Programming scheduled self-assembly of circadian materials

Abstract

Active biological molecules present a powerful, yet largely untapped, opportunity to impart autonomous regulation of materials. Because these systems can function robustly to regulate when and where chemical reactions occur, they have the ability to bring complex, life-like behavior to synthetic materials. Here, we achieve this design feat by using functionalized circadian clock proteins, KaiB and KaiC, to engineer time-dependent crosslinking of colloids. The resulting material self-assembles with programmable kinetics, producing macroscopic changes in material properties, via molecular assembly of KaiB-KaiC complexes. We show that colloid crosslinking depends strictly on the phosphorylation state of KaiC, with kinetics that are synced with KaiB-KaiC complexing. Our microscopic image analyses and computational models indicate that the stability of colloidal super-structures depends sensitively on the number of Kai complexes per colloid connection. Consistent with our model predictions, a high concentration stabilizes the material against dissolution after a robust self-assembly phase, while a low concentration allows for oscillatory material structure. This work introduces the concept of harnessing biological timers to control synthetic materials; and, more generally, opens the door to using protein-based reaction networks to endow synthetic systems with life-like functional properties.

Fig. 1. Harnessing circadian clocks to engineer non-equilibrium materials across scales.

A We functionalize cyanobacteria clock proteins—hexameric KaiC rings (blue), KaiA dimers (cyan), and KaiB monomers (green)—to couple to materials by incorporating biotinylated KaiB (b-KaiB). B KaiB biotinylation: (left) example sites of possible amine-reactive biotinylation (magenta) overlaid on the KaiB crystal structure (green)^,, (right) SDS-PAGE gel of unlabeled KaiB and biotinylated KaiB (b-KaiB), showing successful biotinylation indicated by a mobility shift of the biotinylated product (molecular weight standards [M] are 10 and 15 kDa). C KaiABC reactions in the presence of biotinylated KaiB. Oscillations are measured by fluorescence polarization of FITC-labeled KaiB (0.2 $\mathrm{\mu }\mathrm{M}$), a read-out of KaiB–KaiC complex formation. All conditions contain 3.5 $\mathrm{\mu }\mathrm{M}$ KaiB, with the specified fraction being b-KaiB. Oscillatory association of KaiB with KaiC is sustained with 55% b-KaiB (magenta), the percentage used in subsequent experiments, but not with 80% (pink). Each curve is the mean of two replicates. D KaiB monomers bind cooperatively to KaiC rings in a phosphorylation-dependent manner (indicated by the orange “P” circles), mediated by KaiA, and are subsequently released as KaiC dephosphorylates over a 24-h cycle. We exploit the transition from free KaiB to KaiB fully assembled on a KaiC hexamer to create a time-dependent and phosphorylation-dependent change in crosslinking valency. E We incorporate the “circadian crosslinkers” depicted in (D) into suspensions of 1-$\mathrm{\mu }\mathrm{m}$ streptavidin-coated colloids to drive time-dependent crosslinking of colloids. F Microscope images of fluorescent streptavidin-coated colloids, mixed with KaiB, b-KaiB, and KaiC phosphorylation site mutants that cannot bind KaiB (left) or constitutively bind KaiB (right), show that KaiB–KaiC assembly selectively causes mesoscale clustering and connectivity of colloids. G Sedimentation of colloidal clock suspensions shown in (F) demonstrates pronounced settling of colloids after a day of incubation with the mutant that forms constitutively KaiB–KaiC complexes (right) compared to colloids mixed with the non-binding KaiC mutant (left). Cartoons depict the expected state of the suspension for constitutively binding (left) and non-binding (right) mutants (not drawn to scale). Source data for C are provided as a Source Data file.

Fig. 2. KaiB–KaiC complexes crosslink colloids with high specificity in a phosphorylation-dependent manner.

