Self-replicators from dynamic molecular networks: selection, competition and subsystem
coupling
Komáromy, Dávid
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2019
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Komáromy, D. (2019). Self-replicators from dynamic molecular networks: selection, competition and
subsystem coupling. University of Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Chapter 4
Competition between self-replicators with
different interaction strength, morphology
and stoichiometry
This chapter will be submitted for publication:
D. Komáromy, D. M. Monzón Rodríguez, I. Marić, G. Monréal Santiago, M. Altay, J.
Ottelé, G. Schaeffer, S. Otto manuscript in preparation
106
4.1. Introduction
One of the ultimate goals of chemistry is capturing the transition between the physical and the biological world – in other words, the emergence of the simplest living entities from inanimate matter.[1] In this context, systems chemistry[2,3] has emerged in the last years as a multidisciplinary approach, addressing, among others, the following two important research areas: reconstructing the process of emergence of life in the past (prebiotic systems chemistry)[4–6] and constructing similar processes in the present (creation of de novo life).[7,8] Whereas the former approach focuses on biomolecules as we know them today (and their possible precursors), the latter aims at the understanding basically any molecular network that is capable of producing the same dynamical properties as minimal living systems. These properties encompass the capability for replication, introducing mutations into the replicating species formed and finally selecting for a subset of the formed mutants based on their fitness in a given environment under far from equilibrium conditions.[8,9] Other aspects, such as open-endedness[8] and the capability of forming compartments[10] is usually also regarded as necessary characteristic traits for minimal living systems. Furthermore, in order to understand and mimic the complex regulatory functions of biological systems,[11] the study of feedback loops[12–14] and signaling cascades[15–17] is also an inherent part of the systems chemistry approach.
As indicated above, a key requirement for creating de novo life is the presence of self-replicating molecular species. Self-replicators are a subset of molecules with autocatalytic properties: whereas autocatalysts in general are capable of accelerating their own formation, self-replicators accomplish this more specifically, by catalyzing their own assembly from more than one precursor. Consequently, self-replication necessarily involves a templating step, i.e. the intermolecular binding between precursors and the replicator. In the simplest case, this process involves two precursors and one replicator, forming a ternary complex as a key step, which allows for the accelerated formation of the replicator.[18,19] This minimalistic approach, however, could not serve as a foundation for the high selectivity and tunability featured by biological self-replicating systems. On the one hand, for the introduction of mutations, at least two (even if only slightly different) replicators should be able to emerge from the system.[20] On the other hand, replicators assembled by different intermolecular interactions and into various supramolecular architectures (as observed in the case of DNA) would enable further differentiation between them. Whereas studies featuring competition and/or cooperation between self-replicators differing at the molecular level have been conducted, much less is known about the dynamics of systems based upon self-replicators of different morphologies[21,22] – mainly because the majority of artificial self-replicators studied until now are based on the formation of a ternary complex.[23–27]
107
In the recent years, our group has made substantial progress in the creation and understanding of self-replicators based on dynamic combinatorial chemistry (DCC).[28] This approach is based on fast and reversible reactions, allowing for the straightforward creation of dynamic and diverse mixtures (dynamic combinatorial libraries, DCLs) starting from a few building blocks. Due to the full reversibility of thiol-disulfide exchange, the relative concentrations of the DCL members are determined by multiple equilibria, forming thereby a molecular network. If the building blocks contain chemical moieties which are capable of engaging in specific intermolecular interactions, the equilibria might be shifted towards the formation of specific DCL members, allowing for the templated formation of specific structures and functional molecules, e.g. receptors for external templates,[29–33] catalysts[34–36] interlocked architectures[37–40] or sensors.[41–46] If intermolecular interactions between identical DCL members are strong enough, they might serve as a template for their own formations from their precursors – a scenario, which is basically identical to self-replication.[47,48]
Scheme 4. 1. Dynamic Combinatorial Chemistry of a System Composed of Oligo(ethylene oxide)-Conjugated Dithiol Building
Block 1 and Peptide-Conjugated Dithiol Building Blocks 2 and 3.
In our previous studies, we utilized thiol-disulfide exchange,[49] starting from aromatic dithiol monomers (e.g. 2 and 3, see Scheme 4. 1) covalently linked to a pentapeptide consisting of alternating hydrophobic and hydrophilic amino acids. Upon oxidation, a DCL of macrocyclic disulfides are formed,
108
one of which is capable of templating its own formation, promoted by β-sheet interactions between the peptide chains. Usually, this macrocycle (26 or 38) is initially a minor (thermodynamically not favored) species, whose formation is enabled only by the aforementioned autocatalytic template-assisted process.[50] As a consequence of this mechanism, the self-replicating macrocycles self-assemble into nanoscale fibers. Remarkably, a mixture of two peptidic building blocks forming replicators which are held together by intermolecular interactions of slightly different strength, can give rise to the emergence of two sets of closely related ensembles of replicators, with the second emerging only after the emergence of the first one, featuring an ancestor-offspring relationship.[51] In this case, the process could not be tracked at the supramolecular level, as both replicating sets are assembled into similar fibrous structures. More recently, however, we reported a building block (1, see Scheme 4. 1) which features the same dithiol core but contains a short oligo(ethylene oxide) moiety instead of a peptide chain. Upon oxidation and exchange, the self-replicating 16 macrocycle is formed, self-assembling into more densely packed nanoribbons, held together solely by hydrophobic effects.[52] Here we report the direct comparison of two self-replicators, which are formed based on the same molecular (thiol-disulfide chemistry), but on different supramolecular (β-sheet versus hydrophobic interactions) organization. We show that the strength of supramolecular interactions holding together the assemblies of replicators correlates with their catalytic properties, i.e. replicators organized by stronger intermolecular interactions are capable of cross-catalyzing the formation of those held together by weaker interactions but this relation does not hold the other way.
4.2. Results and discussion
4.2.1. Investigation of the (1+2)-systemStirring an equimolar mixture of building blocks 1 and 2 in aqueous borate buffer (50 mM, pH = 8.2) in the presence of atmospheric oxygen, a mixture of smaller macrocycles (trimers and tetramers) were obtained upon complete oxidation (6 days), as confirmed by UPLC and UPLC-MS analyses. Continued mechanical agitation of the sample resulted in the appearance of larger macrocyclic species, more specifically mixed hexamers and to a lesser extent, mixed pentamers (Figure 4. 1) after 8 days, whereas if stirring was halted, only trimers and tetramers were present even after 30 days. After nearly complete oxidation, hexamers were present in low concentration but after 14 days, they accounted for more than 92% of the overall species detected.
109
Figure 4. 1.UPLC chromatogram of a DCL prepared from equimolar amounts ([1]=[2]=2.0 mM) of building blocks 1 and 2 in aqueous borate buffer (50 mM, pH = 8.5) after 8 days of stirring.
These observations suggested that the hexamers collectively (i.e. regarded as one species) were self-replicators. This assumption seemed to be in line with previous observations, i.e. that separately, each building block was observed to form hexameric self-replicators. For this reason, the time evolution of a DCL prepared from equimolar amounts of building blocks 1 and 2 ([1] = [2] = 2.0 mM) was studied. The amount of the hexamers 1224, 1323 and 1422 (as well as that of pentamers 1322 and 1421) showed a characteristic growth curve, including a lag phase, followed by a sudden increase in relative concentration, leading finally to the halt of growth as the available building blocks were depleted (Figure S4. 55A). Summation of the resulting data by oligomer size suggested that the hexamers (and pentamers) as a distinct set of species are self-replicators (Figure 4. 2A). Notably, the hexamers evolve simultaneously, pointing out that there is no substantial difference in the kinetic barriers of their assembly (Figure S4. 55B).
