxy position of trehalose in both subsite À1 and þ1 (Figure 2), explaining why the bulkier compounds cannot take on the same pose as trehalose. Furthermore, no productive docking result could be obtained in the closed structure.
In contrast to their 4-O-substituted counterparts, 6-O-substituted derivatives show a far more interesting inhibitory profile. A distinction can be made between the ether and carbamate derivatives. For example, the 6-methoxy derivative 17-a triggered minor hydrolytic activity as well as an attractive inhibitory effect. Docking simulations confirmed our hypothesis that the derivative could find a productive conformation in both the open and closed structure ( Figure 3). Interestingly, the substitution is accommodated in subsite À1 rather than þ1, although the latter seems to offer more space (Figure 2). In contrast, substitutions containing four or more carbons are preferentially positioned in subsite þ1, although their inhibitory potential is considerably lower. These findings suggest that a greater affinity can be achieved when the substituent is sufficiently small to be placed in subsite À1.
A trend could be noticed within the 6-O-carbamate derivatives, as the inhibitory potential among these compounds clearly is highest when a phenyl group is present. For instance, complete inhibition of porcine kidney trehalase without hydrolysis was observed with compound 19-f, containing a N-phenethylcarbamoyl group. This trehalose derivative was able to find a fit in the open active site (Figure 3), while its positioning in the closed structure was predicted to have a negative binding energy (i.e. repulsion, Table 3). A strong stacking interaction is established between the compound's aromatic group and F 154 , and again, the substituent was found to be accommodated in subsite À1. Docking simulations were also performed with both other phenylcontaining carbamates 19-d and 19-e and cyclohexyl carbamate 19-c ( Figure 3). The positioning of the benzyl substituent of 19-e resembles that of 19-f. Remarkably, the phenyl carbamate moiety of 19-d is preferentially placed in the þ1 subsite instead, although the interaction with F 154 still is retained. Just as with the alkylated derivatives, the positioning of the substituent in subsite þ1 appears to lower the compound's inhibitory potential. In contrast, the cyclohexyl carbamate 19-c exhibited a different behaviour: its preferred docking poses do not match the one of trehalose, highlighting the importance of the aromatic group.

