Super-resolution microscopy reveals structural diversity in molecular exchange among peptide amphiphile nanofibres

Abstract

The dynamic behaviour of supramolecular systems is an important dimension of their potential functions. Here, we report on the use of stochastic optical reconstruction microscopy to study the molecular exchange of peptide amphiphile nanofibres, supramolecular systems known to have important biomedical functions. Solutions of nanofibres labelled with different dyes (Cy3 and Cy5) were mixed, and the distribution of dyes inserting into initially single-colour nanofibres was quantified using correlative image analysis. Our observations are consistent with an exchange mechanism involving monomers or small clusters of molecules inserting randomly into a fibre. Different exchange rates are observed within the same fibre, suggesting that local cohesive structures exist on the basis of β-sheet discontinuous domains. The results reported here show that peptide amphiphile supramolecular systems can be dynamic and that their intermolecular interactions affect exchange patterns. This information can be used to generate useful aggregate morphologies for improved biomedical function.

ARTICLE

Received 30 Nov 2015 | Accepted 8 Apr 2016 | Published 19 May 2016

Ricardo M.P. da Silva1,2,3, Daan van der Zwaag2, Lorenzo Albertazzi2,4, Sungsoo S. Lee5, E.W. Meijer2

& Samuel I. Stupp1,5,6,7,8

The dynamic behaviour of supramolecular systems is an important dimension of their potential functions. Here, we report on the use of stochastic optical reconstruction microscopy to study the molecular exchange of peptide amphiphile nanobres, supramolecular systems known to have important biomedical functions. Solutions of nanobres labelled with different dyes (Cy3 and Cy5) were mixed, and the distribution of dyes inserting into initially single-colour nanobres was quantied using correlative image analysis. Our observations are consistent with an exchange mechanism involving monomers or small clusters of molecules inserting randomly into a bre. Different exchange rates are observed within the same bre, suggesting that local cohesive structures exist on the basis of b-sheet discontinuous domains.

The results reported here show that peptide amphiphile supramolecular systems can be dynamic and that their intermolecular interactions affect exchange patterns. This information can be used to generate useful aggregate morphologies for improved biomedical function.

1 Simpson Querrey Institute for BioNanotechnology (SQI), Northwestern University, Chicago, Illinois 60611, USA. 2 Laboratory of Macromolecular and Organic Chemistry and Institute for Complex Molecular Systems, Eindhoven University of Technology, Eindhoven MB 5600, The Netherlands. 3 Craniofacial Development & Stem Cell Biology, Kings College London, London, SE1 9RT, UK. 4 Nanoscopy for Nanomedicine Group, Institute for Bioengineering of Catalonia (IBEC), Barcelona 08028, Spain. 5 Department of Materials Science and Engineering, Northwestern University, Evanston, Illinois 60208, USA.

6 Department of Chemistry, Northwestern University, Evanston, Illinois 60208, USA. 7 Department of Medicine, Northwestern University, Chicago, Illinois 60611, USA. 8 Department of Biomedical Engineering, Northwestern University, Evanston, Illinois 60208, USA. Correspondence and requests for materials should be addressed to E.W.M. (email: mailto:[email protected]

Web End [email protected] ) or to S.I.S. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

DOI: 10.1038/ncomms11561 OPEN

Super-resolution microscopy reveals structural diversity in molecular exchange among peptide amphiphile nanobres

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561

Reversible supramolecular interactions are ubiquitous in nature, controlling the self-assembly of ordered functional structures that need to be dynamic to perform their

biological functions. One-dimensional cytoskeletal laments such as actin and tubulin are typical examples of structures that use dynamics to mediate the adaptive behaviour of cells, resulting in cell motility, shape change, cell division, signalling and muscular contraction at larger length scales16. Articial supramolecular materials could offer this bio-inspired dynamic behaviour, thus allowing enhanced interaction with natural systems and increased biomedical functionality. Since many natural processes are carefully regulated, optimization of an articial supramolecular material requires a detailed understanding of its dynamic properties, for example its exchange kinetics.

Molecular mixing experiments typically assess exchange kinetics by utilizing Frster resonance energy transfer (FRET) between a pair of donor and acceptor uorophores7,8, and alternatively, radio-labelled molecules9 or time-resolved small-angle neutron scattering10,11. Although these ensemble experiments can provide the timescale of the processes, they cannot distinguish different mechanisms. Moreover, they fail to detect the structural diversity among bres, or within an individual bre, for example, the occurrence of segregated domains. Local variations of molecular composition can have important biological implications, as they can greatly inuence the signalling potency through multivalency effects1214, and thus understanding the exchange heterogeneity is important to design materials in which function is connected with dynamics for adaptive or responsive behaviour15.

Super-resolution techniques are powerful tools to reveal the spatial distribution of molecules at the nanoscale, but these techniques have thus far been mainly applied to imaging ne details of cellular structures16,17. For instance, a resolution of B20 nm can be achieved using stochastic optical reconstruction microscopy (STORM)18, which is an order of magnitude below the diffraction limit and near the molecular scale. The enhanced resolution is achieved through the accurate localization of single uorescent molecules; to identify individual uorophores only a sparse subset of labels should be active at any given time. This sparse population of uorophores is obtained using probes that can be photo-switched to a temporary non-uorescent off state by light. By repeatedly activating different subsets and overlaying the resulting localizations, an image can be reconstructed. We have previously reported how to apply STORM to probe the dynamics of supramolecular bres19. Using two-colour STORM and quantitative image analysis, we were able to resolve the monomer distribution along the bre backbone. By following the monomer distribution during the molecular exchange process, we were able to infer the exchange mechanism.

Peptide amphiphiles (PAs) that self-assemble into high aspect ratio objects offer exciting opportunities for regenerative medicine and other therapeutic applications2024. As illustrated in Fig. 1a, this class of molecules is composed of an unbranched alkyl chain linked to a peptide segment, which can be further subdivided in several domains. A segment with propensity to form b-sheets is conjugated to the alkyl tail, followed by a charged segment for solubility. Additional domains can be conjugated to the canonical structure to introduce biofunctionality in the nanobres. Whereas the hydrophobic collapse of the aliphatic tails induces self-assembly, experimental25,26 and theoretical27 evidence suggests that the formation of directional hydrogen bonds within the b-sheet domain is an additional important component of the driving force for assembly of the molecules into one-dimensional lamentous shapes. The facile incorporation of multiple bioactive signals at controlled concentrations14,28, together with their structural resemblance with extracellular

matrix bres makes PA assemblies useful as bioactive articial extracellular matrix components for cell signalling. Furthermore, they are also intrinsically biocompatible and biodegradable and can therefore disappear easily after fullling their biological functions. PAs have been extensively studied as a platform for applications that include bone, cartilage, enamel, neuronal regeneration, angiogenesis for ischaemic disease, targeted drug delivery and cancer therapeutics2024,29.

