A β-Glucosyl Sterol Probe for in situ Fluorescent Labelling in Neuronal Cells to Investigate Neurodegenerative Diseases
Graphical Abstract
Aiming at tackling neurodegenerative diseases, the potential effect of β-glucosyl sterol on neuronal cells is investigated by means of fluorescent labelling of a synthetic derivative. Marked differences in neuron accumulation, metabolism, and impact on the lysosomes with respect to the un-glycosylated cholesterol derivative were observed.
Abstract
A β-glucosyl sterol probe bearing a terminal alkyne moiety for fluorescent tagging enables the investigation of the neuronal and intracellular localization of this class of compounds involved in neurodegenerative diseases. The compound showed localization in the neuronal cells, with marked differences in the uptake and metabolism leading to enhanced persistence with respect to the un-glycosylated sterol analogue. In addition, a different impact was observed towards lysosomes, with the simple sterol probe showing the enlargement of the lysosome structures, while the β-glucosyl sterol was less capable to alter the morphology of this specific organelle.
Introduction
While the role of sterols, in particular cholesterol (Chol), has been widely characterized in human physiology and pathophysiology, much less is known for the role of glucosyl-sterols (GlcSt, Figure 1). This group of natural compounds is characterized by covalent conjugation between the 3-hydroxyl group of the sterols (cholesterol or phytosterols) and the C1’ anomeric carbon of d-glucose through a O-(1’→3)-α- or β-configuration.1 GlcSt are found in plants, in fungi and in bacteria as well as in humans.2 In particular, cholesterol-α-d-glucoside (α-GlcChol) found in humans is produced by Helicobacter pylori, known to colonize the gastric tract and involved in severe pathologies such as ulcer and cancer, by means of the specific enzyme cholesterol-α-glucosyltransferase hp0421.3 Conversely, cholesterol-β-d-glucoside (β-GlcChol) is formed in cells by glucosylation of Chol by two ubiquitous enzymes termed glucosylceramidases: the lysosomal GBA1 and the non-lysosomal GBA2,4 which transfer glucose from glucosylceramide to Chol5 leading to the formation of β-GlcChol and ceramide. It was proved that defects in the lysosomal enzyme GBA1, caused by mutations in the gene GBA1, are associated to the lysosomal storage disorder Gaucher Disease (GD) and to Parkinson's disease (PD),6 which is one of the most common neurodegenerative disorders worldwide.7 Moreover, mutations in the gene GBA2, encoding the non-lysosomal glucocerebrosidase, cause a form of Hereditary Spastic Paraplegia (HSP) with ataxia. In both cases, altered levels of β-GlcChol were found in cells and tissue from animal models and patients.2

The molecular structures of the most common natural glucosyl sterols and the molecular probe β-1, analogue of cholesterol-β-d-glucoside (β-GlcChol).
Interestingly, sitosterol-β- d-glucoside (β-GlcSito), which is characterized by a structure similar to β-GlcChol (Figure 1), has been related to a complex neurological disease known as amyotrophic lateral sclerosis-parkinsonism dementia complex (ALS-PDC).8 β-GlcSito is largely present in the seeds of plant Cycas micronesica, usually included in the diet of the Chamorro inhabitants of islands of the Guam Archipelago,9 where a high incidence of ALS-PDC has been observed. All these considerations suggest that β-GlcSt may play an important role in the aetiology of motor neurodegenerative disorders. However, the mechanisms by which β-GlcChol is transferred or compartmentalized in neurons and β-GlcChol physiological function in neuronal cells are still largely unknown.
