Life-like Supramolecular Materials
Serena De Piccoli1, Alessandro Sorrenti2, Jorge Leira-Iglesias, Akihiro Sato & Thomas M. Hermans1
1ISIS, Université de Strasbourg and CNRS (UMR 7006), Strasbourg, France 2ETH Zürich, Institute for Chemical and Bioengineering, Zürich, Switzerland
Supramolecular chemistry aims to develop complex structures through non-covalent intermolecular interactions.1 Living systems use fuel-driven supramolecular polymers to control important cell functions, operating far-from-equilibrium in so-called dissipative steady states (i.e. continuously consume energy to keep their structure and functions).2,3 Fuel molecules like ATP are used to control when and where such polymers should assemble and disassemble. Artificial fuel-driven polymers have been developed, but keeping them in sustained non-equilibrium steady state (NESS) has proven challenging. Assembly and disassembly of our polymer is regulated by phosphorylation and dephosphorylation, respectively. Waste products lead to inhibition, causing the reaction cycle to stop. To go beyond this limit, a membrane reactor has been built and so far, we have demonstrated that we are able to sustain NESS conditions. Right now, our main goal is to achieve enzymatic reaction cycles with designed feedback capable of keeping NESS. The general idea is to have a hydrogel material consisting of supramolecular polymers that are in sustained non-equilibrium states, and that therefore are continuously processing their respective reaction cycles. The latter will give the materials unique properties not currently found in artificial materials.
Figure 4. Enzyme-controlled supramolecular polymerization from stepwise to steady states. (a) Peptide–perylenediimide derivative PDI (half is shown) is phosphorylated on the serine residue by protein kinase A (PKA) to give monophosphorylated p-PDI, and further diphosphorylated p2-PDI, fuelled by ATP to ADP hydrolysis (one eq. per phosphate introduced). Phosphate hydrolysis (scissors) by l-protein phosphatase (lPP) yields inorganic phosphate Pi as waste. Both PDI and p2-PDI can self-assemble to form equilibrium supramolecular polymers. PKA has three binding sites: for ATP (blue), for the LRRASL peptide (green) and for LRRApSL (red). (b) Principle of operation and CAD design
of the continuous flow device based on a clamped dialysis cassette. The average gap between the membranes d1 is 1.9 mm, the width of the flow chambers d2 5.1 mm. (c) Different NESS (plateau regions, no. 1-4), characterized by different molar fractions of the three species PDI, p-PDI and p2-PDI (by LC–MS analysis), can be obtained using the continuous flow device, depending on the influx of the fuel ATP.
References 1. Lehn, J.-M. Supramolecular Chemistry: Concepts and Perspectives. (Wiley, 1995). 2. Fialkowski, M. et al. Principles and Implementations of Dissipative (Dynamic) Self-Assembly. J. Phys. Chem. B 110, 2482–
2496 (2006). 3. Whitesides, G. M. Self-Assembly at All Scales. Science 295, 2418–2421 (2002). 4. Sorrenti, A., Leira-Iglesias, J., Sato, A., Hermans, T. M. Non-equilibrium steady states in supramolecular polymerization.
Nature Comm. 8, 15899 (2017)
plateau at 1.65 after B100 min (Fig. 2b). LC–MS analysis of thesolution at the plateau confirmed the quantitative conversionof PDI to p2-PDI. The observed increase of A504 nm/A540 nm,and the overall decrease in intensity (Fig. 2a) indicates thatphosphorylation promotes further aggregation18,24 as well asgrowth of the aggregates into bigger structures (see below).A negative control where only ATP was added did not result inany change with time of A504 nm/A540 nm.