Ensemble measurements of PA dynamics revealed partial self-healing using rheological techniques30 or the existence of a critical aggregation concentration31, while kinetic spectroscopy showed pathway selection of PAs into different morphologies32,33. However, these approaches do not consider structural heterogeneity and its effect on dynamic molecular exchange at the level of individual laments. In this paper, we use STORM to image individual PA nanobres, addressing the distribution of molecules along the bre during exchange and therefore analysing diversity among supramolecular nanobres.

ResultsNanobre design and preparation. The PA molecule studied in this work is shown in Fig. 1a. It consists of a palmitic acid tail, six alanines in the b-sheet-forming region, followed by three glutamic acids as charged solubilizing moieties. This PA molecule self-assembles in water, forming nanobres around 7 nm in diameter and lengths in the range of micrometres as observed by cryogenic transmission electron microscopy (cryoTEM; Fig. 1e). Amino acids with different propensities to form b-sheets affect considerably the internal order of PA assemblies, as well as nanobre stiffness26. Since alanine has a weaker tendency to form b-sheets than valine34, placing this amino acid in the b-sheet-forming segment was expected to yield relatively dynamic PA bres. Circular dichroism spectroscopy of this PA is consistent with the typical b-sheet conformation found for other PAs26, at physiological pH and ionic strength (Fig. 1c). To perform imaging by uorescence microscopy, PA molecules were labelled with cyanine dyes, namely Cy3 and Cy5 (Fig. 1b). Water soluble sulfonated dyes with a net charge of 1 were used to preserve

the amphiphilic asymmetry of the PA molecule and, since sulfonate groups are fully ionized in the vicinity of physiological pH, their behaviour is insensitive to pH in the range of interest. The Cy3 and Cy5 dyes have been chosen for their suitable photo-physical properties for STORM imaging; moreover, they constitute a good FRET pair with a Frster radius of 50 (ref.35). Single-colour labelled nanobres were created by mixing a stock solution of either Cy3-PA or Cy5-PA with a stock solution of unlabelled PA, lyophilization, brief co-dissolution in triuoroacetic acid (TFA) and immediate TFA removal, lyophilization and nal reconstitution in the aqueous working solution to form labelled supramolecular aggregates. The degree of labelling was accurately controlled by premixing the different PAs at the desired ratios. Fluorescently labelled nanobres revealed a morphology that was indistinguishable from their non-labelled counterparts, as shown by cryoTEM (Fig. 1f).

Ensemble molecular exchange kinetics. The timescale of molecular exchange in PA-based nanobres was measured using FRET kinetic experiments. Two sets of PA nanobres were separately pre-assembled from a mixture of non-labelled PAs with either Cy3-PA (0.5%) or Cy5-PA (0.5%), as illustrated in Fig. 2a. Next, the two solutions were mixed and the FRET ratio, dened as the ratio between the uorescence intensities of Cy5 acceptor and Cy3 donor, was monitored over time. As shown in Fig. 2b, the FRET ratio increases with time, reaching a plateau after several hours. This means that PA molecules are able to migrate between

2 NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561 ARTICLE

a

O

OH

HN H

N

H N

O

O O H N

HO

HO

H N

O

NH N

H

N H

NH O

O O

NH2

SO3H N

O

13 O

O

NH O

O3S

+N

b

O

Cy3

O

OH

R

SO3H

O

HN H

N

O

H N

H N

H N

O

O

O

O

R2: O3S

+N

Cy5

N H

NH2

N H

O

13

O

N H

O

N H

N H

O

O

O

N

HO

O

HO

O

R1:

c d

2 dmol1 103 )

300

60 660 650

Maximum intensity (a.u.)

40 20

200

[afii9838] max(nm)

640

[[afii9835]] MRE (deg cm

0 20

630 620

200 220 260

240 103

100

0

101 100 101 102

[Peptide] (M)

Wavelength (nm)

e f

g h

Figure 1 | PA self-assembly. Molecular structure of (a) non-labelled PA and (b) PA molecules labelled with photo-switchable sulfonated cyanine dyes, namely Cy3 (green) and Cy5 (red). (c) Circular dichroism spectrum and (d) Nile Red assay of non-labelled PA. CryoTEM images of nanobres self-assembled at pH 7.5 and NaCl 150 mM (e) from non-labelled PA alone and (f) from a molecular mixture of non-labelled and Cy5-labelled PAs (scale bar, 200 nm). Diffraction-limited uorescence microscopy images of Cy3- (g) and Cy5-labelled (h) PA nanobres (scale bar, 1 mm).

nanobres, resulting in mixed uorophore bres. Figure 2b depicts kinetic experiments at different temperatures, showing that the exchange rate is remarkably faster at 37 C than at 20 C. On the other hand, we observed a limited effect of concentration

on the exchange rate (Fig. 2c), proving that the exchange process is not diffusion limited in this concentration range. These results resemble observations on self-assembled polymeric micelles, in which unimer expulsion and insertion is thought to be the

NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561

a b

a b

Green fibre

10

8

Red fibre

0.4

0.3

FRET ratio

10

8

37 C

20 C

6

6 4 2 0

Y(m)

Y(m)

4

0 4 8 12 16

20

Time (h)

c

d

20 2

0.4

0.2

0.1

0.0

0.3

4

6

8 10

4

6

8 10

FRET ratio

X (m)

2 X (m)

2

0.5 mM

1 mM

2 mM

5 mM

FRET ratio

0.2

0.1

0.0

0.30.20.10.0

c

d

0.5 mM

1 mM

2 mM

5 mM

20

Counts

20

Counts Autocorrelation

10

10

0 4 8 12 16 20

Time (h)

0.0 0.5 1.0

Time (h)

2

4

0 0 1 3

Arclength (m)

Radius (nm) Radius (nm)

0 0 1 3

Arclength (m)

4

Figure 2 | Molecular exchange by FRET. (a) Schematic representation of a molecular exchange kinetic measurement. The molecular exchange progress over time is estimated by means of FRET ratio (dividing Cy5 by Cy3 uorescence intensities), either at (b) a constant PA concentration(0.5 mM) or at (c,d) a constant temperature (37 C), where d shows the FRET ratio for short timescale.

e

f

Autocorrelation

0.4

0.8

0.6

0.4

0.2

0.0

0.0

0 250 500 750 1,000

0 250 500 750 1,000

rate-determining step of system dynamics10,11, as well as other synthetic supramolecular polymers7. So, while FRET measurements provide useful information about the timescale of the exchange, FRET is an ensemble technique, and it does not provide the spatially resolved information required to elucidate the mechanism of exchange.

Imaging PA nanobres. Due to its high spatial resolution and multicolour imaging ability, STORM microscopy can be used to investigate the spatial details of the exchange process in PA nanobres. Cy3- or Cy5-labelled PA nanobres have been immobilized on a glass slide by physisorption and subsequently have been imaged using STORM (see the Methods Section for details). The resulting images show well-reconstructed aggregates (Fig. 3a,b), indicating that both dyes display appropriate photo-switching behaviour when associated to PA structures. Since the actual width of the nanobre is below the STORM resolution, the apparent thickness of the bre can be used to provide an estimate for the experimental resolution of these measurements, approximately equal to 50 nm (Supplementary Fig. 1). The bre characteristics observed by STORM, for example, rigidity and bre length, match those reported by the cryoTEM images in Fig. 1e,f.