In order to confirm that the corresponding hexamers grow in an autocatalytic manner, a seeding experiment was carried out. A pre-oxidized, non-agitated DCL (food) prepared from equimolar amounts of building blocks 1 and 2 ([1] = [2] = 2.0 mM) was treated with 15 mol% of a mixture of hexamers (seed) prepared from a DCL with the same composition. In the seeded sample, the immediate growth of the corresponding hexamers was observed, while in the unseeded sample the growth of the same hexamers started only after a lag phase of ca. 50 hours (Figure 4. 2B). These results show that the ensemble of the mixed hexamers prepared from an equimolar mixture of 1 and 2 possess are self-replicators (abbreviated as (1+2)6).
110
Figure 4. 2. A) Time evolution of differently sized oligomers in a stirred DCL prepared from equimolar amounts ([1] = [2] = 2.0
mM) of 1 and 2 in aqueous borate buffer (50 mM, pH = 8.5). B) Change of relative concentration of hexamers in a DCL prepared by fast oxidation of an equimolar mixture of 1 and 2, upon seeding with 0 and 15 mol % of hexamers prepared from a DCL with the same building block composition.
As the self-replication of the hexamers arising from building blocks 1 and 2 respectively is promoted by self-assembly into well-defined nanostructures during replication (16 forms nanoribbons, whereas 26 is self-assembled into fibers), we also investigated the morphology of the mixed hexamers. Negative-staining Transition Electron Microscopy (TEM) showed the presence of elongated fibrous aggregates (100-200 nm in length and 10 nm in width). The aggregates, however do not show any sign of helicity, in contrast to helically twisted fiber pairs made from 26, bundled around each other.[53]
These observations prompted us to investigate the supramolecular organization of the DCLs composed of mixed hexamers in more detail. Specifically, we were interested whether, by increasing the concentration of one building block at the expense of the other, the transition from one supramolecular structure to another is gradual or characterized by a sudden phase change.
For this purpose, we prepared mixed DCLs from 1 and 2 with a total building block concentration kept at 4.0 mM (in borate buffer containing 10 V/V % DMF) containing increasing amounts of 1 in 10 mol % increments. These mixtures of the monomers were stirred while exposed to air until the library composition showed no change and the DCL members were exclusively hexamers. UPLC analysis showed that the relative amounts of hexamers corresponded to the expected binomial distribution based on the corresponding monomer composition (Figure 4. 4). These results show that the two building blocks are incorporated into the various hexamers with the same probability. In other words, there is no self-sorting at the molecular level (i.e. neither of the hexamers is formed preferentially, regardless of the original building block composition).
111
Figure 4. 3. TEM micrographs of DCLs composed of equimolar amounts of building blocks 1 and 2, containing A) a statistical
distribution of mixed hexamers, B) obtained by seeding of an oxidized sample of equimolar amounts of building blocks, seeded with 15 % statistical distribution of hexamers of the same composition.
After establishing that no self-sorting takes place at the molecular level (i.e. none of the oligomers are formed in a higher extent than expected from the statistical distribution, as shown on Figure 4. 4L,M), we examined whether the same trend holds at the supramolecular level. We envisaged two different scenarios. In the first, the supramolecular properties of the assemblies would change linearly with the building block composition. In the second scenario, however, the properties of the resulting supramolecular assemblies would show an abrupt change at a certain building block composition, i.e. the system would display phase-change like behavior.
We examined the supramolecular properties of the assemblies with multiple techniques. First, ThT and Nile Red fluorescence were used in parallel to examine the hexamer assemblies formed from DCLs with increasing amounts of 1 (Figure 4. 5). On the one hand, the fibers composed of stacked 26 macrocycles display beta-sheet arrangement of the peptide chains. This feature can be easily detected by the enhanced fluorescence of thioflavin T (ThT) at 480 nm, caused by intercalation of the ThT molecule between the peptide chains. In contrast, as the nanoribbons formed from 16 do not display beta-sheet like structural elements, supramolecular assemblies similar to the nanoribbons would not give a fluorescent response when examined with the same assay. On the other hand, Nile Red shows enhanced fluorescence in nonpolar microenvironments and thus stains the assemblies composed of 16, as shown previously,[52] whereas it is expected to sense a less nonpolar microenvironment in the peptide
112
assemblies rich in polar groups. Thus, enhanced Nile Red fluorescence is expected in the case of assemblies more similar to 16.
Figure 4. 4. Representative UPLC chromatograms of stirred DCLs prepared from building blocks 1 and 2 (total building block
concentration kept at 4.0 mM) containing A) 0, B) 10, C) 20, D) 30, E) 40, F) 50, G) 60, H) 70, I) 80, J) 90, K) 100 mol % of building block 1 ; as well as the corresponding distributions of the relative amounts of mixed 1n26-n hexamers obtained from L)
the experiments above and M) calculations, assuming a binomial distribution of the corresponding hexamers. For hexamers
1224, 1323 and 1422, only one possible positional isomer per hexamer is depicted.
First, we established that increased Nile Red and ThT fluorescence were specific for assemblies of
16 and 26, respectively. The maximal emission intensity characteristic for Nile Red was ca. 6 times higher for 16 than for 26, whereas that for ThT was ca. 17 times higher for 26 than for 16. Second, both assays showed an abrupt change in supramolecular structure at compositions close to that corresponding to the pure replicators: ThT fluorescence dropped to a value 8 times lower already at 10 mol % of 1, whereas Nile Red fluorescence showed a somewhat less abrupt change at ca. 20 mol % of 2 (80 mol % of 1). These results suggest that the specific assemblies do not maintain their well-defined characteristics even upon mixing in only a small (10-20 mol %) fraction of the other building block.
113
Figure 4. 5.Maximum thioflavin T (blue trace) and Nile Red (red trace) intensities of stirred DCLs prepared from building blocks
1 and 2 with increasing molar percentage of 1 (total building block concentration kept at 4.0 mM).
Next, we examined the same samples with CD spectroscopy (Figure 4. 6). The 26-fibers display significant helicity, whereas the nanoribbons formed from 16 are achiral. This supramolecular chirality of the assemblies can be detected and quantified with circular dichroism, with characteristic bands at 211, 230 and 260 nm, corresponding to the circular dichroism the benzene ring of the core. Therefore, higher CD signals would indicate supramolecular assemblies more similar to the chiral fibers. In line with our results from the ThT assay, the CD signals characteristic (Figure 4. 6B) of beta-sheet fibers disappear already at 10 mol % of 1 (Figure 4. 6A, C and Figure S4. 56). At first, this might seem contradictory to the results shown on Figure 4. 4A, which indicate that ca. 50 mol% of the total DCL material is present in form of 26, which is known to form chiral assemblies and therefore is expected to display CD bands comparable in intensity to that corresponding to the pure fibers (Figure 4. 6B). This is only possible if 26, although present in large quantities on the molecular level, are mainly incorporated into achiral nanostructures (at the molecular level).