Fluorescence microscopy images of suspensions of 1-µm diameter colloids taken at 1 h (top), 7 h (middle), and 28 h (bottom) after mixing with KaiC mutants that are frozen in A non-binding (pT) or B binding (pS) states show substantial clustering and assembly of pS-colloids over time that is absent for pT-colloids. Fluorescence microscopy images of suspensions of colloids of 2 µm (C) and 6 µm (D) diameter in the presence of pT and pS KaiC proteins show that timed aggregation, dependent on the phosphorylation state of KaiC, is preserved for different sizes of colloids. The concentrations of colloids, proteins, and reagents, as well as imaging parameters, are identical to those in (A, B). E Images of a suspension of 1-µm diameter colloids undergoing sedimentation in a 12 mm long capillary over 28 h in the presence of pT (left) and pS (right) show that pS-colloids sediment more quickly, as indicated by dark regions extending further down the images. The time that each image is captured is listed at the top. F The same suspension parameters as in (A, B) but without KaiC (including only KaiB and b-KaiB) show minimal clustering over the course of 1 h (top) to 28 h (bottom), demonstrating that the KaiB–KaiC complex formation is essential to the colloidal self-assembly shown in (A–E). G Suspensions of streptavidin-coated colloids, with identical conditions to those in (A, B), but with Kai proteins replaced with alternative biotinylated constructs that could, in principle, crosslink streptavidin-coated colloids: (left) 1 kDa biotin-PEG-biotin with 1 biotin on each end, (middle) 20 kDa biotin-PEG-biotin with 1 biotin on each end, and (right) biotin-BSA with 8-16 biotins. The molarity of PEG and BSA matched the KaiC molarity used in (A–E), and minimal clustering is observed from 1 h (top) to 28 h (bottom), demonstrating that the effect shown in (A–F) is unique to the KaiB–KaiC binding interaction.

Fig. 3. KaiBC crosslinking mediates robustly timed self-assembly of colloidal clusters that are synced with KaiBC complex formation.

Colorized temporal projections of time-lapses of pT KaiC (A, yellow) and pS KaiC (B, cyan) over the course of 28 h, with colors indicating increasing time from dark to light according to the color scales. (Insets) Zoomed-in regions of the projections highlighting pS-specific cluster growth over time that is absent for pT. Pixel intensity probability distributions for pT (C, yellow), and pS (D, cyan) at different times over 28 h, with lighter shades denoting later times according to the color scales in (A, B). Distributions show broadening and emergence of high-intensity peaks at later times for pS. Dashed gray line denotes the full width at 1% of the maximum probability (FW1%), which serves as a clustering metric used in (F). Data are generated from the same images used to generate colormaps in (A, B). E Spatial image autocorrelation functions $g\left(r\right)$ versus radial distance $r$ (in units of colloid diameter) for 5 different times between 1 and 28 h for pT (yellow) and pS (cyan) with color shade indicating time according to the legends in (A, B). The characteristic correlation length $\xi$, determined by fitting each $g\left(r\right)$ curve to an exponential function, is denoted by the intersection of the dashed horizontal line at $g=e^{-1}$. Data shown are the mean and SEM across 36 images from two replicates. F Correlation lengths $\xi$ (open squares), FW1% (half-filled triangles), and median cluster size (filled circles, see Figs. S7 and S8), each normalized by their initial pT value, show that the time course of cluster assembly over 28 h for pT (gold) and pS (cyan) correlate with the fluorescence polarization (FP) of fluorescently labeled KaiB (right axis, mP), which serves as a proxy for KaiBC complex formation. Both the degree of clustering and FP remain at a minimum for pT, while for pS, both steadily increase for the first ~15 h. $\xi$, FW1%, and cluster size data are the mean and SEM across 36 images from two replicates. FP data are the mean and SEM (too small to see) of three replicates. Source data are provided as a Source Data file.

Fig. 4. Kinetic simulations of Kai-mediated crosslinking recapitulate slow formation of colloidal clusters.

A Simulation snapshots showing clustering of colloids (red circles) crosslinked by permanent bonds (blue lines), analogous to the experimental pS-colloid system, at 1 (left), 7 (inset), and 28 (right) hours. Colorized temporal projections of (B) simulation snapshots for colloids with permanent crosslinker bonds (permanent bonds, P) and (C) experimental snapshots for pS-colloids show similar features emerging over the course of a day. Colorized temporal projections of (D) simulation snapshots for colloids with no crosslinker bonds (no bonds, N) and (E) experimental snapshots for pT colloids both show minimal clustering or restructuring over the course of a day. Times and color-coding used in projections are the same as in Fig. 3, as indicated by the color scales. F $g\left(r\right)$ computed for simulation snapshots, taken at times specified in the legend, for colloids with no bonds (N, yellow squares) and permanent bonds (P, cyan triangles). Time course of the (G) correlation lengths $\xi$ and (H) colloid connectivity number CCN determined from simulations with permanent bonds (P, cyan) and no bonds (N, gold). I Multiple metrics of clustering and self-assembly resulting from permanent crosslinker bonding in experiments (pS) and simulations (P), each normalized by its maximum value to indicate the fractional clustering index (left axis) measured using each metric. Metrics include: experimental correlation lengths ($\xi$, open squares), simulated correlation lengths (sim $\xi$, filled squares), full width at 1% (FW1%, half-filled triangles), and median cluster size (cluster size, filled circles). Trends in both simulation and experimental data track with the time course of KaiB fluorescence polarization (right axis (mP), translucent triangles) in a reaction with pS KaiC. All simulation data shown is the mean and SEM across five replicates. Experimental data shown in (C, D, I) are reproduced from Fig. 3. Source data are provided as a Source Data file.

Fig. 5. Oscillatory material properties depend on crosslinker density.

A–C Simulations model oscillatory colloidal crosslinkers with different numbers of KaiB–KaiC complexes ($n$, bonds) participating in each connection between colloids. A Colloid connectivity (CCN) versus time for systems with different numbers of bonds per colloid connection, from light to dark gray: $n=$ 1, 4/3, 5/3, 2, 3. Arrow indicates direction of increasing $n$. Intermediate bond numbers 1 < $n<2$ result in oscillating connectivity, while $n=1$ is not sufficient for pronounced clustering and $n\ge 2$ promotes sustained cluster growth with minimal dissolution. B The fractional clustering index (see text) versus time for bond densities shown in (A) reveal oscillatory clustering for $n<2$ that is most pronounced for $n=5/3$. The colored boxes enclose the data points corresponding to the simulation images with color-matched borders shown in (C). C Simulation snapshots that correspond to troughs (red, orange) and peaks (green, blue) shown in (B) demonstrate that peaks and troughs correspond to substantial and minimal clustering, respectively. D Fluorescence polarization (FP) of KaiB (left axis, triangles) and percentage of phosphorylated KaiCs (%P, right axis, circles) during a KaiB–KaiC reaction. %P measurements were performed in the presence (filled circles) and absence (open circles) of colloids, showing that oscillatory KaiC phosphorylation dynamics are unaffected by the presence of colloids (see Fig. S9). E, F The fractional clustering index versus time for colloid experiments performed with KaiC concentrations of 6.67 µM (1$\times$, dark gray circles), 3.33 µM (0.5$\times$, gray squares), and 1.67 µM (0.25$\times$, light great diamonds) reveal oscillatory clustering for the lowest concentration, similar to the simulated $n=5/3$ case, while the two higher concentrations steadily become increasingly clustered over time, similar to the simulated $n\ge 2$ cases. Colored boxes enclose the data points corresponding to the microscope images with color-matched borders shown in (F). F Microscope images that correspond to troughs (red, orange) and peaks (green, blue) shown in (E) show strong similarities to simulated images and demonstrate minimal and substantial clustering, respectively. Simulated colloid connectivity (G) and fractional clustering index (H) for extended times (30–48 h) for the same bond numbers examined in (A–C). Simulation snapshots are shown for the $n=5/3$ case at 33 (blue), 39 (red), 45 (brown), and 48 (green) hours, as well as the $n=1$ and $n=3$ cases at 48 h (right column). Translucent boxes that match the snapshot borders indicate the corresponding connectivity and fractional clustering index. All simulated and experimental images shown are 50 µm $\times$ 50 µm. All experimental data shown is the mean and SEM across 36 images from two replicates. All simulation data shown is the mean and SEM across five replicates. Source data are provided as a Source Data file.