To analyse the monomer distribution inside PA nanobres, which is crucial to understand the exchange mechanism, we utilized an image analysis routine previously developed in the Meijer laboratory19. This method removes background localizations, identies the contour of the bre backbone to study its mechanical properties (Supplementary Fig. 2) and computes the localization density along the polymer. Figure 3c,d display the localization density plots for Cy5- and Cy3-labelled PA, respectively. These proles contain information about the distribution of the dye-functionalized PA molecules in a nanobre. As can be clearly observed, the number of localizations shows uctuations along the bre. These uctuations can be attributed to two causes: (i) a heterogeneous distribution of monomers or (ii) the stochastic processes taking place during STORM image acquisition, for example, uorophore blinking36 and uorophore bleaching. Therefore, variations in the density of localizations cannot be unequivocally attributed to changes in local monomer concentration. To address this issue,

we use a spatial autocorrelation algorithm, a powerful method to investigate the distribution of these localizations. In purely random distributions, localizations at distance r apart are not correlated with each other, by denition giving an autocorrelation g(r) with a value of zero. Positive values of g(r) represent an increased probability that a second localization is found at a certain distance (r) from a rst given localization, while negative values indicate a decreased probability. The autocorrelation curves of bres in single-colour samples are shown in Fig. 3e,f. The autocorrelation of Cy5-labelled nanobres (Fig. 3f) is strongly positive in the short range (o100 nm) and zero for longer distances. This short-ranged contribution is the typical signature for multiple localizations of the same dye (overcounting) and is present in all STORM images. As previously described, the overcounting contribution in the spatial autocorrelation traces can be modelled using equation (1) (ref. 37),

g r

A1exp r2=4s2

Figure 3 | STORM analysis of single-colour nanobres. PA nanobres were labelled with 5 mol% of either Cy3-PA (a,c,e) or Cy5-PA (b,d,f) before the molecular exchange experiment. Localization maps after applying a clustering algorithm, background cleaning and backbone nding (a,b). Each bre localization distribution prole was determined along the bre backbone (arc length) using a bin size of 25 nm (c,d). Averaged autocorrelation (e,f) was computed from distribution proles of a large set of bre images (nZ19). Solid lines correspond to model ttings (equation (1) for red channel and equation (2) for green channel) and error bars represent 95% condence level.

1 where the standard deviation (s) provides an estimate of the resolution achieved. Equation (1) provides a good t to the autocorrelations of Cy5-labelled single-colour nanobres (Fig. 3f). Therefore, the localization density uctuations visible in the STORM images for these PA nanobres (Fig. 3d) are due to stochastic processes inherent to the technique, while the monomer distribution is homogeneous. On the other hand, the

4 NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561 ARTICLE

autocorrelation function for the Cy3-channel (Fig. 3e) is not well-described by Equation (1), since the Cy3-functionalized bres display a consistent anti-correlation (g(r)o0) at intermediate distances (B500 nm). The autocorrelation behaviour of these bres was correctly described using a micro-emulsion model, that accounts for the existence of microdomains richer in the uorescent monomer, according to equation (2) (ref. 37):

g r

A1exp r2=4s2

A2cos pr=2r0

exp r=a

2 The rst term of this equation is equal to equation (1), since the overcounting phenomenon is still present. In the second term, the parameter r0 is the average size of the microdomains and a is the coherence length of the domains. As dened by equation (2), microdomain corresponds to an extended bre region that is enriched in uorescently labelled molecules compared with the bre average, thus increasing the likelihood of nding those molecules in that particular region. In this context, microdomain does not mean that molecules are completely clustered and conned to those regions.

An improvement in the goodness of the t was not observed for the Cy5-labelled nanobres, therefore it is reasonable to adopt the simpler model with less parameters, which excludes the existence of microdomains. On the other hand, in the case of Cy3-labelled nanobres a considerable improvement on the w2-test (43.5 times lower) is obtained for the model with the microdomain component, compared with the model that only considers the contribution of the overcounting. A microdomain size of B300 nm and coherence length of B500 nm were obtained from the t (Fig. 3e), showing a regular uctuation of localization density in the Cy3-bres at the submicron length scale. The difference observed between the distributions of Cy3 and Cy5-PAs is surprising, because of the great chemical and structural resemblance of these dyes. These results show that a small change in the PA structure may have noticeable effects on the self-assembly process, and that the STORM-based autocorrelation analysis can be used to investigate structure at the single-aggregate level. In addition, it shows that Cy5-functionalized PA is suitable for use in quantitative exchange experiments, since the Cy5 dye does not affect the molecular distribution of the PA monomers.

Fibre bundling and multicolour imaging. Since the nal morphology in supramolecular polymers is highly sensitive to not only the molecular structure, but also the polymerization conditions, it is very important to design a correct sample preparation protocol to obtain the desired structure. A typical problem that may occur is the aggregation of bres into bundles, a phenomenon that is hard to detect using traditional techniques. Avoiding bre bundling is crucial when investigating molecular mixing, because bundles of nanobres with different colours confound the observation of exchanging monomers. In this work we show that STORM-based analysis can be applied to detect PA bre bundling. When nanobre preparation was performed at higher PA concentration ([PA]nal41 mM), extremely long and curved

brillar structures were observed in diffraction-limited uorescence microscopy (Supplementary Fig. 3). However, the increased resolution attained using STORM imaging revealed ner details of the adsorbed structures, showing extensive bre bundling. What appeared to be curvature at low resolution was actually the intersection of smaller stiff bres with different orientations. Two-colour STORM has been performed on PA samples of different concentrations and ionic strengths to unambiguously prove the presence or absence of bundled bres. Two aqueous solutions of PA nanobres labelled with either Cy3 or Cy5 were mixed together and immediately adsorbed onto a glass coverslip at room

temperature, freezing all kinetics. For this short mixing timescale (o1 min), molecular exchange is negligible according to the

FRET measurements shown in Fig. 2b. Therefore, single bres should be either fully green or fully red. If bres are adsorbed at high ionic strength and low concentration (o0.1 mM), this behaviour was indeed observed (Fig. 4a). On the contrary, as shown in Fig. 4b, bres adsorbed at concentrations higher than 1 mM can be observed in both channels simultaneously, indicating bundles of PA nanobres. It is also possible to observe that some of the bres visible in the green channel are not perfectly aligned with the overlapping bres visible in the red channel (Fig. 4b). This provides further evidence that at this concentration bundles are observed instead of fully mixed bres. Therefore, STORM allows us to verify the absence of bre bundling and therefore select the optimal sample preparation to perform the molecular exchange experiments.