114
Figure 4. 6. A) CD intensity (at 211 nm) measured in DCLs containing mixed hexamers of 1 and 2 ([1]+[2] = 4.0 mM, see Figure 4. 4) as a function of composition. As an example, the CD spectra of the DCLs containing B) 0 , C) 10 mol% of building block 1
is shown, showing the disappearance of the characteristic CD signature already at 10 mol % 1.
In order to shed light on the structure of the supramolecular assemblies in a more direct manner, we examined the hexamer-containing DCLs with atomic force microscopy (AFM). AFM enables the direct visualization of the assemblies and provides information of their shape and size in three dimensions. Besides the characteristic difference in the width of the assemblies (10 nm for the 26-fibers and 30-40 nm for the 16-ribbons), they also display different heights (5 nm for the former and 2 nm for the latter). Additionally, the chiral fibers of the peptide macrocycles show pairwise association resulting in helical fiber bundles, which could be directly visualized by AFM. These differences make the system suitable to follow the morphological changes as a function of the stoichiometry with AFM. In the DCL composed exclusively of 26 the macrocycles self-assemble into helical double fibers[53] with a height of ca. 4 nm (Figure 4. 7A).
115
Figure 4. 7.AFM micrographs and selected height profiles of stirred DCLs prepared from building blocks 1 and 2 (total building block concentration kept at 4.0 mM) containing A) 0, B) 10, C) 20, D) 30, E) 40, F) 50, G) 60, H) 70, I) 80, J) 90, K) 100 mol % of building block 1 and of L) a non-stirred, oxidized DCL containing 50 mol % 1. Scale bar, 200 nm. Color code: Red: nanoribbons, Blue: chiral (helical) fibers, Green: achiral fibers.
In general, the morphologies characteristic for the pure replicators disappear in the presence of very low amounts of the other building block and give rise to the formation of an intermittent morphology.
116
More specifically, assemblies showing supramolecular helicity become almost absent upon incorporation of even low amounts of building block 1. This is in line with the sudden disappearance of β-sheets, as observed from the ThT assay (Figure 4. 5). In samples containing 20-70 mol % 1, achiral fibers can be detected, i.e. assemblies displaying several 100 nm in length and approximately the same width as the fibers formed from 26, however, without showing helicity. An unexpected re-appearance of helical assemblies was observed at the sample containing 40 mol% of 2 (Figure 4. 7E). This finding is paralleled in the CD measurements, where the same sample displays strong bands characteristic of β-sheets (Figure 4. 7E) while none of the samples containing building block 2 show any supramolecular chirality. The reason for this behavior is unknown. Furthermore, a control AFM experiment performed on an oxidized but quiescent sample (containing 50 mol % of 1) shows the absence of nanoscale assemblies, pointing out that the observed assemblies are most probably composed of hexamers.
More importantly, upon disappearance of helical fibers with increasing 1-content, the observed assemblies show variable heights, between 2-7 nm. This finding can be rationalized in two ways: First, building block 1, bearing substantially shorter and more flexible side chains is incorporated into the assemblies, resulting in the decrease in height (from 4 to 2 nm), as observed in several cases (Figure
4. 7D-F). Second, as building block 1 is non-charged under the current conditions in contrast to 2
(possessing protonated lysine end groups), these assemblies now bear less positive surface charges compared to fibers of 26. Consequently, these assemblies are more probable to associate vertically, resulting in assemblies much higher than the 26 fibers themselves (Figure 4. 7F-G). The molecular composition and thus the charge content of the single fibers can be assumed to vary considerably, which may lead to a varying degree of association and can hence account for the large variability in fiber height. In samples containing large amounts (≥ 80 mol%) of 1, laterally highly associated nanoribbons (60-80 nm in width) characteristic of 16 were observed (Figure 4. 7I-K), together with the achiral fibers. This change is paralleled in the observed hexamer distributions (Figure 4. 7F), which shows a sudden increase of the amount of 16 at 80 mol% of 1. Similarly, the Nile Red fluorescence shows a sharp increase approximately at the same composition. The fact that the sharp increase of the amount of 16 at the molecular level and the sudden appearance of the corresponding assemblies at the supramolecular level is observed at the same composition, suggests that the observed nanoscale assemblies (i.e. nanoribbons) are composed exclusively of 16. In other words, the probability that the nanoribbons incorporate other hexamers than 16 is fairly low.
The findings from spectroscopy and AFM experiments suggest that the assemblies characteristic of the pure hexamers are not preserved upon the incorporation of even small quantities of the other building block. The nanostructures formed at intermediate (20-80 mol % of 1) composition are however, although devoid of helicity, structurally more closely related to the 26-fibers than to the 16-nanoribbons (especially
117
regarding their width and degree of lateral association). These results suggest that building block 2 has a higher influence on the assembly morphology than building block 1.
Figure 4. 8. Comparison of self-and cross-seeding experiments, showing the time evolution of non-agitated DCLs containing
A) 1-food, B) 2-food, C) (1+2)-food, seeded with 16 (red circles), 26 (blue triangles) and (1+2)6 (cyan triangles) as well as a
non-seeded control (black squares).
These results prompted us to uncover the auto-and cross-catalytic relationships between the three different replicators (16, 26 and (1+2)6) and the corresponding nanoscale assemblies (chiral fibers, achiral fibers and nanoribbons). To this end, we designed seeding experiments in which a partially oxidized mixture (food) containing non-replicating species (mostly trimers and tetramers) is mixed with 10 mol % of pre-formed replicator (seed). The amount of oligomers whose formation can be catalyzed by the addition of the seed (most probably by assembling at the boundaries of the seed assemblies) is expected to increase upon addition of the seed at the expense of other oligomers. In this way, auto-and cross-catalytic relationships can be established.[19,54–56] Three different food solutions were used, prepared from 1, 2 or an equimolar mixture of the two (denoted as 1-food, 2-food (1+2)-food, respectively). These were seeded with 16, 26 and (1+2)6 (the hexamer distribution of (1+2)6 corresponds to that shown in
Figure 4. 4F). The solutions were not agitated in order to ensure that hexamer formation solely results
from the catalytic effect of the added seed. The time evolution of the seeded samples was monitored with UPLC.
The results show that 26 and (1+2)6 are mutually cross-catalytic (Figure 4. 8B, C) but they do not cross-catalyze the formation of 16 (Figure 4. 8A). Moreover, 16 is not cross-catalytic toward 26 and has only a very weak autocatalytic effect toward (1+2)6. Finally, as expected, all seeds catalyze their own formation form their corresponding food (as compared to a non-seeded control). The mutual cross-seeding abilities (Scheme 4. 2A) of 26 and (1+2)6 can be rationalized by their closely related
118
morphologies (similar fiber width) and the similar intermolecular binding motif (oriented peptide-peptide interactions). On the other hand, the ineffective cross-seeding between 16 and the other two replicators can be rationalized in a similar manner: On the one hand, the 16 nanoribbons are organized by the relatively weak hydrophobic effect, whereas the 26 fibers are additionally held together by stronger and highly directional hydrogen bonds. Thus, if 16 molecules were to grow at the catalytic end of the 26 fibers, their packing (which is determined by the seed) would be loose compared to that in native 16 nanoribbons and the resulting net intermolecular interactions would be too weak to hold the resulting assembly together (Scheme 4. 2B).