Thanks to the recent achievements in bio-orthogonal chemical approaches10 such as click chemistry,11 the sensitive and specific detection of modified biomarkers containing azido groups or terminal alkynes has become possible. Although examples of synthetic Chol analogues with terminal acetylenic side chains as biomarkers for fluorescence molecular imaging are known, only one example is known12 for the α anomer α-GlcChol bearing a terminal azide moiety for the study of Helicobacter pylori infections.13
In order to disclose possible localization and effects of β-GlcChol in animal tissues to shed light on potential effects in neurodegenerative diseases, we synthesised the novel molecular probe β-GlcSt (β-1, Figure 1) bearing a terminal alkyne of the chain in position 17 of the sterol unit, that can be functionalized with different commercially available azido fluorophores. The target compound was tested in retrospective click chemistry protocols12e directly in neuronal cells, avoiding any preliminary functionalization that could profoundly alter the physical properties, cell permeability and trafficking of steroid analogues. This allowed us to observe β-1 distribution in neuronal cells and to compare the outcome when neurons were treated with 8, the un-glycosylated form of β-1 (Scheme 1). We observed a more rapid uptake and depletion of 8 compared to β-1, while the latter showed a lower uptake but higher persistence in neurons over time. The impact of both molecular probes on lysosomal compartment was compared as well, in consideration of the importance of cholesterol for lysosomal function and the suspected role of glucosyl sterols in neurodegenerative diseases. Interestingly, we observed that 8 caused the enlargement of the lysosomal compartment. Conversely, β-1, even leading to higher persistence, did not affect the morphology of lysosomes over time and coherently showed a reduced localization to this intracellular compartment.

Synthesis of β-1 probe. Reagents and conditions: (a) THP, PTSA, Tol, react in situ in step c; (b) TBDPSCl, imidazole, DMF 98 %; (c) 2 a, Red-Al®/Tol, 98 %; (d) 2 b; NaBH4, MeOH 98 %; (e) TsCl, KOH, Et2O, 98 %; (f) nBuLi, TMSCl, THF 96 %; (g) nBuLi, TIPSCl, THF, 98 %; (h) 2 a, 2 b, 2 c or 3; NaH; (solvent yield see Table 1 in Supporting Information); (i) PTSA, MeOH-DCM (64 % in two overall steps); (l) NH4F, MeOH, 70 %; (m) TBAF, THF (74 %); (n) BzCl, Py, 94 %; (n) HBr-AcOH DCM, 97 %; (p) ZnBr2, DCM MS 3 Å (32 %); (q) TBAF, THF (54 %); (r) MeONa, MeOH, (49 %).
Results and Discussion
Synthesis of Molecular Probe β-1
The target compound β-1 was obtained following the sequence of reaction reported in Scheme 1, starting from the commercially available building blocks dehydroepiandrosterone (DHEA) 2 as sterol core, 5-hexyn-1-ol 3 or 6-iodohex-1-yne 4 as spacers, and d-glucose 5.
For the preparation of the proper cholesteryl unit, 2 was derivatized with different protecting groups, either THP or TBDPS as function of the final target molecular probe, affording acetal 2 a or silyl ether 2 b. 2 a was reduced with Red-Al®14 to afford 2 c, while 2 b was reduced to 2 d with NaBH4,13b in both cases with almost complete stereoselectivity in position 17.
The synthesis of ethers 6 a–e requested strong efforts to optimize the experimental conditions to ensure acceptable product yields. In fact, the steric hindrance provided by Me20 heavily hampers the O-alkylation, as previously reported for the reaction with 1-azido-6-bromohexane as electrophile.13b Several experimental conditions were tested using alkyne-unprotected electrophilic units with two different leaving groups, iodide 4 or tosylate 3 a15 with different bases, solvents, temperature, and reaction time (see supporting information) observing low yields for the desired product 6 a. This was also partially due to the high volatility of 4, that was consequently replaced also because of the conspicuous base-promoted isomerization of the terminal alkyne to allene 7 (see supporting information). In order to prevent this side reaction, we initially protected the acetylenic moiety with TMS observing an improvement in the yields of 6 b although not exceeding 57 % due to partial loss of the TMS protecting group. Finally, TIPS-protected tosylate 3 c16 turned out to properly react with 2 c or 2 d obtaining protected 6 c and 6 d in up to 65 % isolated yield. Product 6 e maintaining the TIPS on the acetylenic unit was obtained from 6 c by acid hydrolysis with PTSA and from 6 d by reaction with ammonium fluoride (Scheme 1). In order to ascertain the role of the glucose unit in β-GlcChol, we prepared compound 8 by concomitant deprotection of TBDPS and TIPS from 6 d with TBAF, as a control compound lacking the glucose unit with cholesterol-like structure endowed with the terminal acetylenic unit for fluorescent tagging.