Molecular exchange kinetics and mechanisms. The ability of STORM to image individual nanobres with high resolution and to study the distribution of different molecular species inside aggregates (vide supra), makes it the perfect tool to study exchange in PA samples. We prepared samples for these experiments in similar manner to the FRET measurements. First, two sets of single-colour PA nanobres with either Cy3-PA (5%) or Cy5-PA (5%) were separately preassembled in aqueous buffer solution, followed by a 16-h aging period. Subsequently the two aged solutions were brought to 37 C and mixed, allowing molecular exchange. Aliquots of the solution were withdrawn over the course of 48 h and nanobres were adsorbed onto a glass

Bundled fibres

a b

Single fibres

10

8 6 4

10

8 6 4

2 2

Y(m)

Y(m)Y(m)

Cy3 channel Cy3 channel

Cy5 channel Cy5 channel

Merged Merged

2 4 6 8 10

Y(m)

10

8 6 4 2

Y(m)

10

8 6 4 2

10

6 4 2 0

10

6 4 2 0

8

8

Y(m)

2 4 6 8 10

X (m)

X (m)

Figure 4 | Single bre attachment. STORM images (with background) of PA nanobres attached at a nal PA concentration of either (a) 0.1 mM or (b) 1 mM, showing single bres or bundled bres, respectively. Attachment was performed immediately on a glass coverslip after dilution (NaCl 1 M).

NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561

coverslip at room temperature to freeze the exchange. Two-colour STORM images of assemblies were then acquired at the different time points. Fig. 5 shows representative PA nanobres at t 1

min and after 48 h of mixing; as clearly shown in the images and in the corresponding localization density plot (Fig. 5b) the bres initially containing a single label fully mix during the course of the experiment.

We veried that the exchange process is bidirectional by following green bres incorporating red monomers (Supplementary Fig. 4) and vice versa (Supplementary Fig. 5). The intermediate time points provide insight into the exchange mechanism. Visual inspection of STORM images (Supplementary Fig. 6) showed insertion of labelled monomers over the entire length of the nanobres throughout the exchange experiment. This nding suggests an exchange mechanism on the basis of expulsion and reinclusion of PA molecules. Polymerizationdepolymerization at the nanobre ends would have resulted in preferential exchange in those regions, while fragmentationrecombination would have

resulted in a block-like structure; neither of these mechanisms is consistent with the acquired STORM images.

To conrm this observation and further quantify the exchange, we studied the change in distribution of both dyes with time in initially green-labelled nanobres, thus tracking the insertion of Cy5-PA. Figure 6 shows the auto- and cross-correlation plots for the green and red channels at the different time points. The green channel displays the previously observed clustered distribution described by equation (2) rendering it less amenable for quantitative analysis. Therefore, we monitored the autocorrelation of the red channel in time to analyse monomer exchange. The red channel autocorrelation, shown in Fig. 6b, can be t with equation (1) for all time points, suggesting a random exchange of monomers. However, for later time points (that is, after incubation longer than 6 h) the quality of the t deteriorates, indicating some non-random variations of the Cy5-PA concentration in the nanobres other than the single decay originated from overcounting. However, the formation of regular micro-domains does not seem plausible, because equation (2) did not improve the quality of the t or yield reasonable tting parameter values. Since clustering is not observed in single-colour Cy5-PA nanobres (Fig. 3f), this observation points to a heterogeneous exchange and indicates the presence of structural variations along the bre or between bres. This heterogeneity can also be perceived visually by observing single bre images (Supplementary Fig. 6). The cross-correlation between both channels was computed to assess how the distribution of one labelled PA would affect the distribution of the other, as shown in Fig. 6c. A cross-correlation of zero was found for the entire range, indicating that the distribution of Cy5-PA is not inuenced by the presence of Cy3-PA, and vice versa. In other words, the heterogeneous exchange of Cy5-PA is not due to the pre-existing clusters of Cy3-PA, but rather a consequence of structural variation in the PA nanobre.

The observation of this heterogeneity in exchange, not detected by the FRET experiment, was possible due to the ability of STORM to evaluate the progress of molecular exchange for each individual supramolecular nanobre. The variability of the exchange progress between different aggregates could be measured, thus allowing us to further probe the heterogeneity in the system during and after molecular exchange. The linear density of localizations, dened as the number of localizations per nanometre arc length, has been computed and could be used as a rough estimate for the ensemble concentration of labelled molecules19. The density of Cy5-PA in originally single-colour green bres was measured as a function of time, showing an increase in the average value as molecular exchange proceeded (Fig. 7a). However, the standard deviation also increased markedly, another indication for the presence of intrinsic structural diversity. In Fig. 7b, it is possible to observe how the distribution of the Cy5-PA linear density gets progressively wider as molecular exchange evolves in time. This can be conrmed visually by inspecting STORM images acquired after 48 h mixing time (Fig. 7c-f), which displayed a range of different behaviours. It was possible to observe a large subset of bres that had a Cy5-PA density consistent with full mixing (Fig. 7c), as well as bres comprised of domains of Cy3-PA only (Fig. 7d,e), and both subsets co-existed with a smaller subpopulation of nanobres that had undergone very little exchange all over their length (Fig. 7f). The existence of regions displaying minimal or no exchange after 48 h implies that full equilibration of these regions will take weeks to months, making them persistent for most practical purposes. This analysis suggests that the population of bres is considerably more heterogeneous after 48 h than would be expected on the basis of the FRET ensemble measurements.

a

1 min 48 h

10

8

6

4

2

10

Y(m)Y(m)

8

6

4

2

0

2 8 10

6

48 h

1 min

4

4

6

8 10

Green counts

2

X (m)

X (m)

b

10

5 0

40

20

10

5

0 40 20

Rded counts

Rded counts

Green counts

0 0 1 2

Arclength (m)

0 0 1 2 3 4

3 4

Arclength (m)

Figure 5 | Quantitative analysis of STORM imaging during a molecular exchange experiment. (a) Localization maps of PA nanobres immobilized on a glass coverslip at different time points, after applying a clustering algorithm, background cleaning and backbone nding. The better reconstruction of the bres in the green channel reects the selection of initially Cy3-labelled bres for this experiment. (b) Histograms depict the localization density proles along the nanobre backbone (bin size 25 nm).

6 NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561 ARTICLE

a b c

Green channel Red channel Cross-correlation

Autocorrelation

0.80.4

Autocorrelation

0.4

0.0

0.2

Cross-correlation

n=20

n=20

n=23

n=20

n=22

n=20

0.8

0.4

0.0

24 h300 min150 min30 min1min

48 h

0.0

0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000

Autocorrelation

0.80.40.0

Autocorrelation

0.4

0.0

0.2

Cross-correlation

0.8

0.4

0.0

0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000

0.8

0.0

0.2

Autocorrelation

0.80.40.0

Autocorrelation

Cross-correlation

0.8

0.4

0.0

0.6

0.4

0.2

0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000

0.6

0.0

Autocorrelation

0.80.40.0

Autocorrelation

0.2

Cross-correlation

0.4

0.8

0.4

0.0

0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000

0.8

0.0

0.2

Autocorrelation

0.80.40.0

Autocorrelation

0.6

0.4

Cross-correlation

0.8

0.4

0.0

0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000

Autocorrelation

0.80.40.0

Autocorrelation

0.8

0.0

0.6

0.4

Cross-correlation

0.8

0.4

0.0

0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000 0 250 500 750

Radius (nm)

1,000

Figure 6 | Correlation of cyanine dyes along PA nanobres initially only labelled with Cy3 during molecular exchange. Averaged autocorrelation plots for (a) Cy3 and (b) Cy5-labelled PAs and (c) averaged cross-correlation between both dyes. Solid lines correspond to model ttings, namely equation (2) for (a) green channel and equation (1) for (b) red channel (error bars represent 95% condence level, nZ20).

