A dual‐functional BODIPY‐based molecular rotor probe reveals different viscosity of protein aggregates in live cells

Aberrant protein aggregation leads to various human diseases, but little is known about the physical chemical properties of these aggregated proteins in cells. Herein, we developed a boron‐dipyrromethene (BODIPY)‐based HaloTag probe, whose conjugation to HaloTag‐fused proteins allows us to study protein aggregates using both fluorescence intensity and lifetime. Modulation of BODIPY fluorophore reveals key structural features to attain the dual function. The optimized probe exhibits increased fluorescence intensity and elongated fluorescence lifetime in protein aggregates. Fluorescence lifetime imaging using this probe indicates that protein aggregates afford different viscosity in the forms of soluble oligomers and insoluble aggregates in live cells. The strategy presented in this work can be extended to enable a wide class of HaloTag probes that can be used to study a variety of physical properties of protein aggregates, thus helping unravel the pathogenic mechanism and develop therapeutic strategy.

F I G U R E 1 BODIPY fluorophore can be modulated as dual-function probes to detect protein aggregation through fluorescence intensity and lifetime increase.
both the existence and physicochemical properties of protein aggregates in live cells.
Proteins expose hydrophobic residues during misfolding and aggregation, leading to decreased polarity and increased viscosity. Such physical alterations in local environment have been monitored using probes whose fluorescence intensity or spectra (ratiometric) changes according to their microenvironment, including solvatochromic, [6][7][8] molecular-rotor based, [9][10][11][12][13] metallic complex, [14] and aggregation-induced emission probes. [15][16][17][18][19] The most commonly used methods to detect protein aggregation employ turn-on fluorescence upon protein aggregation and thus can visualize protein aggregation in live cells using fluorescence microscopy. One such technology is the recently reported AggTag method, [10,12] which distinguishes aggregated protein-of-interest (POI) from folded POI in live cells. Using this method, a chemically modified fluorescent protein chromophore emits turn-on fluorescence upon binding with aggregated POI, exhibiting a >10-fold fluorescence increase compared with folded POI. However, replying on fluorescence intensity changes could introduce potential artefacts, because intensity changes could also be triggered by local concentrations change of fluorescent probes instead of protein conformational changes. In addition, ratiometric imaging approaches have been developed to report on changes of local polarity upon protein aggregation.
By contrast, fluorescence lifetime is independent of fluorophore concentrations, thus providing more robust measurement of protein aggregation. Importantly, fluorescence lifetime could provide quantitative measure of micro-viscosity (defined as the local viscous environment surrounding the fluorescent probes) in protein aggregates. Herein, we present a boron-dipyrromethene (BODIPY)-based probe to simultaneously visualize and measure viscosity of protein aggregates in live cells, using signals from both intensity and lifetime ( Figure 1). BODIPY fluorophore has been widely used to detect cellular events and analytes. [20][21][22][23][24][25][26][27][28][29][30][31] In particular, it has been reported to probe cellular viscosity distribution using confocal microscopy and fluorescence lifetime imaging (FLIM). [32][33][34][35][36][37][38][39][40][41][42][43] In 2018, Kuimova et al. designed a BODIPY-based viscosity sensitivity probe for mapping microviscosity dynamics in cellular organelles. [44] Inspired by these pioneering work, we envision that conjugating an optimized BODIPY-based HaloTag probe to a HaloTag fused POI will allow us to study aggregation of POI using changes of both fluorescence intensity and lifetime. Such a probe should reveal both cellular location and viscosity of protein aggregates in forms of misfolded oligomers and insoluble aggregates in live cells.