Scheme 4. 2. Schematic representation for the rationalization of A) effective cross-seeding between 1-food and (1+2)6-seed B)
ineffective cross-seeding between 1-food and 2-seed, C) ineffective cross-seeding between 2-food and 1-seed,
On the other hand, one can assume that the 16 molecules are more densely packed within the nanoribbons compared to the 26 molecules in the native fibers, as the former are much smaller than the latter; furthermore, in the latter, the secondary interactions are weaker and thus a higher number of molecules in a given volume is needed hold together the nanostructure. Thus, if 26 molecules were to
119
grow at the catalytic end of the 16 nanoribbon seeds, their packing would be determined by the morphology of the seed. However, this would be sterically too demanding, as the peptide moieties in 2 are much bulkier than the ethylene oxide side chains in 1(Scheme 4. 2C).
It has to be noted that 26 outperforms (1+2)6 both as an auto – and as a cross-catalyst (in terms of extent as well as rate of the induced replication, see Figure 4. 8B-C). This most possibly correlates with the interaction strength between catalytic fiber ends and the incoming food molecules, which is higher for 26 than for (1+2)6.
Figure 4. 9.Time evolution of cross-seeded DCLs, showing the relative change in the concentrations of the individual hexamers: A) 1-food and 26-seed, B) 1-food and (1+2)6-seed, C) 2-food and 16-seed D) 2-food and (1+2)6-seed, E) (1+2)-food and 16
-seed, F) (1+2)-food and 26-seed.
Detailed examination of the time evolution of the individual hexamer macrocycles during cross-seeding experiments also revealed interesting trends (Figure 4. 9). Cross-cross-seeding experiments using 16 as a seed (Figure 4. 9C and E, black squares) showed that the amount of seed decreased abruptly upon its addition to the food solution (to 0.1 and 1 mol %, respectively). In contrast, in experiments using 26 (Figure 4. 9A and F, blue diamonds), the seed persisted during the process. Finally, the amount of (1+2)6-seed, calculated as the total amount of its main components ([1323] + [1422] + [1224]) decreased
120
slightly upon mixing it with 1-food (from 5.8 to 4.0 mol%, Figure 4. 9B) but remained constant when added to 1-food (Figure 4. 9D). It should be noted that upon disappearance, the corresponding seed re-equilibrates to give trimers and tetramers, i.e. it gets converted into the food.
Figure 4. 10. Time evolution of DCLs prepared from a pre-oxidized mixture of 1 and 2 ([1] = [2] = 2.0 mM) and seeded with
equimolar amounts of 16 and 26 (10 mol% each) in the absence of stirring. Time evolution of the A) total mol% of hexamers
(compared to a non-seeded control); B) mol% of each hexamer (in the seeded sample).
Whether a certain seed persists or re-equilibrates upon cross-seeding, is a result of the tradeoff between the enthalpic and entropic terms of the free energy change of the process. On the one hand, the formation of a large number of smaller macrocycles (i.e. formation of trimers and tetramers) is entropically preferred to that of a smaller number of bigger macrocycles.[57,58] On the other hand, if the overall interaction strength (enthalpy) between the macrocycles of the seed assemblies is sufficiently large, it may counterbalance the entropy-driven process of re-equilibration to trimers and tetramers. It should also be noted that a high interaction strength within the seed assemblies also results in a high kinetic barrier for removal of one macrocycle from the assembly. The results shown are in line with these considerations: the interaction strength within the assemblies increases as 16 < (1+2)6 < 26 (vide infra). Analogously, 16 does not persist in cross-seeding experiments, whereas 26 does. In the case of (1+2)6, featuring intermediate interaction strength, the amount of seed remains constant in 2-food, thereby catalyzing the growth of 26 (Figure 4. 9D), whereas it decreases in 1-food, where cross-catalysis is not possible and thus the initially added seed cannot be stabilized by interaction with the potential newly formed 16 macrocycles (Figure 4. 9B) .
121
Finally, it is worth noting that in the case of (1+2)-food, the final distribution of the different hexamers is similar for all seeds, i.e. it is determined by the food composition and only slightly influenced by the added seed (Figure 4. 9E-F and Figure S4. 57).
Figure 4. 11. AFM micrographs and selected height profiles of cross-seeded samples taken at the end of the corresponding
seeding experiments: A) 2-food and 16-seed, B) 2-food and (1+2)6-seed, C) (1+2)-food and 26-seed, D) (1+2)-food and 16-seed.
White arrows indicate chiral fibers (helical twist, which visually appears as breakages in even distances along the fiber). Blue and green arrows indicate the height profiles including chiral or achiral fibers, respectively.
The seeding efficiencies of 16 and 26 on (1+2)-food were also compared directly. For this end, a pre-oxidized DCL prepared from an equimolar amount of building blocks 1 and 2 ([1] = [2] = 2.0 mM) was split into two equal portions, the first treated with equal amounts of 16 and 26 (10 mol % each), while the other sample served as a not seeded control. As expected, mixed hexamers corresponding to the food composition emerged (Figure 4. 10) whereas no emergence was observed in the control sample. In the seeded sample, rapid disintegration of 16 was observed (Figure 4. 10B, black squares), whereas the amount of 26 remained constant (Figure 4. 10B, blue diamonds) during the course of the experiment. It
122
has to be noted that the process was substantially slower than in the case of the treatment with single seeds (Figure 4. 8C), but as the composition of the sample did not differ significantly from the (1+2)-food + 26-seed combination, it must be the result of the inherent variability of the processes (arising presumably from the inhomogeneity of the seeds).
We speculated that, as the 26 seeds are chiral, the seeding process might involve the transfer of chiral information. The treatment of (1+2)-food with 26 seems to represent an interesting scenario, where the food would give rise to achiral assemblies of replicators via the autocatalytic pathway (Figure 4. 7F) but might result in chiral assemblies upon treatment with a chiral seed via a cross-catalytic pathway. Thus, we investigated the morphology of the seeded samples with AFM (Figure 4. 11).
Intriguingly, the samples prepared by cross-seeding 2-food either with 16 (Figure 4. 11A) or (1+2)6 (Figure 4. 11B) give rise to achiral fiber-like assemblies. These observations indicated that, although the main product in both cases was 26 (Figure 4. 9C-D), which would assemble into chiral fibers via the autocatalytic pathway, this scenario could not be realized via the cross-catalytic pathway as the chiral input (seed) is not present. In contrast, in the sample prepared by cross-seeding (1+2)-food with 26, numerous chiral fibers were observed (Figure 4. 11C, white arrows; chirality is visible). That is, upon autocatalysis (with (1+2)6) and cross-catalysis (with 26) the resulting hexamers are formed in the same molecular composition but in a different supramolecular arrangement. As a control, the sample prepared by cross-seeding (1+2)-food with 16 (no chiral input) does not show chiral fibers. Interestingly, the resulting fibers are more uniform as expected for the mixed hexamer assemblies (Figure 4. 7) and their height is ca. 3.5 nm, which is between that of the 16 nanoribbons (2 nm) and the 26 fibers (5 nm).