Previous molecular dynamics simulations of a similar PA provide a theoretical framework for this structural diversity27. In these simulations, a broad distribution of secondary structures was found in the equilibrated bre. The heterogeneous molecular exchange pattern observed in our study might stem from this conformational diversity. The key feature that led to the discovery of PA supramolecular nanobres in the Stupp laboratory was the use of a b-sheet peptide domain to drive one-dimensional self-assembly. Using electron paramagnetic resonance, the b-sheet domain has been recently shown to give rise to locally solid-like behaviour in the interior of nanobre PA assemblies32. On the other hand, the surface moieties in these nanobres have been identied by the electron paramagnetic resonance experiments as regions where liquid-like dynamics prevail. b-Sheets within the nanobres are highly cohesive assemblies and therefore the dynamic exchange among supramolecular nanobres should give rise to a large diversity of supramolecular environments. This is in contrast to supramolecular systems with weaker internal cohesion

that could exchange molecules to produce completely homogeneous environments19. The formulation of supra-molecular systems with high levels of internal order and cohesion such as the PA nanobres will therefore open new avenues to generate structural diversity for functional purposes. One example will be their ability to adapt structurally to bind important bioactive targets.

DiscussionWe have studied the dynamics of PA nanobres using FRET and super-resolution localization microscopy. While ensemble FRET measurements prove that PAs exchange between different bres, two-colour STORM imaging reveals a mechanism on the basis of the transfer of monomers and small clusters. Remarkably, the coexistence of fully dynamic and kinetically inactive areas in the aggregate architecture was observed, demonstrating the existence of structural diversity in PA nanobres. This intriguing dynamic

NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561

0.2

Linear density (nm1 )

0.1

0.0

8

6

a b

Cy5 channel

Equilibrium

48

0 12 24 36Time (h)

1 min

0.5 h

2.5 h

6 h

24 h

48 h

10 c d

(m)

Y(m)

2

0

10

4Y

e

f

8

6

4

2

0

2 8

6

4

2

4

6

8

10 10

X (m) X (m)

Figure 7 | PA molecular exchange kinetics estimated using the linear density of localizations calculated from STORM data. (a) Time course-averaged density of red dye (Cy5) on bres that were initially only labelled with green dyes (Cy3) (nZ20, error bars correspond to standard deviation). The equilibrium guideline is estimated from the localization density of original single-colour red bres. (b) Box chart of data represented on a depicting average (circle), median (box middle line), quartiles (box edges), 5 and 95% percentiles (error bars) and full data set points (diamond symbols, bin size 5 mm 1); (cf) Examples of localization plots (after background cleaning) of individual bres at 48 h.

behaviour is foreseen to have important implications in the biological performance of supramolecular systems.

Methods

Peptide synthesis and purication. PAs were synthesized using standard Fmoc solid-phase peptide synthesis. MBHA rink amide resin, Fmoc-protected amino acids and other solid-phase peptide synthesis reagents were purchased from Novabiochem (USA). Cyanine dyes were supplied by Cyandye (USA). ACS-grade solvents were used for synthesis. Water was puried on an EMD Milipore Milli-Q Integral Water Purication System. For each amino acid coupling, 4.1 equiv of Fmoc-protected amino acid was activated for 1 min with 4 equiv of HBTU (O-benzotriazole-N,N,N0,N0-tetramethyluroniumhexauorophosphate) and 6 equiv

DIPEA (N,N-diisopropylethylamine) in 20 ml DMF. The coupling cocktail was then added to 0.5 mmol of MBHA rink resin and reacted for 1 h. Fmoc removal was performed with 30% piperidine in DMF. Fmoc-Glu(OtBu)-OH (N-a-Fmoc-L-

glutamic acid g-t-butyl ester) (3 ), and Fmoc-Ala-OH (N-a-Fmoc-

L-alanine)

(6 ) were successively coupled to the resin. Finally, the palmitic acid tail

(8.1 equiv) was coupled using 8 equiv of HBTU and 12 equiv DIPEA in DMF/DCM (50:50). PAs were cleaved from the resin using a cleavage solution of 95% TFA,2.5% TIPS (triisopropylsilane) and 2.5% H2O. Cleavage solution was concentrated

by rotary evaporation and the PA precipitated in cold diethyl ether. The crude solid was dissolved in a diluted NH4OH aqueous solution (B10 mg ml 1) and puried by preparative-scale reversed-phase HPLC on a Varian Prostar 210 HPLC system, using acetonitrile/water gradient containing 0.1% NH4OH. All eluents and additives were HPLC grade. Separation was achieved using a Phenomenex C18 Gemini NX column (150 30 mm) with 5 mm particle size and 110 pore size. Product-

containing fractions were conrmed by ESI mass spectrometry (Agilent 6510 Q-TOF LC/MS) and combined. ACN was removed by rotary evaporation, lyophilized to yield a white solid product and stored at 30 C.

Cyanine dye coupling. PAs labelled with cyanine dyes were synthesized rst coupling Fmoc-Lys(Mtt)-OH (N-a-Fmoc-N-e-4-methyltrityl-L-lysine) to the resin. The rest of the synthetic procedure proceeded as described above. After coupling the palmitic acid tail, Lys(Mtt) was selectively deprotected with 3% TFA, 5% TIPS and 92% DCM for 5 min (3 ), to expose the side chain amine moiety.

Disulfo-cyanine carboxylic acid dyes (Cy3 and Cy5) were dissolved in DMF and stock solutions were stored at -30 C. (Benzotriazol-1-yloxy)tripyrrolidinophosphonium hexauorophosphate (PyBOP) was also dissolved in DMF. Cyanine dyes (1 equiv) were activated with PyBOP (1 equiv) and DIPEA (2 equiv) for 1 min and coupled to the lysine-free amine. Cy3- and Cy5-labelled PAs were cleaved and puried as described above. Lyophilization yielded a red powder for Cy3 and blue for Cy5.