Optimization of the BODIPY-HaloTag probe toward balanced viscosity sensitivity and quantum yield
We embarked on exploring the structural features of BOD-IPY to improve the sensitivity of fluorescence intensity and lifetime toward viscosity. First, enhancing steric hindrance of the rotatory phenyl group on BODIPY (group 1, Figure 2A) increased fluorescence intensities (red bars, Figure 2B) and quantum yields ( Figure S1a) in a mixture of 80% glycerol and 20% ethylene glycol. However, the viscosity sensitivity significantly decreased (probes 1b-d in Figure 2C). For these probes, the rotation of phenyl in relative to the BODIPY core is sterically inhibited, thus inhibiting probes forming confirmations that allow for fluorescence quenching due to photo-induced electron transfer. As a result, probes 1b-d would emit fluorescence even at low viscosities, exhibiting a low viscosity sensitivity. Second, substitution of the rotatory benzyl group with heterocycles (group 2, Figure 2A) retained their viscosity sensitivity (green bars, Figure 2C) but significantly quenched fluorescence intensity (green bars in Figure 2B and Figure S1b). Third, we varied functional groups on the rotatory benzyl ring to modulate its electron density (group 3, Figure 2A). While fluorescence intensity was not significantly affected (blue bars in Figure 2B and Figure S1c, with exception of nitro group of 3g), the highest viscosity dependence value was obtained using moderate electron withdrawing groups (probes 3c-e; blue bars, Figure 2C). Finally, we increased bulkiness of the benzyl group (series 4, Figure 2A) and observed marginable effects on both fluorescence intensity and viscosity sensitivity (grey bars in Figure 2B,C, Figure S1d). Spectra and photophysical parameters of these fluorophores were characterized (Figures S2-S10 and Table S1) after careful calibration of sample concentrations using quantitative 19 F Nuclear Magnetic Resonance (NMR) ( Figure S11).
Based on the above results, we chose the amide substituted benzyl-BODIPY (=0.64, probe 3c in Figure 2C) that balances the between quantum yield and viscosity dependence to design the AggTag probe. AggTag is an established method that allows for the visualiztion of misfolding and aggregation of specific proteins in live cells. This method contains two parts: a self-labeling protein tag (HaloTag in this case) that is genetically fused with the POI and the AggTag probe whose fluorescence is only activated when POI misfolds or aggregates. To enable the AggTag probe, the HaloTag reactive warhead, chlorohexane, was coupled with 3c to yield P1 ( Figure 3A, Scheme S2). [45] HaloTag is a modified haloalkane dehalogenase designed to covalently bind to synthetic chloro-alkane chain ligands. A covalent alkylate intermediate was formed by substituting the terminal chlorine of the ligand with the nucleophilic substitution of Asp106, the active site of dehalogenase. The dehalogenase structure provides reaction specificity and efficiency and has no homology to mammalian cell proteins, avoiding cross-interference and thus allowing specific fluorescent labeling of fusion proteins in living cells. P1 exhibited excitation peak at 507 nm and emission peak at 530 nm ( Figure S12); and its fluorescence intensity ( Figure 3B) was dependent on viscosity but not polarity ( Figure 3B; =0.56, =0.48 in a mixture of 80% glycerol and 20% ethylene glycol). Importantly, the fluorescence lifetime of P1 was also prolonged as the viscosity of the solvent increased ( Figure 3C and Figure S14a,b, Table S2).

P1 quantifies local viscosity of protein aggregates using fluorescence intensity and lifetime
Next, we evaluated whether P1 could detect the aggregation of a POI in vitro using both fluorescence intensity and lifetime. We utilized the HaloTag fused superoxide dismutase 1 mutant V31A, SOD1(V31A)-Halo, which is established to undergo heat-induced misfolding and aggregation. [46] Upon heating, we observed five-fold increase in fluorescence intensity when SOD1(V31A)-Halo formed protein aggregates ( Figure 3D). The corresponding absorption spectra are shown in Figure S13. Under the same condition, minimal fluorescence increase was observed for HaloTag (Figures S15 and S16). Further, we monitored the turbidity and tryptophan fluorescence of SOD1(V31A)-Halo during misfolding, which reports on the formation of insoluble aggregates and misfolded oligomers, respectively (Figures S17 and S18). Kinetics of fluorescence was identical to that of protein misfolding as measured by tryptophan fluorescence ( Figure S17) and was faster than that of insoluble aggregates as measured by turbidity ( Figure S18), suggesting that the fluorescent signal originated from misfolded proteins that form before insoluble aggregates. We next examined whether fluorescence lifetime of P1 would response to SOD1(V31A) misfolding and aggregation. When the SOD1(V31A)-Halo•P1 conjugate misfolded and aggregated at elevated temperatures, we observed prolonged lifetime of P1 (green bar in Figure 3E, and Figure S14c), from 1.64 ns (=57 cp based on Figure  S14b) at 25 • C to 3.12 ns (=180 cp) at 50.7 • C. As a control, the HaloTag•P1 conjugate exhibited a short lifetime (∼0.69 ns, =8 cp) at all tested temperatures (grey bar in Figures 3E, and Figure S14d), suggesting that the elongated lifetime was caused by the misfolding of SOD1(V31A).

