Spontaneous Hollow Coacervate Transition of Silk Fibroin via Dilution and Its Transition to Microcapsules

Abstract

Polymeric microcapsules are useful for drug delivery, microreactors, and cargo transport, but traditional fabrication methods require complex processes and harsh conditions. Coacervates, formed by liquid-liquid phase separation (LLPS), offer a promising alternative for microcapsule fabrication. Recent studies have shown that coacervates can spontaneously form hollow cavities under specific conditions. Here, we investigate the spontaneous hollow coacervate transition of silk fibroin (SF). SF coacervates, induced by mixing SF with dextran, calcium ions, and copper ions, transition to hollow coacervates upon dilution. Adding ethylenediaminetetraacetic acid (EDTA) further transforms them into vesicle-like capsule coacervates, which solidify into microcapsules. As a proof-of-concept, we successfully loaded a high-molecular-weight polymer cargo into the hollow cavity and bioactive enzyme cargo into the capsule layer by simply mixing the cargo with the coacervate solution. Our results demonstrate a facile, organic-solvent-free approach for fabricating SF-based microcapsules and provide insight into the mechanisms driving hollow coacervate formation.

Figure 1

LLPS of SF induced by different cation compositions. (a) Representative optical image showing the liquid–liquid phase separation (LLPS) formation by mixing the silk fibroin (SF) solution with a solution containing dextran and copper ions. (b) LLPS of the SF solution containing 1.5% (w/v) SF, 5% (w/v) dextran, and 1 mM copper ions. The nonspherical shapes of the coacervates indicate gelation. (c) LLPS of the SF solution containing 1.5% (w/v) SF, 5% (w/v) dextran, 1 mM copper ions, and 50 mM calcium ions. Scale bar: 20 μm.

Figure 2

Phase separation diagrams of solutions with different compositions. The phase separation diagrams were drawn as a function of silk fibroin (SF) and calcium ion concentrations. Separate phase separation diagrams were drawn for solutions with different copper ion and dextran concentrations: (a) 2% (w/v) dextran and 0 mM copper ions, (b) 5% (w/v) dextran 0 mM copper ions, (c) 10% (w/v) dextran and 0 mM copper ions, (d) 2% (w/v) dextran and 0.2 mM copper ions, (e) 5% (w/v) dextran and 0.2 mM copper ions, (f) 10% (w/v) dextran and 0.2 mM copper ions, (g) 2% (w/v) dextran and 1.0 mM copper ions, (h) 5% (w/v) dextran and 1.0 mM copper ions, and (i) 10% (w/v) dextran and 1.0 mM copper ions. Each data point in the phase separation diagram was noted in three categories: 1. liquid–liquid phase separation (LLPS) observed right after mixing (red circle; 2-phase), 2. no LLPS observed (black X; 1-phase), and 3. small particles or LLPS observed after 5 min (orange triangle). The red binodal line separating the 2-phase and the 1-phase regions was drawn manually based on these data points. The 2-phase region in each phase separation diagram is colored in red.

Figure 3

Hollow coacervate transition through dilution. (a) Optical microscopic image of a liquid–liquid phase separation (LLPS) solution containing 0.5% (w/v) silk fibroin (SF), 16.67 mM calcium ions, 10% (w/v) dextran, and 1 mM copper ions. (b) Optical microscopic image of a hollow coacervate solution formed by diluting the LLPS with deionized water, resulting in final concentrations of 0.1% (w/v) SF, 3.33 mM calcium ions, 2% (w/v) dextran, and 0.2 mM copper ions. A CLSM image of the hollow coacervates is shown in the inset of panel (b). SF labeled with rhodamine B isothiocyanate was used for the CLSM image. Scale bar = 20 μm.

Figure 4

Dilution pathways for the hollow coacervate transition. A liquid–liquid phase separation (LLPS) solution containing 1.5% (w/v) silk fibroin (SF), 50 mM calcium ions, 10% (w/v) dextran, and 1 mM copper ions was diluted with solutions of various compositions to reach specific regions of the phase separation diagram. The data point corresponding to the initial composition of the LLPS solution before dilution is marked as a black circle on the phase separation diagram on the left (identical to the phase separation diagram in Figure 2i). Dilution led to the final compositions shown as points “a”–“d” on the phase separation diagram on the right (identical to the phase separation diagram in Figure 2d). Optical microscopic images show the resulting coacervate structures in the diluted LLPS solutions corresponding to points “a”–“d”. Detailed compositions of the solutions used for the current figure are provided in Table S2. Scale bar: 20 μm.

Figure 5

Effect of the dilution speed on the hollow coacervate transition. A liquid–liquid phase separation (LLPS) solution containing 1.5% (w/v) silk fibroin (SF), 50 mM calcium ions, 10% (w/v) dextran, and 1 mM copper ions was diluted at different speeds to induce a hollow coacervate transition. The data point corresponding to the initial composition of the LLPS solution before dilution is marked as a black circle on the phase separation diagram on the left (identical to the phase separation diagram in Figure 2i). The dilution led to the final compositions shown as black circles on the phase separation diagram on the right (identical to the phase separation diagram in Figure 2d), corresponding to 0.1% (w/v) SF, 10 mM calcium ions, 2% (w/v) dextran, and 0.2 mM copper ions. The dilution speed was controlled by the number of sequential dilutions. The optical microscopic images show the coacervates resulting from the dilution pathways “a”–“c”. Detailed compositions of the solutions used for the current figure are provided in Table S2. Scale bar: 20 μm.

Figure 6

Effect of the dilution speed on the diameter ratio. The diameter of the hollow cavities (I.D.) was normalized by dividing it by the diameter of the hollow coacervates (O.D.) to obtain the diameter ratio (I.D./O.D.). The box-and-whisker plot shows the interquartile range (25–75th percentile) within the box, with the whiskers representing data within 1.5 times the interquartile range. The line inside the box indicates the median, while the white square denotes the mean. The corresponding data points of the box graphs are indicated as black dots. Statistical significance (P < 0.005) was assessed using a one-way analysis of variance (ANOVA) with the Bonferroni post hoc test. Statistical significance is indicated as ***P < 0.005.

Figure 7

Diluting the LLPS solution to the 1-phase region with fixed copper ions or dextran concentrations. A liquid–liquid phase separation (LLPS) solution containing 1.5% (w/v) silk fibroin (SF), 50 mM calcium ions, 10% (w/v) dextran, and 1 mM copper ions, the corresponding data points of which are denoted as a black circle in the phase separation diagram on the left (identical to the phase separation diagram in Figure 2i), was diluted to the 1-phase region with fixed copper ions or dextran concentrations. Dilution pathway “a” results in the final solution with a composition of 0.1% (w/v) SF, 10 mM calcium ions, 10% (w/v) dextran, and 1 mM copper ions (identical to the phase separation diagram in Figure 2i). Dilution pathway “b” results in the final solution with a composition of 0.1% (w/v) SF, 10 mM calcium ions, 10% (w/v) dextran, and 0.2 mM copper ions (identical to the phase separation diagram in Figure 2f). Dilution pathway “c” results in the final solution with a composition of 0.1% (w/v) SF, 10 mM calcium ions, 2% (w/v) dextran, and 1 mM copper ions (identical to the phase separation diagram in Figure 2g). The optical microscopic images show the coacervates resulting from the dilution pathways “a”–“c”. Detailed compositions of the solutions used for the current figure are provided in Table S2. Scale bar: 20 μm.

Figure 8

Hollow coacervate-to-capsule-coacervate transitions by EDTA. (a) Optical microscope and 3D-reconstructed CLSM Z-stack images of a hollow coacervate solution with a composition of 0.1% (w/v) silk fibroin (SF), 3.33 mM calcium ions, 2% (w/v) dextran, and 0.2 mM copper ions. (b) Optical microscope and 3D-reconstructed CLSM Z-stack images of a capsule coacervate solution obtained by mixing the hollow coacervate solution with an equal volume of 0.5 M, pH 8.0 ethylenediaminetetraacetic acid (EDTA) solution. SF labeled with rhodamine B isothiocyanate was used for the CLSM images. The 3D reconstruction of the Z-stacked images was done to the half-height of the coacervates to visualize the hollow structure. Full reconstruction images can be found in Figure S18. Scale bars for the optical microscope and CLSM images represent 50 and 20 μm, respectively.

Figure 9

FRAP analysis of the hollow coacervate and the capsule coacervate. Fluorescence recovery after photobleaching (FRAP) analysis was conducted on the hollow coacervates and the capsule coacervates by bleaching an area 2 μm in diameter on the coacervates. (a) Normalized FRAP intensities of the hollow coacervates and the capsule coacervates. Data are presented as mean ± SD (n = 4). (b) Representative CLSM images of the hollow coacervate and capsule coacervate right after and 5 min after photobleaching. SF labeled with rhodamine B isothiocyanate was used for the FRAP experiments. Scale bar: 10 μm.

Figure 10

Permeability and small molecule partitioning properties of capsule coacervates. (a) Molecular weight-dependent permeability of capsule coacervates was measured from the confocal laser scanning microscopy (CLSM) images of capsule coacervates mixed with an equal volume of 1 mg/mL fluorescein isothiocyanate-dextran (FITC-dextran) solutions of varying molecular weights (4, 20, 150, 250, and 500 kDa). The permeability was calculated by dividing the average fluorescence intensity of the hollow cavities by the average background fluorescence intensity outside of the capsule coacervates. Data are presented as mean ± SD (n = 5). (b) Small molecule partitioning of capsule coacervates was investigated by adding 0.5 μL of a solution containing each small molecule to 20 μL of the capsule coacervate solution. Nile red (hydrophobic), calcein (anionic), propidium iodide (cationic), glucose oxidase (GOx; enzyme), and hydrogen peroxidase (HRP; enzyme) were chosen as small molecules. Scale bar: 50 μm.

Figure 11

Microcapsules after the washing procedure. (a) The images show the stability of the microcapsules after a complete liquid-to-solid transition. The precipitated microcapsules after centrifugation (10,000g, 15 min) can be redispersed without dissociation or aggregation of the microcapsules. (b) The optical microscopic image shows the microcapsules stably redispersed after centrifugation. Scale bar: 20 μm.

Figure 12

Microcapsules as a dual-cargo delivery vessel. (a) Confocal laser scanning microscope (CLSM) images of fluorescein isothiocyanate-dextran (FITC-dextran) loaded inside the microcapsules. FITC-dextran with molecular weights of 150, 250, and 500 kDa was used as a high-molecular-weight cargo model. Scale bar: 20 μm. (b) CLSM images of microcapsules with glucose oxidase (GOx) and hydrogen peroxidase (HRP) loaded in the capsule layer. GOx and HRP were chosen as model bioactive cargos. Scale bar: 20 μm. (c) The 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) assay shows the enzyme cascade reaction of HRP and GOx loaded in the microcapsules. ABTS assay of the solution surrounding the microcapsules was also conducted to investigate the potentially existing free enzyme released from the microcapsules. ABTS assay was conducted at 37 °C using 100 mM ABTS and 100 mM glucose. The ABTS conversion was assessed by measuring the absorbance at 415 nm at 10 min intervals for 12 h. Data are presented as mean ± SD (n = 3).