Purity. Purity was determined by reversed-phase analytical HPLC (see Supplementary Figs 79) using a Phenomenex Gemini C18 column (100 4.6 mm), with 5 mm

particle size and 110 pore size, connected to a HPLC system equipped with an autosampler (Shimadzu SIL-20A XR), degasser (Shimadzu DGU-20A3) and a high-pressure gradient system comprising two LC-20AD XR pumps (Shimadzu). Separation was performed at a ow of 1 ml min 1 and using a 20-60% acetonitrile linear gradient in water containing 0.1% NH4OH. All eluents and additives were HPLC grade. Peptides, cyanine dye-coupled peptides and eventual contaminants were detected using a PDA (Shimadzu SPD-M20A), acquiring a full ultraviolet-visible spectrum between 200 and 750 nm at any time point. Main chromatogram peaks were collected and product mass conrmed by ESI mass spectrometry using a LCQ Deca XP Max (Thermo Finnigan) ion-trap mass spectrometer.

For that, samples were manually injected bypassing the column at a ow rateof 0.20 ml min 1 and positive ion mass spectra were acquired in standard enhanced mode.

Sample preparation. Stock solutions of non-labelled PA (10 mM), Cy3-labelled PA (1 mM) and Cy5-labelled PA (1 mM) were prepared in Milli-Q water, aliquoted and stored at 30 C. Similar to other anionic PA systems, these formed hydrogels

in the presence of CaCl2. The degree of labelling of the bres was accurately controlled by simply mixing PA stock solutions at the desired ratio. Solutions of non-labelled PA were mixed with aqueous solutions of either one or both labels at certain mole ratios, ash-frozen in liquid nitrogen and lyophilized. The obtained PA mixture was molecularly dissolved in TFA and readily evaporated under vacuum. The obtained material was redissolved in NH4OH (aq.), ash-frozen and further lyophilized. These two last steps resembled the cleavage and purication conditions, respectively. These steps were undertaken to provide molecular mixing, as well as to reset PA self-assembly history. Non-labelled PA was treated using the same method before measurements where the uorescence-labelled molecules were absent. The obtained powders were reconstituted in appropriate buffers before measurements. After the harsh acidic conditions (dissolution in TFA) used to create a homogenous molecular mixture, PAs were checked by analytical HPLC to rule out potential degradation of the original molecules. Analytical HPLC was performed as described above. Supplementary Fig. 10 shows that the PA does not degrade during the harsh procedure of molecular mixing in TFA. Fluorescence measurements showed pronounced FRET for nanobres labelled with both Cy3 and Cy5. The FRET intensity increased with concentration of acceptor. This indicates that the sample preparation is suitable to assure co-assembly of the three PAs in a single supramolecular object.

Circular dichroism spectroscopy. Circular dichroism spectra were recorded using Jasco Circular Dichroism Spectrometer (model J-715). PA was dissolved in water at 10 mM. Before each measurement, PA stock solution was diluted in appropriate buffers to yield a nal PA concentration of 0.1 mM. A quartz cuvette of 1 mm path length was used for the measurements. Each trace represents the average of ve scans.

Cryogenic transmission electron microscopy. PA nanobres were separately labelled with either PA-Cy3 (5%) or PA-Cy5 (5%). Both labelled and non-labelled PA nanobres were dissolved at 1 mg ml 1 in 10 mM HEPES buffer (pH 7.5, NaCl 150 mM). CryoTEM was performed using a JEOL 1230 TEM at an accelerating voltage of 100 kV. Using a Vitrobot Mark IV (FEI) vitrication instrument at 25 C with 100% humidity, a 6.5-ml drop of the sample was deposited on a 300-mesh copper grid with lacey carbon support (Electron Microscopy Sciences, EMS), blotted twice, plunge-frozen in liquid ethane, then stored in liquid N2. For imaging,

8 NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561 ARTICLE

the sample was transferred to a Gatan 626 cryo-holder under liquid N2, and images were obtained with a Gatan 831 CCD camera.

Critical micelle concentration by Nile Red assay. Nile Red is a hydrophobic solvatochromic dye (water soluble at low concentration) with high afnity to the hydrophobic pocket of micelles. In aqueous environment, the molecule is weakly uorescent. In hydrophobic environments, uorescence intensity is around 300 times higher and a pronounced blue shift is observed. Therefore, Nile Red is commonly used to probe the existence of hydrophobic pockets formed by surfactants in aqueous environment38. Briey, a Nile Red 2 mM stock solution was prepared in MeOH. Non-labelled PA was dissolved at 10 mM in 10 mM HEPES buffer (pH 7.5, NaCl 150 mM). PA solution was diluted serially to obtain concentrations in the range 200 nM-10 mM. Nile Red stock solution was diluted in the same buffer to obtain a nal concentration of 1 mM. Equal volumes of Nile Red aqueous diluted solution and PA solutions were mixed for each PA concentration. Solutions were aged for 24 h to assure full nanobre disassembly at concentrations below the critical micelle concentration. Fluorescence-emission spectra (excitation 550 nm) were recorded for an emission range between 570 and 700 nm. The maximum intensity and respective wavelength at maximum intensity were both represented as a function of the logarithm of the PA concentration. At concentrations close to critical micelle concentration, it should be observed a sharp increase in uorescence intensity and a hypsochromic effect.

Frster resonance energy transfer. Single-colour PA nanobres labelled with either Cy3 or Cy5 at 1 mol% were assembled in 10 mM HEPES buffer (pH 7.5, NaCl 150 mM). Fluorescence spectra were recorded in a Varian Cary Eclipse Fluorimeter from Agilent Technologies. Emission spectra were acquired using excitation wavelengths of 520 or 615 nm for Cy3 or Cy5, respectively. Excitation spectra were collected using emission wavelengths of 600 or 710 nm for Cy3 or Cy5, respectively. Temperature was kept at 20C using the in-built peltier system. Supplementary Fig. 11a depicts the spectra of the individually labelled PA nanobres, showing signicant overlap between Cy3 emission (donor) and Cy5 excitation (acceptor). On an independent measurement, PA nanobres labelled with both Cy3 (1 mol%) and Cy5 (0, 0.1, 0.2 and 0.5 mol%) were assembled in 10 mM HEPES buffer (pH 7.5, NaCl 150 mM) at a nal PA concentration of 0.5 mM. Fluorescence-emission spectra of dual-labelled PAs were recorded at the excitation wavelength of Cy3 520 nm, at which direct excitation of Cy5 is negligible. Supplementary Fig. 11b shows considerable energy transfer by FRET, which increases monotonically with acceptor concentration, providing evidence for co-assembly of both Cy3- and Cy5-labelled PAs in the same nanobres.

Ensemble molecular exchange kinetics. For molecular exchange kinetic experiments, two sets of PA nanobres were pre-assembled from a mixture of non-labelled PAs and either Cy3 (0.5%) or Cy5 (0.5%). These independently labelled PA solutions were mixed at 1:1 ratio in a 50 ml quartz cuvette and uorescence was followed over around 24 h (lag time B30 s). FRET ratio was determined by dividing Cy5 emission at 668 nm by Cy3 emission at 565 nm, at the Cy3 excitation wavelength (520 nm). The contribution of Cy3 emission at 668 nm was rst subtracted from the Cy5 emission. At the beginning of the kinetics experiment, Cy3-PA and Cy5-PA are distributed in different nanobres, being physically separated by distances much higher than the expected FRET radius35. Therefore, energy transfer at time zero should be negligible and, consequently, FRET ratio should be zero. A total PA concentration range of 0.55 mM was used. Kinetics was recorded at two different temperatures (20 or 37 C).