P1 visualizes protein aggregates via fluorogenic signals in live cells
We next examined whether P1 could report on aggregation of a POI via fluorescence microscopy in live cells. P1 is a suited probe for live cell imaging because its fluorescence intensity was insensitive to pH values from 4.5 to 8.5 ( Figure  S19), and it exhibited low fluorescent signals when incubated with cell lysate, nonpolar solvent, proteins with hydrophobic surfaces, and lipid mimics ( Figure S20). Using the AggTag method, we carried out a dual-probe labeling experiment using P1 and an always-fluorescent TAMRA (TMR) ligand at identical concentrations ( Figure 4A), wherein HaloTag could be equally labeled by both P1 and TMR ( Figure S21). We found that P1 emitted fluorescence in the punctate aggregates formed by Huntingtin exon 1 protein (Htt) with 110 polyglutamine repeat (Htt-Q110-Halo; Figure 4B) [47] ; whereas both Htt-Q25-Halo (known to be soluble) and HaloTag exhibited minimal P1 fluorescence (Figures 4B and S22a). [47] These imaging patterns are consistent with our previously reported results. [10,12,48,49] While Htt-Q25 is known not to form aggregates upon expression, the slight background fluorescence in Htt-Q25-Halo expressing cells was observed ( Figure 4B, and the zoomed-out view in Figure S23), possibly due to its distinctive characteristics of the protein model system. The P1 fluorescence of Htt-Q110-Halo was not due to nonspecific binding of P1 in cells, because HEK293T cells co-expressing Htt-Q110-mCherry, and HaloTag showed no fluorescence of P1 ( Figure S22b).
In addition to Htt, we extended the experimental scope to pathological mutants of SOD1 expressed in HEK293T cells. [47,[50][51][52] While we did not observe significant green fluorescence in cells expressing SOD1(A4V)-Halo in the absence of the proteasome inhibitor MG132 (Figure 4C), diffuse P1 fluorescence of SOD1(A4V)-Halo was found in MG132-stressed cells ( Figure 4C). By contrast, no P1 fluorescence was observed in MG-132 treated cells that co-express TMR-labeled SOD1(A4V)-SNAPf and HaloTag ( Figure S22c), suggesting that the fluorescence was not due to nonspecific binding of P1 in stressed cells. The diffusive fluorescence signal was likely caused by soluble oligomers as indicated by a chemical crosslinking experiment wherein we identified higher molecular weight species (>250 kDa, red box in Figure S24). Similar results were obtained when other pathogenic mutants of SOD1, including G85R-Halo, G93A-Halo and V31A-Halo, [46,47,[50][51][52] were studied using fluorescence confocal microscopy ( Figure S25a-c) and the crosslinking experiment ( Figure S24). These imaging patterns are consistent with our previously reported results. [10,12,48,49] When SOD1(WT)-Halo was expressed with labeling of P1, there was no significant green fluorescence observed both in the absence and the presence of MG132 (Figure S25d), suggesting that the fluorescence of P1-labeled SOD1 mutants faithfully reports on their misfolding and aggregation in MG132-stressed cells. The relatively dark fluorescence of P1 in SOD1(WT)-expressing cells with MG132 treatment ( Figure S25d) was observed as the majority representation of multiple transfected cells in the zoomed-out view ( Figure S26). Minor population of cells, however, exhibited diffusive fluroescence of P1, suggesting misfolding of SOD1(WT) in these individual cells ( Figure S26b). This result is consistent with the previously reported finding, where the SOD1 wild-type protein, unlike SOD1 mutants, is reported not to be polyubiquitylated and degraded by proteasomes, and proteasome inhibition does not lead to the formation of SOD1 wild-type aggregation. [53]