4.2.2. Investigation of the (1+3)-system
From the detailed investigations of the system composed of 1 and 2 we can conclude that the self-replicating species (26 peptide fibers) with stronger intermolecular forces and more ordered supramolecular assemblies imposes its supramolecular properties on the whole system when the two building blocks are mixed. Assemblies of mixed replicators adopt morphologies similar to the peptide fibers already at relatively low (20-30 mol %) amounts of peptide building block. Moreover, the mixed and the pure peptide replicators are mutually cross-catalytic towards each other, while both of them are not mutually cross-catalytic towards the 16 nanoribbons. Finally, we observed that the 26 assemblies can partially transfer their chirality on the mixed replicators via cross-catalysis. This prompted us to investigate the question whether peptide replicators are strong enough to transmit other information such as macrocycle size. Such investigations were not possible in the system studied above as for both building blocks the replicating macrocycles were hexamers. To study this question, we undertook a similar study based on building block 1 and peptide building block 3, the latter featuring an alanine amino
123
acid instead of phenylalanine at position 5. Building block 3 forms self-replicating macrocycles following a mechanism similar to 2. However, in the absence of co-solvents (i.e. in pure aqueous buffer) the replicating macrocycle is the octamer 38. Thus, we asked ourselves whether the peptide replicator 3 also imposes its self-assembly properties upon the mixed systems and if this is indeed the case, does this include the transformation of ring size. In this manner, one would expect that upon seeding 1-or (1+3)-food with 38, mixed octamers could form, which are unable to form autonomously. If, however, On the other hand, if 16 could impose its (hexameric) stoichiometry upon the peptide-containing assemblies, whereby the morphology would remain determined by the latter, we would see an intriguing example of partial information transfer.
Figure 4. 12. Representative UPLC chromatograms of chromatograms of stirred DCLs prepared from building blocks 1 and 3
(total building block concentration kept at 4.0 mM) containing A) 0, B) 3, C) 5, D) 10, E) 20, F) 30, G) 40, H) 50, I) 60, J) 70, K) 80, L) 90, M) 100 mol % of building block 1 ; as well as the N) corresponding distributions of the relative amounts of mixed
1n36/8-n hexamers and octamers obtained from the experiments above and O) calculation assuming a binomial distribution of
124
To this end, we prepared stirred DCLs from 1 and 3 with gradually changing stoichiometries as in the case of 1 and 2 (vide supra). The UPLC chromatograms after 7 days (after no further change in the DCL composition was observed) and the corresponding distribution of the formed macrocycles are shown on
Figure 4. 12. Whereas non-agitated mixed DCLs give rise to trimers and tetramers, agitation leads to
the emergence of hexamers and octamers (Figure S4. 58). Similarly to the (1+2)-system, the time evolution of the hexamers can be characterized by a lag phase followed by a sudden increase in their concentration, where all hexamers evolve as one set of replicators (Figure S4. 59). The relative amount of octamer 38 decreases rapidly in the presence of even minute amounts (< 10 mol%, Figure 4. 12A-D) of 1 and is incorporated into the mixed 1n36-n hexamers. Interestingly, the hexamer 36 is formed in substantial amounts between 3 and 30 mol% 1, although in aqueous buffer, the octamer formed preferentially, whereas the hexamer only becomes a major product upon changing the solvent composition.[59] On the other hand, the only mixed octamer present in detectable, although minor amounts (< 6 mol%) is 1137 (Figure 4. 12B-C). These experiments suggest that the stoichiometry is more strongly dictated by 1 than by 3. The amount of the other hexamers approximately follows the statistical distribution (Figure 4. 12B-O) The amount of 1531 remains below the expected values (Figure 4. 12N-O, red circles) mainly due to the concomitant formation of the mixed pentamers, which incorporate a substantial amount (up to 52 mol%, Figure 4. 12J) of the DCL material. Overall, however, no substantial deviation from the statistical distribution can be observed; thus, similarly to the previous system, no self-sorting takes place on the molecular level.
Figure 4. 13. Maximum thioflavin T (blue trace) and Nile Red (red trace) intensities of stirred DCLs prepared from building
125
In order to investigate the changes of the supramolecular properties of the (1+3)-system, similar spectroscopic and AFM studies as for the previous system were undertaken. The results of the parallel ThT/Nile Red assay (Figure 4. 13) showed slightly different trends for both fluorescent probes. The Nile Red assay indicates that the aggregates resulting from the mixed hexamers and octamers become more lipophilic at lower 1-content (60 mol % 1) than in the (1+2)-system (80 mol% 1). The ThT assay shows more marked differences: the fluorescent signal begins to drop significantly only at high 1-content (80 mol %) in contrast to the (1+2) – system (10 mol%). These results tempted us to assume that the mixed hexamer assemblies could retain their beta-sheet character even upon incorporation of large relative amounts of 1.
Figure 4. 14. CD intensity (at 211 nm) measured in DCLs containing mixed hexamers of 1 and 3 ([1]+[3] = 4.0 mM, Figure 4. 12) as a function of composition.
CD spectroscopy, however, indicated that the helical beta-sheet signature disappears at substantially lower (10-20 mol %) amounts of 1 (Figure 4. 14, Figure S4. 60). This seemingly contradictory result can be resolved by considering that ThT might bind non-specifically to the assemblies (i.e. not only to β-sheet motifs but to other domains as well) or that it might bind to β-β-sheet domains even if they are present only in minor amounts. In contrast, interaction of polarized light with the sample is less prone to yield such artifacts (i.e. CD spectroscopy indicates the global average chirality of the sample.)
In order to resolve this ambiguity, we undertook a detailed AFM investigation of the DCLs prepared (Figure 4. 15). DCLs containing minor amounts of 1 (comprised mainly of 38 and 36) did not feature helical assemblies (Figure 4. 15A-B) as in the case of the 1+2-system, despite of their characteristic CD signatures. More remarkably, these fibers were substantially shorter and more aggregated than the
26-126
fibers, which renders the inspection of their possible helicity more challenging. The strong aggregation also complicated the precise determination of the assembly height: the AFM images feature aggregates comprised of a diverse number of vertically stacked fibers, with heights ranging from 4 to 8 nm (Figure
4. 15A-B).
Figure 4. 15.AFM micrographs and selected height profiles of stirred DCLs prepared from building blocks 1 and 3 (total building
block concentration kept at 4.0 mM) containing A) 0, B) 5, C) 10, D) 20, E) 30, F) 40, G) 50, H) 60, I) 70, J) 80, K), 90 L) 100 mol % of building block 1. Scale bar, 200 nm. Black arrows show the similar heights of the mixed hexamer nanoribbons (5-6 nm). Scale bar, 200 nm. Color code: Red: nanoribbons, Blue: short, associated fibers, Green: achiral fibers.
127
In samples containing 10-70 mol % 1 (Figure 4. 15C-I), ribbon-like structures were detected. These objects were, however, clearly distinguishable from the 16 nanoribbons, based on their height (5-6 nm, versus 2 nm for 16). Notably, the height distribution was more uniform (Figure 4. 15C-I, black arrows) than for the assemblies comprised mainly of 3. At 80 mol% 1 (Figure 4. 15J), two clearly distinguishable assemblies were present (similarly to the corresponding case of the 1+2-system, Figure 4. 7J): Narrower nanoribbons with a height of 5 nm (green traces, arising most possibly from the mixed hexamers) were interspersed with wider nanoribbons with a height of 2 nm (red trace, corresponding most possibly to assemblies of 16).Finally, assemblies with high 1-content feature nanoribbons with a height of 2 nm (Figure 4. 15K-L). TEM images analogously feature three different morphologies across the stoichiometric window: strongly aggregated, 20-60 nm long and 5-15 nm wide fibers at low amounts of
1 (Figure S4. 61A); less aggregated, 80-160 nm long and 25-50 nm wide ribbons at intermediate levels
of 1 (Figure S4. 61E-F) and several hundred (>200) nm long nanoribbons, consisting of several uniform, 6-7 nm wide stripes (Figure S4. 61I). These results show that, analogously to the (1+2)-system, the (1+3)-system features three different supramolecular assembly modes, two at the narrow extremes (5-10 and 80-(5-100 mol% of 1) of the stoichiometric window and one in the broad range of intermediate compositions. The absence of coexisting structures points out that the structure of the supramolecular assemblies of replicators undergo a phase-change like transformation with linearly changing stoichiometry.