Discussion
In our work, we have evaluated a series of 4-and/or 6-O-substituted trehalose derivatives towards trehalase degradation. Generally, all compounds were resistant to hydrolysis for both tested trehalases, with the exception of 6-O-alkylated derivatives with short carbon chain length (C 1 -C 3 ). For porcine kidney trehalase, as representative of GH37 trehalases, this trend was confirmed by docking studies: the hydrolysable 6-O-substituted derivatives could find a productive fit in the closed (i.e. hydrolytically active) enzyme structure, whereas that was not the case for analogues containing larger substituents at this position or for any of the 4-substituted derivatives. This is consistent with the  Table 1).
findings of Asano et al., who pointed out that the enzyme is most tolerant towards substitutions at the 6-O-position 38 .
To evaluate whether the non-hydrolysable trehalose derivatives are still able to bind to the enzyme, inhibition experiments were performed with the trehalases from porcine kidney (family GH37) and M. smegmatis (family GH15).
From a pharmaceutical perspective, inhibitors of intestinal glycosidases, including trehalase (GH37 family), can be interesting and may be orally administered in the treatment of diabetes (type II) to regulate the absorption of carbohydrates 39 . A distinction can be made between transition state analogues and substrate analogues 40 . The former are inhibitors that optimally bind the transition state of trehalase, by mimicking the glucosyl-oxocarbenium ion which is developing during enzymatic action 35,40 . Examples are validamycin A 35, validoxylamine A 41 36 and trehazolin 35 37, which have IC 50 values in the nanomolar range for porcine kidney trehalase (Figure 4). In contrast, substrate analogues like mannotrehalose 42,43 38, 5-thiotrehalose 42  Trehalases could also be attractive drug targets in the ongoing battle against pathogenic microorganisms due to the essentiality of trehalose and its metabolic derivatives for their survival 23 . Trehalose is a building block of the cell wall in mycobacteria and corynebacteria, as it is a basic component of their glycolipids 6,10 . For example, the cell wall of the pathogen Mycobacterium tuberculosis contains trehalose-6,6 0 -dimycolate, a toxic lipid that has been identified as the main virulence factor of tuberculosis and is responsible for the low permeability of the cell wall, leading to drug resistance 10,45 . Interestingly, Shleeva and co-workers investigated the importance of trehalase activity on the resuscitation of dormant mycobacterial cells 46 . Validamycin A 35 (Figure 4), a trehalase inhibitor that mimics the transition-state, had a negative  potent inhibitors with IC 50 values down to 0.35 mM for 31. To the best of our knowledge, this is the first report of substrate analogues that inhibit the activity of M. smegmatis trehalase.
Finally, clear differences can be noticed between the two studied trehalases concerning interaction with our series of trehalose derivatives; generally, M. smegmatis trehalase is more sensitive to inhibitory effects in comparison to porcine kidney trehalase, reflecting their classification in different families (GH15 vs. GH37, respectively). Carrol and co-workers already pointed this out, as it was found that the transition-state mimic trehazolin 35,47 , a known inhibitor of porcine kidney trehalase, had no inhibitory effect towards M. smegmatis trehalase 32 (Figure 4). In our study, this effect is particularly noticeable in the performance of 4-O-substituted derivatives; several compounds (2, 21-a, 23-c, 23-d, 23-f, 29, 31) show an interesting inhibition of M. smegmatis trehalase (>90% inhibition at 20 mM concentration) but lack inhibitory effect towards porcine kidney trehalase. Notably, the trehalase hydrolysis-resistant lentztrehalose A 2 inhibited the activity of M. smegmatis trehalase completely, whereas it only had a poor effect on porcine kidney trehalase activity (1% inhibition) at the same concentration, the latter confirming previously reported data 25 . This binding selectivity is also reflected in the IC 50 values of these 4-O-substituted compounds, which are in the lower-to sub-mM range (0.35-5.27 mM) for M. smegmatis trehalase, but could not be determined for porcine kidney trehalase (>20 mM). Due to this selective inhibition profile and the hydrolytic stability, these 4- Table 3. Binding energy (kcal mol À1 ) and the possible presence of a productive conformation determined by simulated molecular docking experiments versus experimentally observed hydrolysis and inhibition.  Table 1). b Non-hydrolysable compound (no glycosidic linkage). substituted trehalose derivatives could be investigated as potential selective anti-pathogenic agents that exert a limited inhibitory effect in the human and/or mammalian gastro-intestinal tract without being easily degraded by intestinal trehalases.

Conclusions
In this study, a series of trehalose derivatives was successfully synthesised and the effects of their substitution pattern was explored towards trehalase susceptibility. All compounds were shown to be fully or strongly resistant to enzymatic hydrolysis, and may thus have a higher bioavailability than trehalose due to their increased resistance against trehalase activity in the human gastro-intestinal tract, which could be promising in eventual drug applications. Amongst all tested analogues, only trehalose derivatives bearing short alkoxyl chains on the 6-O-position were marginally hydrolysed. A number of the synthesised compounds were found to have an interesting substantial and selective inhibitory effect against M. smegmatis trehalase. Studies on the implication of these trehalose derivatives on the growth of pathogens like M. tuberculosis are ongoing.
Docking simulations in a trehalase originating from the GH37 family confirmed our experimental findings on porcine kidney trehalase. None of the 4-O-substituted derivatives could find a productive binding conformation in the open enzymatic state which supports their low inhibitory effect; they are resistant against hydrolysis as the closed formation cannot be reached. However, compounds with substituents at the 6-O-position showed a greater binding affinity in the open form, explaining their inhibition potential, while a successful fit in the closed form could only be achieved by the hydrolysable derivatives containing small 6-Oalkyl groups.
In conclusion, an exploration of 34 trehalose derivatives was performed by investigating a variety of substituents on the 4-and 6-position. Our work reveals their interaction with two relevant trehalases, i.e. one from family GH37 and one from GH15, through in vitro and in silico experiments. In this way, preliminary steps have been taken to unlock their potential use as therapeutic agents.

Disclosure statement
No potential conflict of interest was reported by the author (s