The mechanism of supramolecular polymerization forboth PDI and p2-PDI was studied by temperature-dependentultraviolet–visible experiments (see Supplementary Discussion).The latter showed in both cases a continuous decrease of the ratioA504 nm/A540 nm upon heating from 283 to 368 K, indicative ofdisassembly (Supplementary Fig. 10a). The observed spectralchanges were reversible upon consecutive heating–coolingcycles (at a rate of 1 K min! 1), suggesting that the polymersare equilibrium structures with exchange dynamics on the secondtimescale. The A504 nm/A540 nm versus temperature curves could bedescribed with an isodesmic (equal-K) polymerization model25,26
for both PDI and p2-PDI. This analysis (SupplementaryDiscussion/Supplementary Table 1) shows that p2-PDI polymers
(Keq (298 K)¼ 3.8# 104 M! 1, Tm (230mM)¼ 368.9±0.1 K) aremore stable as compared with PDI polymers (Keq (298 K)¼2.1# 104 M! 1, Tm (230mM)¼ 358.5±0.1 K). Perylenediimidederivatives are well known for their propensity to assemble intoone-dimensional columnar aggregates18,19,27. In the case ofp2-PDI we could observe bundles of long fibrous aggregates byatomic force microscopy (mm-sized, 1±0.2 nm height for thesmallest fibres) in samples drop cast on mica (Fig. 2d andSupplementary Fig. 11). Shorter fibres could also be imagedby transmission electron microscopy (Supplementary Fig. 12).However, our main focus was on assessing the changes inducedby phosphorylation on the supramolecular polymer in thereaction buffer. To this end, we performed dynamic light-scattering (DLS) measurements that confirmed the increase inthe polymer size upon phosphorylation. As shown in Fig. 2c,the normalized cross-correlation function g(1)(t) of p2-PDI(red squares) is clearly shifted to higher lag time t whencompared with that of PDI (black circles), corresponding tohydrodynamic radii RH of 690±40 and 440±10 nm for p2-PDIand PDI, respectively (at 1 mM). In addition, the correspondingrelaxation G showed a slope of 2 versus scattering vector squared
x x
O
HN N
H
HN N
H
HN N
H
O
NH
NH2
O
+
+
O
OO
O
HN
H2 N
OHN
2O
OP
O
O––ONH2
H2 N
N
O
O
++ +
O
+
+
HN N
H
HN N
H
HN N
H
O
NH
NH2
O
O
OO
O
HN
H2 N
OHN
2O
OH
H2 N
NH2
N
O
O
+te te
d
t0
ATP
Transient state
t0
ATP
Steady states
tet0
PKAATP λPP
Stimuli responsive
a
b c
ATP
Continuousinflux
Batchaddition
PKAdomains
λPP domain
LRRApSLbinding
LRRASLbinding
ATPbinding
LRRApSLbinding
p-PDI
H2OPiλPP
H2OPiλPP
PKAADPATP
PKAADPATP
Keq K ′eq
p2-PDIPDI
Figure 1 | Enzyme-controlled supramolecular polymerization from stepwise to steady states. (a) Peptide–perylenediimide derivative PDI (half is shown)is phosphorylated on the serine residue by protein kinase A (PKA) to give monophosphorylated p-PDI, and further diphosphorylated p2-PDI, fuelled byATP to ADP hydrolysis (one eq. per phosphate introduced). Phosphate hydrolysis (scissors) by l-protein phosphatase (lPP) yields inorganic phosphate Pias waste. Both PDI and p2-PDI can self-assemble to form equilibrium supramolecular polymers. PKA has three binding sites: for ATP (blue), for the LRRASLpeptide (green) and for LRRApSL (red). (b ) Stimuli responsiveness: the addition of ATP and PKA to a solution of PDI results in p2-PDI and a consequentchange of the supramolecular structure of the polymer. A second stimulus, that is, the addition of lPP, is needed to reset the polymer to its originalnonphosphorylated state. (c) Transient state: a single input, that is, the addition of ATP to a solution of PDI, in the presence of PKA and lPP, leads to atransient change of the supramolecular structure and chirality of the polymer. (d) Supramolecular non-equilibrium steady states (NESS). The system is keptin a dissipative steady state by continuous influx of ATP. Depending on the level of the chemical fuel supplied different dissipative steady states can beaccessed.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15899 ARTICLE
NATURE COMMUNICATIONS | 8:15899 | DOI: 10.1038/ncomms15899 | www.nature.com/naturecommunications 3
enzymes PKA and lPP, are compartmentalized inside thecassette, whereas ATP and the waste (ADP, Pi) can easily passthrough the membrane (Fig. 4a). As a result, the fuel can beadded to (and waste removed from) the small fluid volume(B400 ml) within the two membranes by continuously flushingthe flow chambers with fresh ATP solutions.
