A pH-Sensitive Double Chromophore Fluorescent Dye for Live-Tracking of Lipophagy
Graphical Abstract
Live monitoring the fate of lipid droplets (LD) within cells! To visualize and quantify lipophagy in an easily applicable, non-toxic and protein label-free approach by means of fluorescence microscopy, a unique dye called Lipo-Fluddy-1 was developed. This synthetic dye is characterized by a high lipophilicity and contains two chromophores, one of which only fluoresces at low pH. Its usefulness was demonstrated by the flow cytometry investigation of lipophagy in several cell lines.
Abstract
Lipid droplet (LD) degradation provides metabolic energy and important building blocks for various cellular processes. The two major LD degradation pathways include autophagy (lipophagy), which involves delivery of LDs to autolysosomes, and lipolysis, which is mediated by lipases. While abnormalities in LD degradation are associated with various pathological disorders, our understanding of lipophagy is still rudimentary. In this study, we describe the development of a lipophilic dye containing two fluorophores, one of which is pH-sensitive and the other pH-stable. We further demonstrate that this “Lipo-Fluddy” can be used to visualize and quantify lipophagy in living cells, in an easily applicable and protein label-free approach. After estimating the ability of compound candidates to penetrate LDs, we synthesized several BODIPY and (pH-switchable) rhodol dyes, whose fluorescence properties (incl. their photophysical compatibility) were analyzed. Of three Lipo-Fluddy dyes synthesized, one exhibited the desired properties and allowed observation of lipophagy by fluorescence microscopy. Also, this dye proved to be non-toxic and suitable for the examination of various cell lines. Moreover, a method was developed to quantify the lipophagy process using flow cytometry, which could be applied in the future in the identification of lipophagy-related genes or in the screening of potential drugs against lipophagy-related diseases.
Introduction
The biosynthesis and degradation of macromolecules within the cell are rigorously regulated, as either an excessive accumulation or a deficiency can cause physiological dysregulation and trigger pathological conditions.1 Lipid homeostasis is of particular importance in this context, as lipids are essential building blocks for all living organisms and are involved in a wide range of functions, including energy storage, signaling or the formation of biomembranes.2 Neutral lipids are stored within LDs, which represent evolutionarily-conserved organelles contained in nearly all types of cells. LDs consist of a hydrophobic core containing triacylglycerols and cholesterol esters surrounded by a phospholipid monolayer.3 The degradation of LDs provides metabolic energy for several cellular processes as well as important building blocks for new cellular compartments and can proceeds through a process called lipophagy. This selective form of autophagy starts via the formation of an autophagosome (surrounded by a phospholipidic bilayer membrane), which then fuses with a lysosome to form an autolysosome with increased acidity (pH 4–5), leading to the digestion of the autophagosome content (Figure 1a).4 While dysregulated LD degradation has recently been associated with various metabolic diseases such as obesity,5 type 2 diabetes mellitus,6 alcoholic7 and non-alcoholic8 fatty liver, as well as neurodegenerative diseases such as hereditary spastic paraplegia (HSP),9 Alzheimer's disease,10 and Parkinson's disease,11 our understanding of lipophagy and its regulation still remains rudimentary.12
An established method for studying selective autophagy in living cells makes use of organellar-targeted fluorescent reporters, comprising a target protein fused to the two fluorescent markers mCherry and GFP.13 While the red fluorescence of mCherry is constant throughout the pH range of the cell, GFP shows no fluorescence at a pH of 4–5, which is characteristic of autolysosomes. This allows to track the movement of organelles into lysosomes, which is correlated to a decrease in the green fluorescence. Due to the stoichiometric coupling of the two dyes, the changing ratio of fluorescence allows a quantitative detection of the rate of autophagy. This elegant method has been applied to study the autophagy of proteins and organelles.14 An extension of application to LDs, however, is limited due to a lack of proteins within the lipid core. While proteins in the LD membrane can be used to install fluorescent protein labels, their mode of degradation may be different from the degradation of lipids,15 and lipids cannot be genetically tagged with protein labels.