Figure 4. 16. Comparison of self-and cross-seeding experiments, showing the time evolution of non-agitated DCLs containing
A) 1-food, B) 3-food, C) (1+3)-food, seeded with 16 (red circles), 38 (blue triangles), 36* (purple pentagons), (1+3)6 (cyan
triangles) as well as a non-seeded control (black squares).
Based on the results from the cross-seeding experiments from the (1+2)-system, we hypothesized that the replicators held together by stronger secondary interactions dictate not only the morphology but also the stoichiometry in cross-catalysis, i.e. that the outcome of the cross-seeding is enthalpically driven.
128
Based on this assumption we expected that in the (1+3)-system 38 could catalyze the growth of mixed octamers. Quite on the contrary, however, we observed that for mixed replicators, already minor amounts of 1 can reverse the preferred stoichiometry, i.e. produce hexamers instead of octamers (Figure 4. 12B-C). This observation is in line with the reasonable assumption that the assembly of larger macrocycles is entropically less favored than that of smaller ones. We were therefore interested whether the outcome of the cross-seeding is enthalpically or entropically driven. Thus, we set out to perform similar seeding experiments as in the case of the (1+2)-system to probe whether by cross-seeding it is possible to catalyze the growth of replicators with stoichiometries which are inaccessible by self-seeding, i.e. the growth of 1-rich octamers and 3-rich hexamers.
Figure 4. 17. Relative amounts of 36 (A,C) and 38 (B,D) upon seeding of 3-food (A,B) and (1+3)-food (C,D) with different seeds.
For this purpose, food samples with similar stoichiometries as for the (1+2)-system were used, denoted as 1-food, 3-food and (1+3)-food. Four different seeds were used, denoted as 16, 38, 36* and
(1+3)6. 16 and 38 were composed of the corresponding pure hexamer and octamer, respectively. Seeds (1+3)6 and 36* contained mixed hexamers in a statistical distribution, prepared from a 1:1 and a 9:1
129
mixture of 3 and 1, respectively (compositions shown at Figure 4. 12C and H, respectively). Seed 36*
contained more than 50 mol% 36 and only traces of 38, i.e. it is composed mainly of the unfavored replicator formed from 3 (under aqueous conditions[59]), and upon addition to 3-food should result in a sample in which building block identity and replicator stoichiometry are decoupled.
The results of the cross-seeding experiments show that 1) as in the case of the (1+2)-system, 16 is mutually not cross-catalytic with the peptide-containing replicators (Figure 4. 16A), 2) 36* and (1+3)6 has a detectable but less pronounced cross-catalytic effect on the formation of 3-only replicators (Figure 4.
16B) and 3) similarly, 38 exerts only a minor weak cross-catalytic effect on the formation of mixed hexamer replicators (Figure 4. 16C). More specifically, by seeding 3-food with hexameric seeds (36* or
(1+3)6) mainly the formation of 36 (and not that of 38) is catalyzed, whereas upon seeding with the octameric 38 seed, mostly 38 (with only traces of 36) is formed (Figure 4. 17A, B). Similar trends can be observed for the (1+3)-food (Figure 4. 17C, D). In other words, no significant cross-catalysis takes place between replicators of different stoichiometries. Notably, only trace amounts of 1137 were detected during cross-seeding experiments with 38 seed (Figure S4. 62D). Similarly to the (1+2)-system, in the presence of non-self food, 16 undergoes rapid disintegration (Figure S4. 62A,C), whereas the amount of 38 and (1+3)6 remains approximately constant (Figure S4. 62B, D).
The experiments described above show that, although mutual cross-catalytic relationships exist between the peptide-containing systems, their relationship is not fully symmetrical as in the (1+2)-system. Specifically, octamers are only formed when both the required seed (38) and the food (3-food) are present. In other words, mixed octamers tend not to form in this system, in contrast to a reported mixed DCL based on 1 and an octamer-forming peptide replicator (similar to 3).[51] Thus, although the replicator held together by stronger interactions (enthalpic effect) imposes its morphology upon the assembly formed by cross-catalysis, the same does not hold for the stoichiometry, which can be ascribed to an effect of entropy.
4.3. Conclusion
In this study we investigated the interplay between self-replicating molecules whose formation is based on the same dynamic covalent chemistry but are assembled through secondary interactions of different strength and directionality and, consequently, show different morphologies. All building blocks form self-replicating oligomers; 16 is organized into nanoribbons held together by relatively unspecific and weak hydrophobic interactions, whereas 26 and 38 form helical fibers stabilized by β-sheet interactions between the side chains. Remarkably, DCLs prepared from samples containing the building blocks in various ratios all give rise to self-replicating entities. The resulting mixed (1+2)6 and (1+3)6 hexamers, as an ensemble, display kinetic properties characteristic of self-replicators. Moreover, in the
130
case of the (1+3)-system, the original stoichiometry of the self-replicators arising from 3 is not preserved: the amount of 38 decreases dramatically even in the presence of 5 mol% of 1 and only traces of mixed octamers are formed. In the mixed DCLs no preferred replicator emerges at the molecular level, i.e. the hexamers are formed according to a statistical distribution based on the stoichiometry of the building blocks. At the supramolecular level, however, such continuous transition from one morphology to the other cannot be observed: The morphology of the single-building block replicators is not maintained upon the incorporation of even a small amount (10-20 mol%) of the other building block. Instead, the mixed replicators self-assemble into nanostructures of an intermediate morphology, consisting of non-helical fibers. The morphological differences are translated into intricate cross-catalytic relationships between the replicators (Table 4. 1A). In the case of the (1+2)-system, 16 cannot catalyze the formation of 26 or (1+2)6 from the respective food niches and decomposes to give mixed trimers and tetramers. This lack of cross-catalytic activity is indicative of the low strength of favorable interactions between the seed assemblies and food molecules. Similarly, neither 26 nor (1+2)6, can catalyze the emergence of 16, possibly for the same reason. On the other hand, 26 and (1+2)6 show a mutual cross-catalytic relationship. This is most possibly due to similar strength and orientation of supramolecular interactions within the aggregates, as the β-sheet interactions are partially preserved upon incorporation of 1 units.
Table 4. 1. Auto- and cross-catalytic relationships between replicators A) 16, 26 and (1+2)6; B) 16, 38, 36* and (1+3)6.
empty = no catalytic relationship; blue = positive catalytic relationship, preservation of food and seed stoichiometry; red = positive catalytic relationship, loss of either food or seed stoichiometry
For the (1+3)-system, similar cross-catalytic relations could be observed. However, cross-catalysis between replicators of different stoichiometries is less effective than between those of similar stoichiometries: whereas (1+3)6 and 36* can cross-catalyze the formation of 36 under conditions where
131
38 is the preferred self-replicator, 38 cannot induce the formation of octameric species. This shows that although enthalpic effects (interaction strength) might determine which self-replicator impose its supramolecular properties (morphology and chirality) on the system, entropic effects also play a decisive role, determining molecular properties (stoichiometry).
