Multichannel Dual Protein Sensing Using Amphiphilic Supramolecular Assemblies

Abstract

Protein sensing strategies have implications in detection of many human pathologies. Here, a supramolecular strategy for sensing two different proteins using a multichannel readout approach is outlined. Protein-ligand binding or enzymatic cleavage can both be programmed to induce supramolecular disassembly, which leads to fluorescence enhancement via aggregation-induced emission (AIE), protein-induced fluorescence enhancement (PIFE), or disassembly-induced fluorescence enhancement (DIFE). The accompanying signal change from two different fluorophores and their patterns are then used for specific protein sensing.

Dysregulation of proteins or enzymes is often associated with pathological developments.^1–5 Protein sensing offers a great tool for understanding these processes as well as for biomarker detection and disease diagnosis.^6,7 Most frequently used approaches for protein sensing include antibody-based immunoassays and ligand-based fluorescent probes.^8–14 The former method provides high specificity in detection, but this approach is time-consuming and requires costly development of a new antibody for each protein.¹⁵ The latter approach is relatively simple, but is mainly utilized for enzymes that exhibit fast reaction kinetics and high selectivity.¹⁶ Also, both approaches can only detect a single protein at a time with a single signal readout, while multianalyte sensing is desirable as sensing strategies advance. Array-based sensing provides one way to detect multiple proteins depending on ‘fingerprint’ identified from multichannel signals; however, since the platforms are hypothesis-free and based on non-specific interactions, the lack of detailed mechanism understanding and unpredictability of this approach impede further development.¹⁷ Therefore, a simple sensing platform with enhanced sensitivity and robustness for multiple protein sensing is desired.

Here we present a supramolecular approach embedded with protein-ligand binding feature and enzyme-cleavable moiety for dual protein sensing based on multichannel fluorescence signals (Figure 1). In the sensing process, the first signal is generated when the ligand-installed supramolecular assemblies interact with the target protein via the lock-key principle. The second signal arises from enzyme-induced assembly transformation. These assemblies are also expected to undergo disassembly resulting in distinct signal changes from both channels. The signal generation and alteration accompanied with the assembly deformation offer multiple pathways to confirm the sensing event. The clear mechanistic basis in the proposed approach allows us to understand the whole sensing regime and potentially adapt it to other types of protein analytes.

Figure 1.

Schematic representation of the proposed protein sensing assembly where the signal is generated from binding events and enzymatic activities

Our sensing strategy of multichannel signal generation relies on three distinct mechanisms, protein-induced fluorescence enhancement (PIFE), disassembly-induced fluorescence enhancement (DIFE), and aggregation-induced emission (AIE). Firstly, recognition of target protein takes place at the interface of ligand-installed assemblies and the protein analyte, which brings the protein to the proximity of the fluorescent probe. This phenomenon restrains the probe conformation on the assembly surface, resulting in a fluorescence enhancement.^18–25 Secondly, AIE fluorogens that are attached to the assemblies display a weak emission, but rapidly aggregate upon enzyme-induced cleavage and disassembly to produce enhanced emission.^26–28 Thirdly, protein binding and enzymatic cleavage can result in supramolecular disassembly^29–35, which leads to molecular rearrangement of cyanine dye from a quenched organized stacking state to a free soluble state, resulting in significant fluorescence enhancement.

We aimed to design a supramolecular assembly that generates PIFE, DIFE, and AIE in presence of specific proteins. Bovine carbonic anhydrase (bCA) is used as a model protein for its well-studied binding with sulfonamide ligand^36–38, while porcine liver esterase (plE) is chosen as a model enzyme to cleave the amphiphiles used for assembly formation. Both bCA and plE are disease-relevant and have human analogues that are overexpressed in cancers.^39–46 For the initial proof of concept, we first tested whether protein-ligand binding can result in fluorescence enhancement with a small molecule probe 1, which contained sulfo-Cy3 as the fluorescent probe and phenylsulfonamide as the ligand for bCA (Figure 2a). A significant fluorescence enhancement was observed only with bCA, suggesting specificity in ligand-protein interactions, while no fluorescence increase was observed with other proteins, including protein tyrosine phosphatase 1B (PTP1B) and cyclooxygenase 2 (COX2) which have phenyl sulphonamide moiety as a substructure of their ligands^47,48 (Figure 2c). Also, the fluorescence increased in a concentration-dependent manner, highlighting the potential for protein quantification (Figure 2d).

Figure 2.

(a) Structure of the PIFE probe 1 (b) Structure of the TPE amphiphilie probe 2 (c) Cy3 fluorescence change of 1 after 100 mol% addition of various proteins (d) Cy3 fluorescence change of 1 after 50 – 150 mol% bCA addition (e) TPE fluorescence change of 2 after 1 mol% addition of various proteins followed by 12 hr incubation (f) TPE fluorescence change of 2 after 0.1 – 1 mol% plE addition

Next, we investigated the AIE-based mechanism for enzyme sensing. Amphiphilic probe 2, which combines AIE fluorogen, alkylated tetraphenylethylene (TPE), and poly(ethylene glycol) (PEG) chain through an ester linkage, was synthesized to form assemblies and evaluate their response in the presence of plE (Figure 2b). Probe 2 was able to form nanoassemblies with a size around 173 nm in aqueous solution (Figure S1). When we added plE to the obtained assembly, TPE fluorescence signal at 450 nm increased over time, suggesting the aggregation of TPE molecules after enzyme cleavage. Nonetheless, this fluorescence increase was not observed with the addition of other proteins, demonstrating the specificity of this approach (Figure 2e). Also, the rate of signal enhancement was found to be dependent on plE concentration within the range of 0.1 to 1 mol% (0.14 to 1.37 μM of the whole solution) (Figure 2f and Figure S2). We also noticed a size change of the assembly from 173 nm to 20 nm after 48 hours of plE addition suggesting plE induced cleavage (Figure S1).³⁰

The successful demonstration of bCA-induced PIFE and plE-induced AIE allows us to combine these features into one scaffold, which can be used for detecting two proteins with multichannel signal readout. To this end, we synthesized molecule 3 by conjugating sulfonamide-based small molecule probe with pentaethylene glycol extension to the TPE-based amphiphile (Figure 3a). Nanoassemblies were generated by combining amphiphiles 2 and 3 in a 1:1 ratio, which had a size around 300 nm. Then we examined the possibility of utilizing this assembly to sense the presence of bCA and plE with multichannel signals. To mimic the protein sensing in a complex environment, we performed all experiments by the spiking model proteins in 10% fetal bovine serum solution. First, when 1 equiv of bCA (relative to the sulfonamide moiety) is added to the assembly solution, around 20% fluorescence enhancement from the TPE channel is observed (Figure 3b), attributed to the microenvironment change of TPE as a result of binding-induced disassembly (Figure S3). Then with the addition of plE (1 mol% relative to the ester group) more than 7-fold TPE enhancement is observed after 48 hours, attributed to the enzymatic cleavage of the ester bond generated decyl TPE and the resultant AIE. Also, we observed ~40% increase in fluorescence intensity of the Cy3 channel after bCA addition and another 5-fold increase after the addition of plE (Figure 4c). The former signal enhancement is likely due to the ligand-bCA binding, while the latter observation is attributed to: i) the molecular rearrangement of Cy3 during the supramolecular disassembly, because the initial fluorescence of Cy3 in the nanoassembly is quenched by 65% compared to 1 (Figure S4), the disassembly of which offers more freedom for the Cy3 moiety; ii) disassembly led to more accessible binding of sulfonamide with bCA and thus higher PIFE, compared to the assembly where the binding sites are more sterically hindered.

Figure 3.

(a) color-coded structure of amphiphile 3 (b) TPE fluorescence change before, after 100 mol% bCA addition, and after 100 mol% bCA then 1 mol% plE addition with 48 hr incubation (c) Cy3 fluorescence change before, after 100 mol% bCA addition, and after 100 mol% bCA then 1 mol% plE addition with 48 hr incubation (d) TPE fluorescence change upon addition of different bCA concentration (e) Cy3 fluorescence change upon addition of different bCA concentration (f) TPE fluorescence change after addition of different bCA then plE concentration with 12 hr incubation (g) Cy3 fluorescence change after addition of different bCA then plE concentration with 12 hr incubation

Figure 4.

(a) TPE fluorescence change before, after 1 mol% plE addition with 48 hr incubation, and after 48 hr incubation with 1 mol% plE then 100 mol% bCA addition (b) Cy3 fluorescence change before, after 1 mol% plE addition with 48 hr incubation, and after 48 hr incubation with 1 mol% plE then 100 mol% bCA addition (c) TPE fluorescence change upon addition of different plE concentration followed by 12 hr incubation (d) Cy3 fluorescence change upon addition of different plE concentration followed by 12 hr incubation (e) TPE fluorescence change after 48 hr incubation with different plE then bCA concentration (f) Cy3 fluorescence change after 48 hr incubation with different plE then bCA concentration

Next, to explore whether the signal changes are concentration dependent, we introduced different amounts of bCA and plE to the assemblies. With the addition of bCA increased from 0.5 eq. to 1.5 eq., steady increase was achieved in the Cy3 channel (Figure 3g), suggesting the binding of bCA and Cy3 is concentration dependent. Meanwhile little increase was observed from the TPE channel, which is understandable because the bCA binding does not induce TPE aggregation (Figure 3f). Similar trend was still observed for both Cy3 and TPE after 0.5 mol% plE was added to the above solution, other than that the signal intensity of TPE increased 1.5–2 fold and the Cy3 increased 3–5 fold, again confirming the AIE and DIFE sensing mechanism. Later the amount of plE was increased from 0.5 to 2 mol%, we noticed a linear increase in the TPE channel but not in the Cy3 channel, meaning the AIE signal generation is dependent on the decyl TPE generation. However, the Cy3 signal did not increase much with additional plE, suggesting the first dose of plE is enough to induce the disassembly and facilitate the Cy3 fluorescence recovery. These findings together show that the fluorescence change of TPE channel is dependent mostly on plE concentration not bCA, and the Cy3 channel is dependent on the bCA concentration, but with plE addition, the Cy3 signal can be further enhanced.

To illustrate that the sensing scaffold could show unique fluorescence enhancement regardless of the order of protein additions, we switched the order of treating the assembly by firstly adding plE then bCA. After introducing 1 mol% plE into the assembly solution, the TPE signal showed 9-fold increase (Figure 4a), in which the increasing trend aligned with our previous observation in Figure 3f. 3-fold increase was also found in the Cy3 channel (Figure 4b) and this increase was much higher than the bCA-treated solution (as shown in Figure 3c), which is mainly due to DIFE. After that, 100 mol% of bCA was mixed into the sensing solution, which showed no effect on the TPE signal but led to a 2.4-fold Cy3 fluorescence enhancement. This is due to the combination of DIFE and PIFE.

To gain more quantitative insights, varied concentrations of both plE and bCA were added into the assembly solution. The sensing assembly was first treated with 0.5 – 2 mol% plE. After 12 hours of incubation, the TPE signal was enhanced in a concentration-dependent manner, 1.5, 3, and 6-fold increase for 0.5, 1, and 2 mol% plE respectively. (Figure 4c), similar to results in Figure 3f. The Cy3 fluorescence also demonstrated gradual enhancement for 0.5, 1, and 2 mol% plE. (Figure 4d) This trend suggests that increasing plE concentration can lead to more Cy3 fluorescence recovery from the quenched form on the assembly. Afterwards, varied doses of bCA (0.5–1.5 equiv) were introduced to the solution. The fluorescence of Cy3 further increased by 1.5–2.5 fold (Figure 4f), which was much more drastic compared to that without plE (Figure 3e). This indicates that PIFE is more prominent with freely dissolved ligands than the ones embedded in assemblies. Also, the variation of bCA did not change the signal of TPE channels dramatically (Figure 4e), similar to the observations in Figure 4d. These observations illustrate that the presence of plE alone can be recognized by enhancement in TPE and Cy3 channels via AIE and DIFE mechanisms respectively. The addition of bCA mainly affected Cy3 fluorescence which is consistent to the readouts from the assembly added with bCA then plE.

In this work, we illustrate a supramolecular approach for multichannel protein sensing with concurrent selectivity towards two different proteins. We show that 1) PIFE, AIE, and DIFE concepts can be harnessed together for quantitative fluorescence enhancement; 2) the sulfo-Cy3 fluorescence change depends on specific protein-ligand binding and molecular rearrangement; 3) the fluorescence change of the TPE channel is mainly due to the enzyme-induced TPE aggregation; and 4) the unique sensing patterns allow sensing, quantification, and verification at the same time. Overall, we show that supramolecular strategies can be used for multichannel sensing for use in protein detection. These strategies could form the basis for further development of disease relevant multi-protein biomarker detection and quantification.