To develop a method for the study of lipid degradation via lipophagy, we envisioned that a synthetic lipid containing both a pH-stable and a pH-sensitive fluorophore might solve the above-mentioned problems. In this work we designed such a lipophilic double chromophore dye (which we named “Lipo-Fluddy”), in which a green pH-sensitive dye unit is linked to a red pH-stable dye (Figure 1b). Lipophilic side chains are required to ensure that the molecule behaves like a lipid and penetrates LDs. With such a dye, it should be possible to monitor lipophagy under the microscope and quantify its flux. The covalent linkage of the chromophores should enable ratiometric detection of the fluorescence signals. This in turn would allow normalization and thus the identification of cells with abnormalities in LD degradation compared to healthy cells. In addition, such a Lipo-Fluddy could be used to identify gene mutations that alter lipophagy or to search for drugs that enhance or inhibit this process. Due to the small molecular size of the Lipo-Fluddy, the method could be applied to any type of cell without the need for prior genetic modification to introduce fluorescent markers. We herein report the successful development of a first Lipo-Fluddy dye suitable for the live fluorescence tracking of lipophagy.
Results and Discussion
Selection of Suitable Chromophore Types
Notably, several challenges accompanied the search for a suitable Lipo-Fluddy structure. Firstly, its lipophilicity had to be sufficiently high to ensure entry into LDs, so an efficient method for predicting the lipophilicity of a potential target structure was needed. The second challenge was to choose the fluorophores in such a way that energy transfer processes such as Förster resonance energy transfer (FRET) should not occur, because otherwise only the red emission would be visible.
A modified (deoxy) rhodol system was chosen as the pH-switchable fluorophore, which had originally been developed by More et al.16 for lysosomal staining, because this system shows fluorescence only at the low pH of lysosomes and autolysosomes (Figure 2). By altering the groups R and R’, respectively, a fine-tuning of photophysical and chemical properties could be achieved.
As the pH-stable fluorophore, a BODIPY moiety was selected because these dyes are known for their high brightness, stability, lipophilicity and membrane permeability.17 Additionally, their spectral properties can easily be altered (and even shifted to the deep-red range) by the introduction of aromatic substituents (R), which at the same time would serve as lipophilic side chains. Ideally, the resulting lipo-fluddy would stain the LDs red, with additional green fluorescence becoming visible as soon as lipophagy occurs.
Lipophilicity Prediction
To predict whether a new dye would enter LDs, we applied computational tools to calculate the logarithmic n-octanol/water partition coefficient (Log P) and the total polar surface area (TPSA) from the chemical structure.18, 19 As a reference, we first analyzed a series of known dyes established as LD markers, which due to their high lipophilicity automatically localize in the hydrophobic core of LDs.20 From these calculations (see Table S1 in the Supporting Information) we deduced that the structure of a Lipo-Fluddy should possess a calculated log P value of >4 and a TPSA below 80 to ensure sufficient lipophilicity. Keeping this in mind, we investigated which type of alkyl and aryl side chains in our selected fluorophores would be suitable. An example of this approach is depicted in Figure 3, where the introduction of n-octyl or styryl sidechains to a BODIPY core leads to a clear increase of the predicted Log P value, to ensure the required lipophilicity of the dye, while the TPSA is not affected.
Notably, the introduction of styryl side chains also leads to a red shift in BODIPY fluorescence beneficial for the use of this dye in combination with the green deoxy-rhodol fluorophore. As explained in more detail below, BODIPY 3 proved to be an extremely useful and previously undescribed marker for the selective staining of LDs in the deep red spectral region. This demonstrates the utility of this predictive method in the search for new LD markers. The method was also used in the search for a suitable substituent and linker at the deoxy-rhodol fluorophore (see Supporting Information for details), resulting in the suggestion of a very promising general structure (4) for the Lipo-Fluddy (Figure 4). For these predictions, the cyclic form of the deoxy-rhodol substructure (compare Figure 2) was chosen because it should be present at neutral pH in both the cytosol and LDs. Upon protonation in the autolysosome, opening of the cyclic ether would lead to a considerable reduction in lipophilicity (see Supporting Information, Figure S7). However, a high lipophilicity of the dyes is no longer required once degradation of the LDs has begun. Instead, the lower lipophilicity of the ring-opened form might lead to a lower permeability through the autolysosomal membrane and a higher probability that the dye will remain inside. Increased polarity in acidic medium may thus even be advantageous for a Lipo-Fluddy.