FLIM of P1 reveals the multistep process of protein aggregation in live cells
Next, we utilized P1 to dissect the viscosity of protein aggregates in live cells using FLIM, wherein fluorescence brightness represents intensity of P1 and warmness of color displays lifetime of P1 ( Figure 5). In cells expressing HaloTag proteins (left panel of Figure 5A; Figures S27 and  S28), P1 exhibited both low fluorescent intensity and short average lifetime (2.4 ns; viscosity of ∼150 cp). By contrast, cells expressing Htt-Q110-Halo displayed the average lifetime of 3.4 ns (viscosity of 220 cp; right panel of Figure 5A; Figure S29 and S30). The punctate stucture of Htt-Q110-Halo-expressing cells is consistent with the confocal and DIC images ( Figure 4B) and our previous imaging results, where the punctate structures of Q110 aggregates are localized to the perinuclear regions. [10,12,48,49] The distribution of lifetime in cells expressing HaloTag (blue curve in Figure 5B) was broader than that of lifetime in cells expressing Htt-Q110-Halo (red curve in Figure 5B), suggesting a more rigid environment in punctuates of the Htt-Q110 aggregates than the folded Halo proteins. Inside the Htt-Q110 aggregates, the edge ( Figures S31 and S33) and the center ( Figures  S32 and S34) of punctuates exhibited similar lifetime profiles, suggesting that the rigidity of these punctuates was homogeneous. In addition to cells analyzed in Figure 5A, we observed similar lifetime values (3.5 ns) of Htt-Q110 punctuates in other cells, supporting the representativity (Figures S35). When cells expressing SOD1(A4V)-Halo were analyzed by FLIM, we found that the average lifetime of P1 was 3.0 ns (left panel of Figure 5C; Figures S36 and S37) and exhibited a broad distribution (blue curve in Figure 5D). In cells treated with MG132, we observed both diffusive and punctate fluorescent structures (right panel of Figure 5C), whose lifetime profiles were analyzed separately. While the average lifetime of the diffusive fluorescence was 3.0 ns ( Figures S38 and S40), the average lifetime of the punctate fluorescence was 3.4-3.6 ns (Figures S39 and S41) and displayed a broad shoulder (red curve in Figure 5D). Similar lifetime values were obtained when we imaged other MG132-treated cells expressing SOD1(A4V)-Halo, supporting the representativity ( Figures S42). These results suggest that the misfolded oligomers of SOD1(A4V) exist in a low viscosity and loosely-packed conformation, which is consistent with the hypothesis that misfolded oligomers exhibit liquid-like properties. [3,50,54,55] Whereas, the puncta structures of SOD1(A4V) aggregates exhibit higher viscosity, likely resulted from insoluble protein aggregates.

CONCLUSION
In summary, we have developed a BODIPY-based AggTag probe that is capable of detecting protein aggregates in live cells. Importantly, BODIPY probes, when reporting on misfolded and aggregated proteins, exhibit increase of signal in both fluorescence intensity and lifetime. Thus, these probes allow us to reveal the morphology, viscosity, and spatiotemporal information of protein aggregates in live cells.
Our results highlight the capacity of using BODIPY-based AggTag probes to quantitatively measure the differential viscosity of misfolded oligomers and insoluble aggregates. We envision that novel class of AggTag probes can be enabled by the vast pool of small molecule ligands probing various physical chemical properties, such as pH, redox, and metal. These probes will allow for the unprecedented detection of physicochemical properties of protein aggregates in live cells, promoting studies in their pathogenic mechanism and development of therapeutic strategies.

C O N F L I C T O F I N T E R E S T
The authors declare no competing interests.