The glycosidic synthon 5 b was prepared in multigram scale by exhaustive benzoylation17 of glucose 5 leading to 5 a followed by selective α-bromination at the anomeric position.17, 18
To ensure high β stereo-selectivity, the glycosylation reaction was performed according to the method described by Murakami.19 This protocol was preferred to other strategies, which require per-O-benzoylated-glycosyl trichloroacetimidate as glucose synthon,20, 18 that needs two more synthetic steps from 5 b, or the use of expensive stoichiometric AgOTf21, 17 or Ag2O.21a, 22 The Murakami protocol applied to 5 b with 6 e in the presence of ZnBr2 under strictly anhydrous conditions with molecular sieves in DCM afforded the product 9 a in 32 % yield after flash chromatography purification (Scheme 1). The complete β stereoselectivity was confirmed by the chemical shift value of the doublet corresponding to the anomeric proton H1’ at 4.94 ppm with a J=7.9 Hz. These data are consistent with values reported for tetra-O-benzoyl-β-d-glucopyranoside derivates of cholesterol.23 Subsequently, the final target product β-1 was obtained by TIPS deprotection with tetrabutylammonium fluoride13b and benzoyl deprotection from the glucose moiety with catalytic sodium methoxide.24 The identity of the compound β-1 was confirmed by 1H, 13C{1H} NMR and HRMS (see Supporting Information). In particular the 1H NMR spectrum confirmed the retained β configuration after final deprotections of the anomeric carbon atom of the glucose. In fact, the resonance at 4.22 ppm attributed to the anomeric H atom and its coupling constant of 7.8 Hz are characteristic for the β stereoisomer (See Supporting Information).25
Interaction of β-1 and 8 with Neurons and the Lysosomal Compartment
The target compound β-1 was tested in primary mice neuronal cells to verify its capability of entering the cells. Primary mice neurons at day in vitro (DIV) 14 were treated with 10 μM β-1 for 5, 15 or 60 minutes and staining with azide-Alexa488 for imaging was performed after extensive washes and fixation with 4 % PFA, following a previously developed protocol.12d Cells were then imaged using confocal microscopy (Figure 2a). Neurons average intensity and intensity distributions were measured using Fiji and are shown in Figure 2b and 2c for each experimental condition. Alexa488 signal was present also in untreated neurons (NT), likely due to non-specific staining or autofluorescence. At 5 minutes, the signal was comparable to the one measured for NT neurons, suggesting that the amount of β-1 able to enter the cells at this time point is very limited. According to these experiments, β-1 shows a trend in the increase over time (despite being non-significant at 15’), reaching a significant difference in the intensity value at 60’ as compared to the other time point. No changes were observed in the signal from untreated cells over time, as expected.

A. Representative confocal microscopy images showing the azide-Alexa488 staining of untreated neurons (NT), of neurons treated for 5 minutes (5’), 15 minutes (15’) or 60 minutes (60’) with β-1. B. Average fluorescence intensity of neuronal cells in the conditions previously described. C. Fluorescence intensity distributions of each pixel in neuronal cells in the conditions previously described (ns=non-significant, **<0.01; NT vs 5’ p>0.999, NT vs 15’ p=0.5203, NT vs 60’ p=0.0015, 5’ vs 15’ p=0.7190, 5’ vs 60’ p=0.0013, 15’ vs 60’ p=0.3093, Kruskal–Wallis test, Dunn's multiple comparison test). Experiments were performed at least twice and n>5 cells were analyzed. Scale bar 10 μm.