Identification of a Suitable Chromophore Combination
In the search for a suitable fluorophore combination, several pH-sensitive (deoxy-rhodols) and pH-stable (BODIPY) dyes were synthesized (see Supporting Information), and their absorption and fluorescence properties were analyzed by UV/VIS and fluorescence spectroscopy (Figure 5). The synthesis of deoxy-rhodols 5 and 6 was analogous to the methodology described by More et al.,16 but using a MOM protecting group at the phenol residue to allow the later introduction of a linker at this position. An n-hexyl-substituent was introduced at the aniline moiety to increase lipophilicity. Notably, while the introduction of an additional N-methyl group (rhodol 6) was predicted to further increase lipophilicity, it was also accompanied by an unfavorable bathochromic shift, which makes this compound less suitable in combination with a red fluorescent dye. Therefore, rhodol 5 was chosen as the pH-switchable dye unit. As pH-stable fluorophore units, the symmetric BODIPY dyes 7, 8, 9 and 10, which all contain a methyl benzoate moiety that can later be used for coupling to the linker, were synthesized following the methods described by Frank et al.21, 22 As expected, the nature of the substituents adjacent to the N-atoms were found to significantly influence the emission wavelengths (Figure 5). Compared to the phenyl-substituted compound 7, the p-fluorophenyl-substituted analogue 8 showed no advantages and was therefore not considered further. The other three BODIPY derivatives, on the other hand, were further investigated with regard to their photophysical compatibility with the rhodol 5.
To first confirm the pH-switch ability of 5, its fluorescence was measured in phosphate buffer at both pH=7.4 and pH=4.5. As expected, emission was only detectable in the acidic medium (Figure 6, left). Performing the same experiment with the BODIPYs proved difficult due to their low solubility in aqueous medium. Therefore, an alternative solvent pair was employed, consisting of ethanol (as a pH-neutral solvent), and a 1 : 1 mixture of acetic acid and ethanol (to mimic the lysosomal pH). In these solvents, the pH-switchable rhodol 5 showed virtually the same behavior as that in the aqueous system (Figure 6, right).
In a similar set of experiments using either ethanol or acetic acid/ethanol (1 : 1) as a solvent, the BODIPY dyes (7–10) showed a more or less pH-independent fluorescence behavior with no differences in their emission wavelengths. Only in the case of 7 the intensity decreased markedly at lower pH (Figure 7).
To pre-investigate whether energy transfer processes (such as FRET) could pose problems upon combining two different fluorophores in a single Lipo-Fluddy molecule, solutions of rhodol 5 and one of each of the possible BODIPY units (7, 9, and 10) were measured, again in ethanol and a 1 : 1 mixture of acetic acid and ethanol (see Figure 10, left). In the acidic medium, the characteristic fluorescence signals (emission bands) of both fluorophores were visible for all three combinations. However, in the case of 7 and 10, excitation of the green dye (500 nm) also resulted in the emission of the respective red dyes. This can be explained by analyzing the absorption spectra of the BODIPY dyes with the fluorescence emission spectrum of the rhodol 5 (Figure 8). Here, a distinct overlap of the absorption bands of 7 and 10 with the emission band of 5 becomes apparent, while the overlap is much smaller for the BODIPY 9 (shown in blue).
Nonetheless, coupled molecules were synthesized corresponding to all three combinations (see Supporting Information), resulting in the fluorescent double dyes Lipo-Fluddy-1, Lipo-Fluddy-2 and Lipo-Fluddy-3 (Figure 9).
The fluorescence behavior of these double chromophore dyes was then investigated in the same fashion as before by measuring the emission upon excitation at the respective wavelengths of both fluorophores in ethanol and a 1 : 1 mixture of acetic acid and ethanol. The results were compared with the results of the combinations (mixtures) of the separate dyes (Figure 10, right). Much to our satisfaction, Lipo-Fluddy-1 exhibited the expected behavior, as irradiation at 500 nm resulted in a strong green fluorescence only in the acidic medium, while the red fluorescence proved to be pH-independent (Figure 10a). Nevertheless, a certain amount of red emission was now visible during excitation at 500 nm at a lower pH value, which was not the case with the non-coupled dyes. This indicates that some energy transfer takes place, possibly caused by the closer proximity of the chromophores after covalent linkage. However, the change in green fluorescence (and correspondingly the ratio of green to red signal) from a pH-neutral to an acidic medium is quite significant, so that Lipo-Fluddy 1 appears to be well suited for the desired application. In contrast, no green fluorescence could be detected in the emission spectra of Lipo-Fluddy-2 and Lipo-Fluddy-3 (Figure 10b/c), probably as a consequence of a pronounced energy transfer in these cases. Overall, these results indicate that even a small energy transfer between the chromophores cause problems as soon as they are coupled. The absorption band of the low-energy fluorophore must therefore not overlap with the emission of the higher-energy fluorophore, as otherwise only one emission is visible. Thanks to the low overlap of dyes 5 and 9, the energy transfer between the fluorophores within Lipo-Fluddy 1 is negligible, making this compound a most promising candidate for LD imaging applications.
Improved Synthesis of Lipo-Fluddy-1
For the further studies, we elaborated an improved synthesis of Lipo-Fluddy-1, starting with the optimization of the preparation of building block 16 corresponding to BODIPY 9 (Scheme 1).
In contrast to Frank et al.,21 who introduced substituents at the BODIPY core at a later stage through cross-coupling, we condensed methyl 4-formylbenzoate (12) directly with 2-styrylpyrrole 13.23 The crude dipyrromethane 14 was oxidized with chloranil followed by the reaction with boron trifluoride to yield BODIPY 9 in 34 % over 4 steps, which is a significant improvement over the original method which yielded only 3 % of 9 over 6 steps (see supporting information, Figure S2). Methyl ester hydrolysis of 9 finally afforded BODIPY 16 in 61 % yield.
The optimized synthesis of the deoxy-rhodol moiety 23 (Scheme 2) avoids a temporary MOM-protection (see Supporting information) by direct and early introduction of an ethylene glycol linker unit. Thus, after reacting fluorescein 17 with an excess of bromide 18 under SN2 conditions, the resulting product 19 was reduced with LiAlH4 and re-oxidized with chloranil to yield the deoxy-rhodol 20 in high yield. Triflation of the phenol function afforded 21 which was then converted to the aminated product 22 by Pd-catalyzed coupling with n-hexylamine.24 Cleavage of the TBS protection group with TBAF afforded the alcohol 23 which was finally esterified with the BODIPY acid 16 under Steglich conditions to give Lipo-Fluddy-1 in improved overall yield (19 % over 7 steps).
Lipo-Fluddy-1 Staining Colocalizes with LDs and LC3-Positive Autophagosomes
We next investigated the behavior of Lipo-Fluddy-1 in live microscopy using HeLa cells. Incubation with Lipo-Fluddy-1 (10 μM) produces distinct signals when imaging was performed with the canonical settings for the green (488 laser and filter 525/50) and deep red (640 laser and filter 705/90) channels according to the absorption and emission spectra previously measured in vitro (Figure 11). Surprisingly, the green signal never overlapped with the deep red, suggesting that the latter is quenched in the acidic aqueous compartment of autophagosomes and lysosomes.
To validate this hypothesis, we measured the fluorescence of Lipo-Fluddy-1 in vitro in different solvents (PBS, ethanol and oleic acid) under normal and acidic conditions (pH<6) to mimic the environment of different cell compartments. These experiments showed that while the green signal is present in all solvents under acidic conditions, the deep red signal is only present in ethanol and oleic acid, while it is lost in an aqueous environment, probably as a consequence of aggregation (Figure S9). To further demonstrate the specificity of Lipo-Fluddy-1 in staining the desired compartments, HeLa cells were incubated with Lipo-Fluddy-1 and fixed with 4 % PFA. This treatment brings the pH of the sample to 7.4, completely abolishing the signal coming from the green part of the fluorophore. The cells were further stained with the commercially available BODIPY 493/503, a marker of LDs (Figure 12). The deep red signal now completely overlapped with the green signal from BODIPY 493/503, demonstrating that Lipo-Fluddy-1 does indeed stain LDs. To further confirm that the deep red signal originated from LDs, we treated the cells with oleic acid, a treatment known to increase the size and number of LDs. And indeed, the deep red colored structures increased significantly in size and number after exposure to oleic acid. (Figure S10).