From a broader perspective, this study represents the first step toward scenarios of self-replication featuring markedly different replicators, which can nevertheless engage in mutual competition. However, although the self-assembly properties of the replicators are highly distinctive, the displayed differences are not characteristic enough to construct a scenario with molecularly well-defined replicators. More closely, we expected that due to the pronounced differences in 1) assembly mode (nanoribbons vs. nanofibers) and 2) stoichiometry (1 forms hexamers, whereas 3 forms octamers) the replicators arising from the two food niches might emerge as two distinct sets, with different compositions and likely in temporal succession. In ideal case, two replicators, each with a well-defined stoichiometry (see Chapter 5), might emerge after each other (specialist replicators). In contrast, we observed in this chapter that in all cases, replicators emerge at the same time and stoichiometry (generalist replicators). This situation hampers the creation of more advanced life-like scenarios, in which two (or more) replicators coexist and e.g. compete for the same building block or display a function cooperatively, which would not be possible in isolation. In order to create a system comprised of specialist replicators, replicators with much pronounced differences in the kinetic (length of lag phase) and/or thermodynamic (strength and directionality of secondary interactions holding together the assemblies) properties have to be designed. For example, nucleotide-appended building blocks with a large number of highly directional complementary hydrogen bonds[60] might result in the formation of thermodynamically preferred assemblies with specific secondary interactions. In this case, the formation of mixed replicators (containing both nucleobase-appended and nucleobase-free building blocks) would result in a high energetic penalty compared to the formation of self-sorted replicators.. Moreover, disulfide exchange is a fast reaction compared to nucleation. Thus, either conditions have to be developed where the rate of disulfide exchange is diminished, or the kinetic barrier of nucleation events is lower; or, as an alternative solution, dynamic chemistries with lower rates than thiol-disulfide exchange have to be utilized.
We envision that taken our results and the improvements suggested above, the difference in assembly strengths and the consequent asymmetric cross-catalytic relationships can be further elaborated to more intriguing systems. For example, if a specific chemical functionality (e.g. catalytic activity applied to an external substrate) is possessed by only one of the replicators and can be coupled to replication, it becomes possible to mimic simple ecological scenarios, e.g. parasitic relationships between the two replicators. Research toward this direction is currently ongoing in our laboratory and will be reported in due course.
132
4.4. Supporting Information
4.4.1. General remarksAll chemicals, unless otherwise stated, were purchased from Sigma-Aldrich and used as received. Acetonitrile (ULC-MS grade), water (ULC-MS grade), DMF (HPLC grade) and trifluoroacetic acid (HPLC grade) were purchased from Biosolve BV. Building blocks 2 and 3 were purchased from Cambridge Peptides Ltd. (Birmingham, U.K.) by coupling 3,5-bis(tritylthio)-benzoic acid, which was synthesized via a previously reported procedure[47] at the N-termini. Phosphate buffer (50 mM, pH = 8.16) was
prepared by dissolving Na2HPO4.7H2O (6.43 g, 24.0 mmol) and NaH2PO4.H2O (0.166 g, 1.20 mmol) in 500 mL doubly distilled
water. The pH was adjusted either with 1 M NaOH or with solid NaH2PO4.H2O (ca. 90 mg). The buffer was filtered through a
Whatman 0.2 μm PTFE syringe filter (GE Healthcare) just prior to library preparation. This filtrations step is very important to prevent the growth of microorganisms in the buffer, which otherwise sets in after ca. 5 days. We observed that after this period the peptide building blocks begins to get hydrolyzed, presumably by proteases released by microorganisms.
UPLC analysis
Generally, samples were diluted to 400 μM with water (UPLC grade) and 10 μL of this diluted sample was injected for UPLC analysis. Samples containing building block 1 as food were diluted to 400 μM with a 7:3 V:V mixture of water (UPLC grade) and isopropanol (HPLC grade), containing 0.1 V/V% TFA (HPLC grade). For the other seeding experiments, 5 μL of the sample was transferred into a UPLC insert, diluted by the UPLC instrument with 20 μL solvent (10 V/V % acetonitrile in water) and after 1 minute of mixing time, 5 μL of this sample was injected onto the UPLC column.
UPLC measurements were performed on a Waters Acquity H-class system equipped with a PDA detector. UPLC analyses, unless otherwise stated, were performed on an Aeris 1.7 μm XB-C18 (150 × 2.1 mm) column, purchased from Phenomenex, using UPLC-MS grade water (eluent A) and UPLC-MS grade acetonitrile (eluent B), containing 0.1 V/V % TFA as a modifier. A flow rate of 0.3 mL/min (unless otherwise indicated) and a column temperature of 35 °C were applied. Spectra were recorded at a detection wavelength of 254 nm.
Methods for the analysis of DCLs prepared from 1 and 2: t / min % A 0 90 1 90 2 80 5 73 20 54 22 5 24 5 25 90 28 90 Method for the analysis of DCLs prepared from 1 and 3:
t / min % A 0 90 1 90 2 80 6 73 23 59 27 15 28 5 29 5 30* 90 32* 90 * flow rate: 0.45 ml/min
133
UPLC-MS analysis
UPLC-MS measurements were performed using a Waters Acquity UPLC H-class system coupled to a Waters Xevo-G2 TOF. The mass spectrometer was operated in the positive electrospray ionization mode with the following ionization parameters: capillary voltage: 3 kV, sampling cone voltage: 20 V, extraction cone voltage: 4 V, source gas temperature: 120°C, desolvation gas temperature: 450°C, cone gas flow (nitrogen): 1 L/h, desolvation gas flow (nitrogen): 800 L/h.
Negative-staining Transmission Electron Microscopy
Samples were diluted to 40-fold using UPLC grade water. A small drop (5 µL) of sample was then deposited on a 400 mesh copper grid covered with a thin carbon film (supplied by Agar Scientific). After 30 seconds, the droplet was blotted on filter paper. The sample was then stained with a solution of 2% uranyl acetate (4 µL) deposited on the grid, subsequently washed and blotted on filter paper after 30 seconds. The staining procedure was repeated a second time, this time without the washing and blotting step. The grids were observed in a Philips CM12 electron microscope operating at 120 kV. Images were recorded on a slow scan CCD camera.
Fluorescence Measurements
Fluorescence measurements were performed on a JASCO FP 6200 fluorimeter using quartz cuvettes with 1 cm path length. For the ThT measurements, a freshly prepared solution of thioflavin T (dissolved to 2.2 mM in 50 mM borate buffer at pH = 8.2 and filtered with a 0.2 µm syringe filter) was diluted to 22 µM with the same buffer. 450 µL of this diluted solution was transferred to the cuvette, followed by the addition of 80 µL of sample (diluted with borate buffer to 80 µM total building block concentration just prior to the measurement). Spectra were recorded after an incubation time of 2 minutes. The excitation wavelength was set at 440 nm and spectra were recorded in the range of 480-700 nm with a slit width of 5 nm, using a cutoff filter at 480 nm. The blank spectrum (i.e. fluorescence of the corresponding dye in solvent only) was subtracted from each spectrum.