Sample preparation for microscopy. Glass microscope coverslips were cleaned by successively immersing in acetone, isopropanol and Milli-Q water. Bath sonication was performed for 10 min with each solvent, followed by drying under N2 ow.

The glass coverslips were then etched with a fresh Piranha solution (3:1 v/v H2SO4 (98%):H2O2 (30%)) for 30 min. To nish the cleaning procedure, the slides were washed thoroughly with Milli-Q water and rinsed with acetone before drying under N2 ow. To determine the labelled PA distribution along the nanobres length, single bres should be immobilized on a glass surface, and remain still during STORM acquisition. A ow chamber was assembled using a glass slide and a clean coverslip separated by double-sided tape. PA nanobres were immobilized by adsorption onto the surface of the clean coverslip by ushing diluted PA solutions at high ionic strength (NaCl 1 M). After incubation for 1 min, the unbound nanobres were washed out of the chamber by ushing with HEPES buffer (2 )

followed by STORM buffer (2 ). STORM buffer is composed by 50 mM HEPES

buffer (pH 8.5), 1 M NaCl, an oxygen-scavenging system (0.5 mg ml 1 glucose

oxidase, 40 mg ml 1 catalase, 5 wt% glucose) and 200 mM 2-aminoethanethiol. The high ionic strength conditions were used to screen the electrostatic repulsion between highly charged PA nanobres and the glass surface. However, repulsion between PA nanobres is also reduced, causing bre bundling. To avoid bundling, PA solutions were rapidly diluted at low ionic strength (1 mM). After a nal (10 )

dilution step in NaCl 1M, nanobres were immediately attached on the glass coverslip. At this concentration isolated nanobres are mainly observed in the microscopic mounting.

Stochastic optical reconstruction microscopy. STORM images were acquired using a Nikon N-STORM system congured for total internal reection uorescence imaging. Excitation inclination was tuned to maximize the signal-to-noise ratio of the glass-absorbed bres. Cy3- and Cy5-labelled samples were illuminated by 561 and 647 nm laser lines. No activation ultraviolet light was used. Fluorescence was collected by means of a Nikon 100 , 1.4NA oil immersion objective

and passed through a quad-band pass dichroic lter (97335 Nikon). All time-lapses were recorded onto a 64 64 pixel region (pixel size 0.17 mm) of an EMCCD

camera (ixon3, Andor). For each channel, 20,000 frames were sequentially acquired. STORM movies were analysed with the STORM module of the NIS Elements Nikon software.

Molecular imaging of bres during exchange. PA nanobres were separately labelled with either PA-Cy3 (5%) or PA-Cy5 (5%) at 0.5 mM in 10 mM HEPES buffer (pH 7.5, NaCl 150 mM). Subsequently, 250 ml-aliquots of each solution were mixed in a microcentrifuge tube and gently shaken at 37 C for different amounts of time. At predened time points, 2 ml of the solution was withdrawn, diluted and adsorbed on a glass surface as described above. We veried that once adsorbed on the slide, bres are not able anymore to considerably exchange monomers with the solution in the time frame of the experiment. STORM imaging of Cy5 and Cy3 channels was performed sequentially. Fibres that are initially labelled only with Cy3 and incorporate Cy5 monomers over time were selected observing the uorescence intensity in the low-resolution images. Original single-colour nanobres were also observed before mixing as a control.

Image analysis. During STORM imaging, the Nikon software generates a list of localizations by 2D Gaussian tting of blinking chromophores in the acquired movie of conventional microscopic images. This localizations list was subsequently analysed with our custom-made Matlab scripts, as described in detail elsewhere19. Briey, a rst script uses a density-based clustering algorithm to automatically identify the bres in the image and to remove the background. Examples of output localization maps are shown in Figs 3a,b and 5a. A second script is run on the well-reconstructed bres obtained in the Cy3-channel to nd the bre backbone. The lower localization density obtained in the Cy5-channel for shorter time periods in the molecular exchange experiment is not enough to fully reconstruct the nanobres. The polymer backbone coordinates are used to obtain structural information. Next, histograms of localization density proles along the backbone are generated for both channels, as shown in Figs 3c,d and 5b. The spatial autocorrelation (that is, the average correlation between the localization density at one backbone point in the bre and the density at another backbone point, as a function of the distance between these points) was also computed using the Matlab-function xcov with unbiased normalization. The spatial cross-correlation of the two channels was calculated in a similar manner. Averaging over multiple bres is required to obtain accurate correlation decay graphs, such as depicted in Figs 3e,f and 6. Investigation of the average correlation decays yields the distribution of dyes in a bre, without interference from the stochastic uctuations inherent to the technique.

References

1. Dominguez, R. & Holmes, K. C. Actin structure and function. Annu. Rev. Biophys. 40, 169186 (2011).

2. Janke, C. & Kneussel, M. Tubulin post-translational modications: encoding functions on the neuronal microtubule cytoskeleton. Trends Neurosci. 33, 362372 (2010).

3. Pollard, T. D. & Borisy, G. G. Cellular motility driven by assembly and disassembly of actin laments. Cell 112, 453465 (2003).

4. Pollard, T. D., Blanchoin, L. & Mullins, R. D. Molecular mechanisms controlling actin lament dynamics in nonmuscle cells. Annu. Rev. Biophys. Biomol. Struct. 29, 545576 (2000).

5. Stournaras, C., Gravanis, A., Margioris, A. N. & Lang, F. The actin cytoskeleton in rapid steroid hormone actions. Cytoskeleton 71, 285293 (2014).

6. Symons, M. & Rusk, N. Control of vesicular trafcking by Rho GTPases. Curr. Biol. 13, R409R418 (2003).

7. Albertazzi, L. et al. Spatiotemporal control and superselectivity in supramolecular polymers using multivalency. Proc. Natl Acad. Sci. USA 110, 1220312208 (2013).

8. Marchi-Artzner, V. et al. Selective adhesion, lipid exchange and membrane-fusion processes between vesicles of various sizes bearing complementary molecular recognition groups. Chemphyschem Eur. J. Chem. Phys. Phys. Chem. 2, 367376 (2001).

9. Ferrell, J. E., Lee, K. J. & Huestis, W. H. Lipid transfer between phosphatidylcholine vesicles and human-erythrocytesexponential decrease in rate with increasing acyl chain-length. Biochemistry 24, 28572864 (1985).

10. Lund, R., Willner, L., Richter, D. & Dormidontova, E. E. Equilibrium chain exchange kinetics of diblock copolymer micelles: tuning and logarithmic relaxation. Macromolecules 39, 45664575 (2006).