To demonstrate the colocalization of the green signal with autophagosomes, MAP1LC3 A, a gene encoding the well-known LC3 autophagosome marker, was cloned into a plasmid carrying a gene encoding a blue fluorescent reporter protein (BFP) (Supplementary Material mTagBFP2-LC3). Cells transfected with the construct and treated with Lipo-Fluddy-1 show colocalization of the green signal with BFP-LC3, indicating that the green signal from Lipo-Fluddy-1 originates from the acidic environment of the LC3-positive autophagosome. As expected, not all blue autophagosomes stain green, because at a given time, not all autophagosomes would be involved solely in lipophagy, but also in other autophagic processes (Figure 13).
Notably, we proved that Lipo-Fluddy-1 treatment does not induce cell death, as measured by monitoring PARP cleavage by caspases. Moreover, it also does not increase the levels of the chaperon BiP, a readout of endoplasmic reticulum stress (Figure S11).
Lipo-Fluddy-1 Enables the Monitoring of Lipophagy Using Flow Cytometry
To test whether the Lipo-Fluddy-1 could be used to quantitatively monitor lipophagy in living cells using flow cytometry, we induced lipophagy by nutrient deprivation incubating the cells with EBSS medium for 4 hours. Indeed, after 4 hours of starvation in EBSS, the number of cells showing green dots increased (Figure S10). As a negative control, the induction of lipophagy was prevented by treatment with bafilomycin,25 which blocks the fusion of autophagosomes and lysosomes and prevents lysosomal acidification. Thus, we could show that Lipo-Fluddy-1 can be successfully used in flow cytometry (Figure 14). As previously reported26, 27 starvation not only induces lipophagy and general autophagy (Figure 14), but also increases LD production as shown by the increased cell count detected with the 640 nm laser (deep red signal) (Figure 14B).
Lipophagy inhibition with bafilomycin after EBSS starvation completely abolishes the green signal without affecting the level of the deep red signal. In addition, quadrant analysis, where each cell is visualized as a dot in a scatter plot, was performed to plot the deep red over the green signal to calculate the percentage of cells undergoing lipophagy during starvation. Therefore, the scatter plot of all the cells analyzed was divided into four regions, with the lower left representing the unstained cells, the lower right representing the stained cells that are not undergoing lipophagy and the upper right representing that are actively doing lipophagy (Figure 15).
As expected, nutrient deprivation with EBSS for 4 hours leads to a significant acidification of the LDs, a phenotype that is completely rescued by bafilomycin (Figure 16).
To further demonstrate that Lipo-Fluddy-1 indeed monitors an autophagic process, we used a mouse embryonic fibroblast (MEF) cell line knockout for the key autophagy regulator ATG5, which is essential for both autophagosome formation and further fusion with the lysosome.28 Interestingly, while WT MEFs activate lipophagy similarly to the HeLa cell line, depletion of ATG5 prevents the process both under normal conditions (Figure 17) and after oleic acid loading (Figure S12). Thus, Lipo-Fluddy-1 can be successfully used in flow cytometry to monitor lipophagy quantitatively, which will be useful for genetic or drug screening applications.
Conclusions
We have described the design, synthesis and spectroscopic characterization of the unique lipophilic double chromophore dye Lipo-Fluddy-1, which is the first dye of its kind to enable monitoring of lipophagy, i. e. the degradation of lipid droplets (LD) in living cells by means of fluorescence microscopy without the need for genetically targeted overexpression of LD coat proteins. The elaborated synthesis of Lipo-Fluddy-1 proceeds in 19 % overall yield (7 linear steps) and allows structural variations. We have also shown that Lipo-Fluddy-1, which exhibits no cellular toxicity, is suitable for LD imaging in HeLa and MEF cell lines and can even be used in high-throughput procedures such as flow cytometry. Using this methodology, genetic screening should allow the identification of new genes involved in the process of lipophagy that may be dysregulated in diseases related to lipid metabolism, such as non-alcoholic fatty liver disease. In the future, the methodology could also be used in the screening of compound libraries to identify molecules that affect lipophagy as potential therapeutics for lipophagy-related diseases.
Experimental Part
All experimental details are given in the Supporting Information.
Supporting Information
The authors have cited additional references within the Supporting Information (Ref. [29,30,31,32]).
Acknowledgments
This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - Project-ID 411422114 – GRK 2550 (RELOC) and by the Neupert Foundation to M.V. Open Access funding enabled and organized by Projekt DEAL.
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.