To further understand the entry capacity of β-1 within neurons, we treated the cells with control molecule 8, at the same concentration and compared the results obtained for the same time points (Figure 3a and 3b). The analysis performed to quantify the average fluorescence intensity of the stained cells allowed to establish that the two molecules are both able to enter the cells (Figure 3b), but to a different extent. While at 15 min fluorescence signal of 8 in neurons was the highest, it declined at 60 min. Conversely, the same did not occur for β-1, which presented a lower fluorescence signal at 15 min (even if non-significant) when compared to 8-treated cells at the same time point. This result suggests that the two molecules have different capacity to enter the cells and/or to be processed or metabolized. One possible explanation for this difference is that the β-1 molecule is less prone to partition into the membrane being more water soluble because of the presence of the glucose moiety, which makes the molecule larger and possibly more hydrophilic. Another possibility is that the cholesterol transport proteins regulating Chol homeostasis in cells26 play a different role in the regulation of β-1 transport leading to different outcomes when comparing 8 and β-1.

A. Representative confocal microscopy images of neurons stained with azide-Alexa488 after treatment with 8 or β-1 for 15’ or 60’ (click labelling) compared with time 0, and with the antibody against the lysosomal marker LAMP1. For the nuclei, Hoechst was used. B. Violin plot showing the average lipid intensity at 15’ and 60’ with the two treatments (ns=non-significant, *<0.05, **<0.01, ***<0.001, ****<0.0001; 8–15′ vs. 8–60′ p<0.0001; 8–15′ vs. β-1 −15′ p=0.072; 8–15′ vs. β-1 −60′ p=0.0013; 8–60′ vs. β-1 −15′ p=0.0043; 8–60′ vs. β-1 −60′ p =0.0764; β-1 −15′ vs. β-1 −60′ p>0.999, Kruskal-Wallis test, Dunn's multiple comparison test). C. Violin plot of the lysosomal integrated fluorescence intensity normalized by neuronal area at 15’ and 60’ with the two treatments (ns=non-significant, *<0.05, **<0.01, ***<0.001, ****<0.0001; 8–15′ vs. 8–60′ p=0.016; 8–15′ vs. β-1 −15′ p>0.999; 8–15′ vs. β-1 −60′ p=0.579; 8–60′ vs. β-1 −15′ p=0.0034; 8–60′ vs. β-1 −60′ p =0.834; β-1 −15′ vs. β-1 −60′ p=0.187, Kruskal–Wallis test, Dunn's multiple comparison test). D. Violin plot of the Pearson's coefficient between lipid fluorescence signal and LAMP1 signal per image at 15’ and 60’ with the two treatments (ns=non-significant, *<0.05, **<0.01, ***<0.001, ****<0.0001; 8–15′ vs. 8–60′ p<0.0001; 8–15′ vs. β-1 −15′ p=0.0011; 8–15′ vs. β-1 −60′ p=0.0003; 8–60′ vs. β-1 −15′ p=0.400; 8–60′ vs. β-1 −60′ p =0.794; β-1 −15′ vs. β-1 −60′ p>0.999, Kruskal–Wallis test, Dunn's multiple comparison test). Scale bar 10 μm.
To ensure that the two molecular probes 8 and β-1 are sufficiently stable during the studies with neuronal cells and do not undergo interconversion, we analysed Folch lipid extraction samples of neuron-like cells treated with 8 and β-1 after different times using UHPLC-HRMS. In the lipid extract of cells treated with only 8, no traces of β-1 were detected. Similarly, in the lipid extract of cells treated with only β-1, no traces of 8 were detected (see Supporting Information). These results confirm that the molecular probes 8 and β-1 were basically rather stable in the time period of the experiments. This is in agreement with the low rate of glucosyltransferase activity of glucocerebrosides as reported in the literature.27
Lysosomes are crucial for the regulation of Chol homeostasis, which is key for the functioning of endo-lysosomal compartment. Chol sensing at the lysosomes regulates cell signalling, proliferation, and autophagy.28 Moreover, one of the two enzymes involved in the metabolism of β-GlcChol is GBA1, which is localized at the lysosomes. Thus, we decided to investigate whether 8 and β-1 reach and affect the lysosomal compartment in a comparable manner, or if they behave differently.
To evaluate this aspect, we stained neurons with the antibody against the lysosomal marker LAMP1 (Figure 3a). This approach allowed to evaluate the average amount of lysosomes per neuron: to obtain this result, we quantified the intensity of the lysosomal signal above a certain threshold and normalized the value by the neuron area (Figure 3c).
Moreover, we also quantified the colocalization between each of the two lipid species β-1 and 8 with the lysosomal marker (Figure 3d). Interestingly, the lysosomal compartment was enlarged when treating neurons with 8 for 60 min, compared to the 15 min treatment. The same did not happened when treating cells for the same amount of time with β-1, thus suggesting that they do not impact similarly on lysosomes (Figure 3c). Coherently, when evaluating the level of colocalization between the lipids and the LAMP1-positive intracellular compartment by calculating the correlation between the two fluorescence signals via Pearson's coefficient (Figure 3d), we found that localization of 8 to the lysosomes was reduced upon longer incubation times (60 min), while it was the same at 15 min or 60 min for β-1. This result may suggest that the two lipids undergo different transport routes to reach lysosomes and to leave the organelles.
Given the limited information on the activity of β-GlcSt in neurons and considering the crucial role that sterols have in lysosomal activity and the importance of lysosomes for sterols homeostasis,26 these experiments show for the first time that both glucosyl sterol β-1 and simple sterol 8 can enter neurons, but impact differently on the lysosomal compartment. In particular, while 8 enlarges the lysosomes and is exported at 60 min, likely via its specific protein transporters at the lysosomal membrane (NPC1, NPC1 and LIMP-2), β-1 does not seem to induce significant changes in the lysosomal compartment over time and the β-1 levels at 15 min and 60 min remain stable. These results underline the marked difference in transport, metabolism, and lysosomal biology in neurons imparted by the β-glucosyl moiety on sterols.
Conclusions
In conclusion, herein we described the synthesis and the application in neurons of a versatile molecular probe β-1 for fluorescence imaging mimicking the structure of β-GlcSt, which is considered an important target compound potentially involved in neurodegenerative diseases. Compound β-1 was obtained from three synthons (glucose, a sterol-like unit, and a spacer with alkyne moiety for click conjugation with azide fluorescent units), after optimization of the connection between the sterol unit and the alkyne spacer. β-1 and the corresponding un-glycosylated derivative 8 showed marked differences in neuron accumulation and metabolism. In particular, the glycosylated derivative showed lower accumulation but longer time persistence in neuronal cells and a reduced impact on the lysosomes with respect to 8. The different behaviour observed will spur further investigation on compounds 8 and β-1 to ascertain the complex role of glucosyl sterols in neuronal biology and lysosomal function, especially in relationship to neurodegenerative diseases, but also in the function of other membranous organelles, such as mitochondria and endoplasmic reticulum.
Supporting Information
Detailed experimental procedures, characterization data, 1D and 2D NMR spectra of all novel synthesized compounds. The authors have cited additional references within the Supporting Information.29-32
Acknowledgments
FF and AS acknowledge Università Ca’ Foscari di Venezia and NP acknowledges Università degli Studi di Padova for financial support. AS and GB are grateful to Dr. A. Bonetto for HPLC-HRMS analyses.
Conflict of interests
The authors declare no conflict of interest.
Open Research
Data Availability Statement
The data that support the findings of this study are available in the supplementary material of this article.