11. Zinn, T., Willner, L., Lund, R., Pipich, V. & Richter, D. Equilibrium exchange kinetics in n-alkylPEO polymeric micelles: single exponential relaxation and chain length dependence. Soft Matter 8, 623626 (2011).

NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms11561

12. Fasting, C. et al. Multivalency as a chemical organization and action principle. Angew. Chem. Int. Ed. 51, 1047210498 (2012).

13. Helms, B. A. et al. High-afnity peptide-based collagen targeting using synthetic phage mimics: from phage display to dendrimer display. J. Am. Chem. Soc. 131, 11683 (2009).

14. Silva, G. A. et al. Selective differentiation of neural progenitor cells by high-epitope density nanobers. Science 303, 13521355 (2004).

15. Baker, M. B. et al. Consequences of chirality on the dynamics of a water-soluble supramolecular polymer. Nat. Commun. 6, 6234 (2015).

16. Bates, M., Huang, B., Dempsey, G. T. & Zhuang, X. Multicolor super-resolution imaging with photo-switchable uorescent probes. Science 317, 17491753 (2007).

17. Shim, S.-H. et al. Super-resolution uorescence imaging of organelles in live cells with photoswitchable membrane probes. Proc. Natl Acad. Sci. USA 109, 1397813983 (2012).

18. Huang, B., Bates, M. & Zhuang, X. Super-resolution uorescence microscopy. Annu. Rev. Biochem. 78, 9931016 (2009).

19. Albertazzi, L. et al. Probing exchange pathways in one-dimensional aggregates with super-resolution microscopy. Science 344, 491495 (2014).

20. Webber, M. J., Berns, E. J. & Stupp, S. I. Supramolecular nanobers of peptide amphiphiles for medicine. Isr. J. Chem. 53, 530554 (2013).

21. Boekhoven, J. & Stupp, S. I. 25th anniversary article: supramolecular materials for regenerative medicine. Adv. Mater. 26, 16421659 (2014).

22. Stupp, S. I. & Palmer, L. C. Supramolecular chemistry and self-assembly in organic materials design. Chem. Mater. 26, 507518 (2014).

23. Soukasene, S. et al. Antitumor activity of peptide amphiphile nanober-encapsulated camptothecin. ACS Nano 5, 91139121 (2011).

24. Moyer, T. J. et al. Shape-dependent targeting of injured blood vessels by peptide amphiphile supramolecular nanostructures. Small 11, 27502755 (2015).

25. Paramonov, S. E., Jun, H.-W. & Hartgerink, J. D. Self-assembly of peptide amphiphile nanobers: the roles of hydrogen bonding and

amphiphilic packing. J. Am. Chem. Soc. 128, 72917298 (2006).26. Pashuck, E. T., Cui, H. G. & Stupp, S. I. Tuning supramolecular rigidity of peptide bers through molecular structure. J. Am. Chem. Soc. 132, 60416046 (2010).

27. Lee, O.-S., Stupp, S. I. & Schatz, G. C. Atomistic molecular dynamics simulations of peptide amphiphile self-assembly into cylindrical nanobers.J. Am. Chem. Soc. 133, 36773683 (2011).28. Storrie, H. et al. Supramolecular crafting of cell adhesion. Biomaterials 28, 46084618 (2007).

29. Stephanopoulos, N., Ortony, J. H. & Stupp, S. I. Self-assembly for the synthesis of functional biomaterials. Acta Mater. 61, 912930 (2013).

30. Greeneld, M. A., Hoffman, J. R., de la Cruz, M. O. & Stupp, S. I. Tunable mechanics of peptide nanober gels. Langmuir 26, 36413647 (2010).

31. Newcomb, C. J. et al. Cell death versus cell survival instructed by supramolecular cohesion of nanostructuresl. Nat. Commun. 5, 3321 (2014).

32. Korevaar, P. A., Newcomb, C. J., Meijer, E. W. & Stupp, S. I. Pathway selection in peptide amphiphile assembly. J. Am. Chem. Soc. 136, 85408543 (2014).

33. Tantakitti, F. et al. Energy landscapes and functions of supramolecular systems. Nat. Mater. 15, 469476 (2016).

34. Levitt, M. Conformational preferences of amino acids in globular proteins. Biochemistry 17, 42774285 (1978).

35. Bastiaens, P. I. & Jovin, T. M. Microspectroscopic imaging tracks the intracellular processing of a signal transduction protein: uorescent-labelled protein kinase C beta I. Proc. Natl Acad. Sci. USA 93, 84078412 (1996).

36. Rust, M. J., Bates, M. & Zhuang, X. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat. Methods 3, 793796 (2006).

37. Veatch, S. L. et al. Correlation functions quantify super-resolution images and estimate apparent clustering due to over-counting. PLoS ONE 7, e31457 (2012).

38. Stuart, M. C. A., van de Pas, J. C. & Engberts, J. B. F. N. The use of Nile Red to monitor the aggregation behavior in ternary surfactantwaterorganic solvent systems. J. Phys. Org. Chem. 18, 929934 (2005).

Acknowledgements

Molecular synthesis, purication and characterization were supported by the U.S.

Department of Energy, Ofce of Science, Basic Energy Sciences, under Award

# DE-FG02-00ER45810. The work in Eindhoven was supported by was nanced by the

Dutch Ministry of Education, Culture and Science (Gravity program 024.001.035) and

the European Research Council (FP7/2007-2013, ERC Grant Agreement 246829).

R.M.P.d.S. and his research activities were funded by the Marie Curie FP7-PEOPLE-

2010-IOF program (ref. no. 273295). Compound preparation, purication and char

acterization were partially performed at the Peptide Synthesis Core at Simpson Querrey

Institute. The Biological Imaging Facility (BIF) and Keck Biophysics facilities at

Northwestern provided further instrumentation used in this work. Mark Seniw is

acknowledged for the preparation of graphics.

Author contributions

R.M.P.d.S., L.A., D.v.d.Z., S.I.S. and E.W.M. conceived the project and contributed to the

concept of the manuscript. R.M.P.d.S., L.A., D.v.d.Z. designed the experiments and

analysed the results. R.M.P.d.S. performed peptide synthesis, spectroscopic and STORM

experiments. S.S.L. performed cryoTEM imaging. R.M.P.d.S., L.A., D.v.d.Z., S.I.S. and

E.W.M. wrote the manuscript and contributed to discussions on the project.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/

http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/

http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: da Silva, R. M. P. et al. Super-resolution microscopy reveals

structural diversity in molecular exchange among peptide amphiphile nanobres.

Nat. Commun. 7:11561 doi: 10.1038/ncomms11561 (2016).

This work is licensed under a Creative Commons Attribution 4.0

International License. The images or other third party material in this

article are included in the articles Creative Commons license, unless indicated otherwise

in the credit line; if the material is not included under the Creative Commons license,

users will need to obtain permission from the license holder to reproduce the material.

To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

10 NATURE COMMUNICATIONS | 7:11561 | DOI: 10.1038/ncomms11561 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications