Instructed-Assembly of Peptides for Intracellular Enzyme Sequestration

Abstract

Liquid-like droplets of biomacromolecules are emerging as a fundamental mechanism of cellular signaling, but designing synthetic mimics to form such membraneless organelles remains unexplored. Here we report the use of supramolecular assemblies of small peptides, as a mimic of biomacromolecular condensates, for intracellular sequestration of enzymes on endoplasmic reticulum (ER). Specifically, integrating a short peptide with naproxen (a nonsteroidal anti-inflammatory drug (NSAID) and a ligand of cyclooxygenase-2 (COX-2)) generates an enzymatic substrate that acts as a precursor for instructed-assembly. Slowly dephosphorylating the precursors by phosphatases forms the corresponding hydrogelators in cellular environment, which results in the supramolecular assemblies on ER. Consisting of the precursor and the hydrogelator molecules, the assemblies enable the sequestration of COX-2 and protein-tyrosine phosphatase 1B (PTP1B) on ER. Further structure-activity investigation reveals that the co-localization of COX-2 and PTP1B relies on the NSAID motif, the phosphotyrosine, and the enzymatic dephosphorylation of the precursor. This work, for the first time, illustrates the use of supramolecular processes for associating enzymes in cells, and may provide insights for understanding intracellular liquid condensates and a new strategy for modulating protein-protein interactions.

Graphical Abstract:

It is known that biomacromolecules are able to form liquid-phase condensates, such as Cajal bodies,¹ U and P bodies,² and nucleolus,³ inside cells via liquid-liquid phase separation. Recent studies have suggested that such a liquid phase condensation likely represents a ubiquitous cellular process for organizing intracellular space and for cellular signalings.⁴ For example, the binding of DNA to cyclic GMP–AMP synthase (cGAS) results in liquid droplets that concentrate an enzyme and reactants to activate innate immune signaling.⁵ Despite these exciting developments^6,7 in biology, the use of synthetic molecules to mimic these membraneless organelles has yet to be explored. The key feature of liquid phase condensates is the non-covalent association of molecules at a high concentration to result in a liquid-liquid phase transition. This process coincides with sol-to-gel transition, a type of liquid-liquid phase transition, which is a usual consequence of enzyme-instructed self-assembly (EISA).^8,9 Thus, we decide to use EISA to generate supramolecular assemblies of peptides in cellular environment to mimic the membraneless organelles for enzyme sequestration.

As a form of instructed-assembly,¹⁰ EISA refers to the formation of ordered superstructures of molecules as the consequence of enzymatic reactions. Like self-assembly, EISA relies on non-covalent interactions to form the assemblies from small, simple building blocks; unlike self-assembly, EISA includes an enzymatic reaction, a process that is away from equilibrium (ΔG < 0) and inherently irreversible.¹⁰ Recent studies have already demonstrated the promising applications of supramolecular assemblies in biomedicine,^11–14 particularly the use of EISA of small molecules for selectively inhibiting cancer cells in vitro,^15–19 for slowing tumor progression in animal models,^20–23 and for molecular imaging applications.^24–26 These promising developments imply that EISA generated supramolecular assemblies of small molecules should be able to mimic the biomacromolecular condensates (e.g., membraneless organelles) in cellular environment.

To demonstrate the concept of EISA for sequestration of enzymes, we use EISA to generate supramolecular assemblies to interact with multiple numbers of cyclooxygenase-2 (COX-2) and protein-tyrosine phosphatase 1B (PTP1B) simultaneously. Based on above concept, we designed EISA precursor Npx-1P, which consisted of a NSAID drug (naproxen), a self-assembling D-peptide backbone (D-Phe-D-Phe), an enzyme trigger (D-phosphotyrosine), and a positively charged homoarginine residue at the C-terminal. Such a design allows Npx-1P to interact selectively with COX-2²⁷ and to serve as a substrate for PTP1B²⁸ simultaneously. Moreover, the use of homoarginine allows the supramolecular assemblies to form on ER via EISA as evidenced by our previous report that the fluorescent analogue of Npx-1P forms enzymatic assemblies on ER of cancer cells.²⁹ As illustrated in Scheme 1, partially dephosphorylated by phosphatases, the precursor (Npx-1P) and its corresponding hydrogelator (Npx-1) co-assemble to form supramolecular assemblies that promote the association of COX-2 and PTP1B on endoplasmic reticulum (ER). Critical micelle concentration (CMC) measurement and transmission electron microscopy (TEM) confirm that enzymatic conversion of Npx-1P to Npx-1 results in increased self-assembling propensity and the phase transition (sol-gel). The formed supramolecular assemblies enrich COX-2 and PTP1B in a cell free condition, as evidenced by the TEM images and pull down analysis. Immunofluorescence staining indicates that the nanofibers, forming in cellular environment, interact with COX-2 and PTP1B and induce their association at ER of cells (Saos-2³⁰). Structure-activity relationship studies reveal that the COX-2 binding NSAID motif and the phosphatase substrate are essential for the association of the enzymes. As the first demonstration of EISA as a molecular process to enable intracellular sequestration of enzymes, this work illustrates enzymatic non-covalent synthesis of molecular condensates in cellular environment for conferring biological functions to supramolecular assemblies, a future direction of supramolecular chemistry.³¹

Scheme 1.

Illustration of instructed-assembly for intracellular sequestration of PTP1B and COX-2.

After synthesizing the precursor Npx-1P via solid-phase peptide synthesis,³² we first examined its liquid-liquid phase transition. Upon the addition of alkaline phosphatase (ALP), the Npx-1P undergoes a sol–gel transition when the concentration is about 400 µM (Figure 1A, Figure S9) in PBS buffer. We used liquid chromatography–mass spectrometry (LC-MS) and TEM to examine the enzymatic dephosphrylation of Npx-1P and the generation of supramolecular assemblies. Npx-1P hardly forms any ordered nanostructure at the concentration of 12.5 µM (Figure S10), agreeing with the CMC of Npx-1P (35 µM) (Figure 1B) in PBS buffer. The addition of ALP to the solution of Npx-1P in PBS for half an hour dephosphorylates 15% of Npx-1P to Npx-1 (Figure 1C) to form aggregates (Figure 1D). After 1 h incubation, the amount of dephosphorylated Npx-1P increases to 29%, which yields an Npx-1 concentration of 3.6 µM. While the concentration of Npx-1 in the above solution (3.6 µM) is lower than the CMC of Npx-1 (5.7 µM), short nanofibers form with a width of 7 ± 2 nm, suggesting the participation of Npx-1P at the mixture in the assembly process. The structural similarity between Npx-1P and Npx-1 favors their co-assemblies, as evidenced by the nanofibers with a uniform width of 7 nm in Figure 1D, since Npx-1P itself only forms uniform nanofibers with a width of 5 nm. (Figure S11) These results are similar to the previous reports of peptide coassembly.^33,34 From 1 h to 2 h, the short nanofibers grow to long uniform nanofibers as 50% of Npx-1P being converted into Npx-1. Moreover, the dephosphorylation rate of Npx-1P by ALP decreased after 2 h incubation, (Figure S12) likely due to the co-assembly of Npx-1P and Npx-1, which reduces the amount of free Npx-1P. These results verify the phase/morphological transition resulted from EISA of Npx-1P and indicate the co-assembly of Npx-1P and Npx-1.

Figure 1.

(A) Sol-gel transition of Npx-1P (400 µM) upon adding ALP (1 U/mL). (B) CMCs of Npx-1P and Npx-1. (C) Dephosphorylation of Npx-1P (12.5 µM) after incubating with 0.1 U/mL ALP at different time. (D) TEM images of nanostructures formed before and after adding ALP (0.1 U/mL) to the solution of Npx-1P (12.5 µM)(scale bar = 100 nm). All the solutions were prepared in pH 7.4 PBS buffer.

We then investigated the interactions between enzymatically formed assemblies and the enzymes (COX-2 and PTP1B). Being co-incubated with Npx-1P and after the addition of ALP, either PTP1B or COX-2 adheres to the nanofibers of Npx-1P and Npx-1 (Figure S13). The incubation of COX-2 and PTP1B with Npx-1P and ALP results in larger protein aggregates on the nanofibers. Some of the proteins appear to be away from the nanofibers in the TEM image, which are likely due to the dynamic nature of enzymatic assemblies. These observations suggest interaction and enrichment of COX-2 and PTP1B on the non-covalent assemblies formed by enzymatically dephosphorylation of Npx-1P. To further confirm the interactions between assemblies and proteins, we conducted the pull-down assay of PTP1B and COX-2 with the enzymatic assemblies (Figure 2), which reveals that the pellet, compared with the supernatant, of assemblies abundantly enrich both PTP1B and COX-2, further validating the enzyme sequestration by the supramolecular assemblies formed by EISA. In addition, after 1 h incubation, PTP1B (45 U/mL) converted 50%, 40% and 25% of Npx-1P to Npx-1 when the concentration of Npx-1P is 50 µM, 100 µM and 200 µM, respectively, but hardly hydrolyzed more Npx-1P from 1 h to 24 h (Figure S14). This observation likely results from the sequestration of PTP1B from solution to the assemblies, which reduces the amount of free enzymes.

Figure 2.

Pull down of COX-2 and PTP1B with enzymatic assemblies formed by Npx-1P in PBS. The image shows Coomassie staining of pellet (P) and supernatant (S) fractions.

To verify the enzymatic transformation and self-assembly of Npx-1P in live cells, we quantified the conversion of Npx-1P after incubating it with Saos-2 cells at different time points. Figure 3A indicates that endogenous phosphatases convert 19%, 29%, and 46% of Npx-1P to Npx-1 after 0.5 h, 1 h, and 2 h incubation, respectively. This result implies the co-assembly of Npx-1P and Npx-1 in live cells, agreeing with the results from cell free condition (Figure 1). We next used immunofluorescence staining to verify whether the assemblies formed by co-assembly of Npx-1P and Npx-1 sequestrated the enzymes. Being incubated Saos-2 cells with Npx-1P (12.5 µM) after 0.5 h (Figure S15), both COX-2 and PTP1B form puncta and the green fluorescence from COX-2 start to overlap to red fluorescence from PTP1B at some regions. After 1 h incubation, more yellow fluorescence appears at the site of ER, indicating increased level of co-localization of the two enzymes (Figure 3B and C). Orthogonal Z-stack scanning (Figure 3D) also indicates the intracellular association of PTP1B and COX-2 upon incubation the cells with Npx-1P. We also examined the enzyme sequestration in HS-5 cells (as a control of Saos-2 cells since HS-5 express low level of ALPL³⁵) and found that the addition of Npx-1P hardly induced the sequestration or co-localization of COX-2 and PTP1B in the cells (Figure S16). Together with our previous report²⁹ that EISA precursors hardly form assemblies in HS-5 cells, this result indicates cell selective formation of molecular condensates for enzyme sequestration by EISA.

Figure 3.

(A) Dephosphorylation of Npx-1P (12.5 µM) after incubating with Saos-2 cells at different time. (B) CLSM images of Saos-2 cells treated with Npx-1P (12.5 µM) for 1 h and then stained with antibodies of PTP1B (red) and COX-2 (green). (C) The enlarged region of co-localization in (B) (scale bar = 2 μm). (D) Orthogonal Z-stack of co-localization of PTP1B and COX2.

To evaluate the specificity of intracellular enzyme sequestration by the assemblies formed by EISA, we used immunofluorescence staining to examine the interactions of COX-1, an isozyme of COX-2. As shown in Figure 4, COX-1 hardly forms puncta inside Saos-2 cells, indicating that the supramolecular assemblies of Npx-1P and Npx-1 scarcely interact with COX-1 in Saos-2 cells. This results is likely due to the enhanced selectivity of Npx-1P towards COX-2,²⁷ suggesting that the strong interaction between the naproxen-peptide conjugates and COX-2 is critical for the sequestration of COX-2. To further verify the molecular basis of Npx-1P for the sequestration of COX-2 and PTP1B, we designed two control molecules (Nap-1P, Npx-2P) by mutating fragments in Npx-1P (Scheme S1). Replacing the naproxen with naphthalene in Npx-1P yields Nap-1P, which hardly induces the puncta formation of COX-2 or PTP1B, further confirming the importance of the COX-2 binding moiety (i.e., naproxen). Having a D-phosphoserine to replace the D-phosphotyrosine (in Npx-1P), Npx-2P fails to cause the association of PTP1B or COX-2, agreeing with that phosphoserine significantly diminishes PTP1B-substrate interaction^36,37 and self-assembling ability. These results, collectively, confirm the essential role of phosphotyrosine and naproxen for the sequestration of PTP1B and COX-2 by the assemblies of Npx-1P and Npx-1.

Figure 4.

CLSM images of Saos-2 cells stained with antibodies of PTP1B (red) and COX-1/COX-2 (green) after treating with Npx-1P, Nap-1P and Npx-2P for 1 h (scale bar = 10 μm).

In conclusion, this work illustrated a novel strategy to sequestrate intracellular enzymes selectively via instructed-assembly. Incorporating with specific protein interaction motifs, the precursors, via EISA process, form supramolecular assemblies that display the protein-binding motifs on the surface of the molecular assemblies due to the dynamic and adaptive properties of the peptide assemblies, thus interacting with multiple proteins in a manner to concentrate specific proteins (e.g., PTP1B and COX-2 in this work). This process is similar to formation of higher-order assemblies of proteins³⁸ or protein phase transition,³⁹ a ubiquitous phenomenon in biology for cell signaling. Moreover, employing synthetic small molecules to generate intracellular molecular assemblies (or condensates) represents a general way to create a multifunctional signal hub for modulating diverse protein-protein interactions,⁴⁰ especially for controlling protein-protein interaction involving tertiary epitopes that are much more dynamic.⁴¹ Further exploration along this direction may lead to remarkable therapeutic benefits, as shown by the drugs that control protein-protein interactions.^42,43