Using LC–MS we tracked PDI, p-PDI and p2-PDI bysampling the solution in the cassette at different times(Supplementary Discussion). Figure 4c shows that, starting froma solution containing only PDI polymers, we could get a firstNESS characterized by the presence of all three species PDI,p-PDI and p2-PDI (mole fractions B0.35, 0.45 and 0.20,respectively) by constantly flowing 1 mM ATP (SupplementaryFig. 19). Remarkably, we could sustain such a state for 420 h(plateau no. 1 in Fig. 4c), that is, for as long as we kept the influxof ATP (and outflux of ADP and Pi) constant. We have to pointout that during this time the phosphorylation/dephosphorylationreactions of the PDI derivatives are continuously occurring whileconsuming fuel. When increasing the ATP concentration to2.5 mM (second tap in Fig. 4c), the system responded by reachinga different NESS (plateau no. 2 in Fig. 4c) characterized by mostlyp2-PDI polymers (Supplementary Fig. 20). From this point,we switched to flow only buffer without ATP (third tap att¼ 3,170 min in Fig. 4c), and as a consequence the system relaxedback to the nonphosphorylated state, that is, mostly PDI present(plateau no. 3 in Fig. 4c, and Supplementary Fig. 21). In fact,the absence of ATP causes the phosphorylation to stop, while the
dephosphorylation keeps progressing. The latter demonstratesthat a continuous influx of ATP (and removal of waste) is indeedneeded to maintain a phosphorylated NESS, which is acontinuously dissipating state. Finally, upon flowing 5 mM ATP(fourth tap at t¼ 4,440 min in Fig. 4c) we could once againstabilize a NESS characterized by p2-PDI polymers (plateau no. 4in Fig. 4c and Supplementary Fig. 22), thus demonstrating thatour supramolecular system is perfectly reversible when operatingin open flow conditions, and it does not suffer from poisoning.In other words, by modulating the influx of ATP, we maintainour supramolecular system in distinct steady states driven byfuel-consuming enzymatic (de)phosphorylation. By the reactionrates obtained from the model, we can estimate that eachserine cycles on average 8 times between phosphorylated anddephosphorylated states during the entire experiment shown inFig. 4c (6,000 min).
We want to stress that our NESS conditions are not justshifting coupled chemical equilibria as described by Le Chatelier’sprinciple (cf., red crosses in Fig. 1a). Instead, we have threespecies that cannot equilibrate, but can only interconvert in thepresence of the enzymes and chemical fuels. More generally, if afuelled self-assembly process has fast assembly kinetics comparedwith reaction kinetics, the assemblies are in a local equilibriumstate (seemingly obeying Le Chatelier’s principle). That is, thefuelled reactions change the average monomer concentration, butthe assemblies themselves are determined by the thermodynamicsparameters (that is, Gibbs energy, concentration, temperature
ATP
ADP
ATP
ADP
ATP
H2O
Waste
Dialysismembranes
ATP
AD
P
Waste
ATP
Flowchamber
Flowchamber
Dialysiscassette
QuartzwindowQuartz
window
Mol
e fr
actio
n
Time (min)
PDI
p2-P
DI
p-PDI
1 mM ATP
2.5 mMATP
No ATP5 mM ATP
a b
c
d1 d2d2
Pi Pi
Pi
#1
#2
#3
#4
1.0
0.8
0.6
0.4
0.2
0.0
0 2,000 4,000 6,000
Figure 4 | Continuous flow device and supramolecular non-equilibrium steady states. (a) Principle of operation and CAD design of the continuous flowdevice based on a clamped dialysis cassette. The average gap between the membranes d1 is 1.9 mm, the width of the flow chambers d2 5.1 mm. (b) Frontand side view of the three-dimensionally (3D) printed device, also showing the inlet and outlet needles inserted through the silicon spacers. (c) DifferentNESS (plateau regions, no. 1"4), characterized by different molar fractions of the three species PDI, p-PDI and p2-PDI (by LC–MS analysis), can beobtained using the continuous flow device, depending on the influx of the fuel ATP. Solid lines are drawn to guide the eye. The taps are placed incorrespondence of the time(s) at which we changed the ATP concentration of the solution that is continuously flowed through the lateral flow chambers.Namely, at t¼ 50 min we started to flow ATP 1 mM, at t¼ 1,770 min we switched to ATP 2.5 mM, at t¼ 3,170 min we started to flow buffer without ATPand at t¼4,440 min we switched to ATP 5 mM. For all NESS experiments: 0.085mM PKA and 0.097mM lPP. The error bars were set to 6% of thecorresponding value. The latter is the maximum error observed in the determination of the molar fraction when injecting 3 times in LC–MS solutions ofknown composition.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15899
6 NATURE COMMUNICATIONS | 8:15899 | DOI: 10.1038/ncomms15899 | www.nature.com/naturecommunications
enzymes PKA and lPP, are compartmentalized inside thecassette, whereas ATP and the waste (ADP, Pi) can easily passthrough the membrane (Fig. 4a). As a result, the fuel can beadded to (and waste removed from) the small fluid volume(B400 ml) within the two membranes by continuously flushingthe flow chambers with fresh ATP solutions.
Using LC–MS we tracked PDI, p-PDI and p2-PDI bysampling the solution in the cassette at different times(Supplementary Discussion). Figure 4c shows that, starting froma solution containing only PDI polymers, we could get a firstNESS characterized by the presence of all three species PDI,p-PDI and p2-PDI (mole fractions B0.35, 0.45 and 0.20,respectively) by constantly flowing 1 mM ATP (SupplementaryFig. 19). Remarkably, we could sustain such a state for 420 h(plateau no. 1 in Fig. 4c), that is, for as long as we kept the influxof ATP (and outflux of ADP and Pi) constant. We have to pointout that during this time the phosphorylation/dephosphorylationreactions of the PDI derivatives are continuously occurring whileconsuming fuel. When increasing the ATP concentration to2.5 mM (second tap in Fig. 4c), the system responded by reachinga different NESS (plateau no. 2 in Fig. 4c) characterized by mostlyp2-PDI polymers (Supplementary Fig. 20). From this point,we switched to flow only buffer without ATP (third tap att¼ 3,170 min in Fig. 4c), and as a consequence the system relaxedback to the nonphosphorylated state, that is, mostly PDI present(plateau no. 3 in Fig. 4c, and Supplementary Fig. 21). In fact,the absence of ATP causes the phosphorylation to stop, while the
dephosphorylation keeps progressing. The latter demonstratesthat a continuous influx of ATP (and removal of waste) is indeedneeded to maintain a phosphorylated NESS, which is acontinuously dissipating state. Finally, upon flowing 5 mM ATP(fourth tap at t¼ 4,440 min in Fig. 4c) we could once againstabilize a NESS characterized by p2-PDI polymers (plateau no. 4in Fig. 4c and Supplementary Fig. 22), thus demonstrating thatour supramolecular system is perfectly reversible when operatingin open flow conditions, and it does not suffer from poisoning.In other words, by modulating the influx of ATP, we maintainour supramolecular system in distinct steady states driven byfuel-consuming enzymatic (de)phosphorylation. By the reactionrates obtained from the model, we can estimate that eachserine cycles on average 8 times between phosphorylated anddephosphorylated states during the entire experiment shown inFig. 4c (6,000 min).
We want to stress that our NESS conditions are not justshifting coupled chemical equilibria as described by Le Chatelier’sprinciple (cf., red crosses in Fig. 1a). Instead, we have threespecies that cannot equilibrate, but can only interconvert in thepresence of the enzymes and chemical fuels. More generally, if afuelled self-assembly process has fast assembly kinetics comparedwith reaction kinetics, the assemblies are in a local equilibriumstate (seemingly obeying Le Chatelier’s principle). That is, thefuelled reactions change the average monomer concentration, butthe assemblies themselves are determined by the thermodynamicsparameters (that is, Gibbs energy, concentration, temperature
ATP
ADP
ATP
ADP
ATP
H2O
Waste
Dialysismembranes
ATP
AD
P
Waste
ATP
Flowchamber
Flowchamber
Dialysiscassette
QuartzwindowQuartz
window
Mol
e fr
actio
n
Time (min)
PDI
p2-P
DI
p-PDI
1 mM ATP
2.5 mMATP
No ATP5 mM ATP
a b
c
d1 d2d2
Pi Pi
Pi
#1
#2
#3
#4
1.0
0.8
0.6
0.4
0.2
0.0
0 2,000 4,000 6,000
Figure 4 | Continuous flow device and supramolecular non-equilibrium steady states. (a) Principle of operation and CAD design of the continuous flowdevice based on a clamped dialysis cassette. The average gap between the membranes d1 is 1.9 mm, the width of the flow chambers d2 5.1 mm. (b) Frontand side view of the three-dimensionally (3D) printed device, also showing the inlet and outlet needles inserted through the silicon spacers. (c) DifferentNESS (plateau regions, no. 1"4), characterized by different molar fractions of the three species PDI, p-PDI and p2-PDI (by LC–MS analysis), can beobtained using the continuous flow device, depending on the influx of the fuel ATP. Solid lines are drawn to guide the eye. The taps are placed incorrespondence of the time(s) at which we changed the ATP concentration of the solution that is continuously flowed through the lateral flow chambers.Namely, at t¼ 50 min we started to flow ATP 1 mM, at t¼ 1,770 min we switched to ATP 2.5 mM, at t¼ 3,170 min we started to flow buffer without ATPand at t¼4,440 min we switched to ATP 5 mM. For all NESS experiments: 0.085mM PKA and 0.097mM lPP. The error bars were set to 6% of thecorresponding value. The latter is the maximum error observed in the determination of the molar fraction when injecting 3 times in LC–MS solutions ofknown composition.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15899
6 NATURE COMMUNICATIONS | 8:15899 | DOI: 10.1038/ncomms15899 | www.nature.com/naturecommunications
a
c
b