For the Nile Red measurements, a freshly prepared solution of Nile Red (dissolved to 160 µM in UPLC grade methanol) was used. A sample volume of 10 µL (4.0 mM original total building block concentration, diluted to 80 µM with borate buffer) was added to a mixture of borate buffer (480 µL) and 10 µL of Nile Red solution in the cuvette. Spectra were recorded after an incubation time of 2 minutes. The excitation wavelength was set at 553 nm and spectra were recorded in the range of 600-750 nm with a slit width of 5 nm, using a cutoff filter at 575 nm. All spectra were recorded as an average of 3 repeats. The blank spectrum (i.e. fluorescence of the corresponding dye in solvent only) was subtracted from each spectrum.
CD Spectroscopy
Samples were diluted to 250 μM with water (UPLC grade). Measurements were performed on a JASCO J-715 CD spectrophotometer using a HELMA quartz cuvette with a path length of 1 mm. Spectra were recorded in continuous mode between 190-300 nm at room temperature at a scan rate of 125 nm/min with 2 nm step-interval in millidegrees. All recorded spectra were an average of 3 repeats with solvent spectrum subtracted from each spectrum.
Atomic Force Microscopy
AFM samples were prepared by depositing 100 µL of the sample (diluted to 10 µM building block concentration with UPLC grade water) onto a clean mica surface (Grade V1, Van Loenen Instruments). Subsequently, the solvent was evaporated in a gentle stream of air in ca. 10 minutes. The surface was then washed with 100 µL of UPLC grade water and blotted into a piece of paper twice and finally air-dried. The AFM measurements have been performed using a Bruker Multimode 8 instrument in Scan Asyst-Air imaging mode. Measurements were performed in air at room temperature. As a probe, a ScanAsyst Air (Bruker) silicon tip on a nitride cantilever was used with the following parameters: length: 115 µm, width: 25 µm, resonance frequency: 70 kHz, force constant: 0.4 N/m. The images were recorded with frequencies between 0.5 and 1.5 Hz and analyzed with NanoScope Analysis 1.50 software (Bruker Corporation, 2015).
134
4.4.2. Library preparation
4.4.2.1. DCLs prepared from 1 and 2
Samples prepared in buffer only:
The building blocks (8.8 µmol) were separately dissolved borate buffer (2200 µL). The dissolution of 1 required sonication for 20-30 minutes to give a clear solution (or occasionally a fine suspension). The resulting samples of building block 1 were immediately portioned into 11 separate vials (see below), followed by the addition of the corresponding portions of the peptide building blocks. The samples were stirred at 1200 rpm. The volumes of the portions were the following:
Mol % of 1 V(1, 4.0 mM) / µL V(2 or 3, 4.0 mM) / µL 0 0 400 10 40 360 20 80 320 30 120 280 40 160 240 50 200 200 60 240 160 70 280 120 80 320 80 90 360 40 100 400 0 Samples prepared in 10% DMF:
Building block 1 (8.8 µmol) was dissolved in DMF (HPLC grade, 220 µL). Building blocks 2 and 3 (8.8 µmol) were separately dissolved in the corresponding buffer (2200 µL). The stock solutions of the building blocks were portioned into 10 separate vials (see below) and the volume was adjusted with DMF and buffer to give solutions with the same volume and DMF content. The volumes of the portions were the following:
Mol % of 1 V(1, 40 mM) / µL V(2 or 3, 4.0 mM) / µL V(DMF) / µL V(buffer) / µL 10 4 360 36 0 20 8 320 32 40 30 12 280 28 80 40 16 240 24 120 50 20 200 20 160 60 24 160 16 200 70 28 120 12 240 80 32 80 8 280 90 36 40 4 320 100 40 0 0 360
A sample containing 100 mol % of building block 2 was prepared by dissolving the monomer in a 10:90 V/V % mixture of DMF and the corresponding buffer. The samples were stirred at 1200 rpm.
4.4.2.2. DCLs prepared from 1 and 3
The DCLs with increasing relative amount of 1 were prepared similarly to those described for the (1+2) system, with two important differences. First, instead of borate buffer, phosphate buffer (50 mM, pH = 8.2) was used (always filtered through a 0.2 µm syringe filter before use, see General remarks). Second, the libraries were always prepared without the addition of DMF as this hampered the formation of 38 in a 3-only library. Notably, the DCL material was converted into replicators (pentamers,
hexamers and octamers) much faster than for the (1+2) system (6-7 days instead of 14-15 days).
4.4.3. Seeding experiments 4.4.3.1. DCLs prepared from 1 and 2
Seeding experiments involving 1 as food were performed by dissolving 1.91 mg (5.77 μmol) 1 in a mixture of 1191 μL borate buffer and 144 μL isopropanol (IPA). Subsequently, a solution of NaBO3 (108 μL, 40 mM, in borate buffer) was added in
135
100% was indicated by UPLC analysis. Thus, 28.9 μL of 10 mM TCEP solution in borate buffer was added to reach an overall oxidation level of 95%. The resulting solution was 4.0 mM regarding 1 and contained 10 V/V % IPA. Four equal portions (300 μL each) were taken and the corresponding aliquots were seeded with 30 μL of 4.0 mM samples of 16, 26 or (1+2)6 as well as
with 30 μL buffer (blank). The (1+2)6 sample was prepared by stirring an equimolar mixture of 1 and 2 ([1] = [2] = 2.0 mM) in
aqueous borate buffer until UPLC analysis indicated the conversion of library material into hexamers (ca. 14 days). The resulting solutions were left standing. At every 2 hours, 5 μL of the corresponding DCL was diluted with 45 μL of a 9:1 V:V mixture of water and DMF, containing 0.1 V/V % TFA and injected on the UPLC. Seeding experiments involving 2 as food were performed by dissolving 4.85 mg (4.91 μmol) 2 in 1135 μL borate buffer. Subsequently, a solution of NaBO3 (92 μL, 40 mM, in borate
buffer) was added in equal portions over 22 minutes. This corresponded to 75% oxidation. However, due to the presence of air, an oxidation level of 98% was indicated by UPLC analysis. Thus, 19.6 μL of 20 mM TCEP solution in borate buffer was added to reach an overall oxidation level of 90%. The resulting solution was 4.0 mM regarding 2. Four equal portions (280 μL each) were taken and the corresponding aliquots were seeded with 28 μL of 4.0 mM samples of 16, 26 or (1+2)6 as well as with 30 μL
buffer (blank). The samples were left standing. For UPLC analysis (ca. every 12 hours), a 5 μL of the sample was transferred into a UPLC insert, diluted by the UPLC instrument with 20 μL solvent (10 V/V % acetonitrile in water) and after 1 minute of mixing time, 5 μL of this sample was injected onto the UPLC column.
Seeding experiments involving (1+2) as food and single seeds were performed by dissolving 1.05 mg (3.17 μmol) 1 in 761 μL borate buffer. Subsequently, a solution of NaBO3 (32 μL, 40 mM, in borate buffer) was added in equal portions over 12
minutes. This corresponded to 40% oxidation. In a separate vial, 2.55 mg (2.75 μmol) 2 was dissolved in 600 μL borate buffer. Subsequently, a solution of NaBO3 (45 μL, 40 mM, in borate buffer) was added in equal portions over 20 minutes. This
corresponded to 70% oxidation. To the latter DCL was added a 600 μL aliquot of the former DCL and the mixture was left to equilibrate and oxidize for 3 days. Subsequently, 24 μL of a 20 mM TCEP solution was added to reach an oxidation level of 90 %. Subsequent treatment was similar to that of the DCL prepared from 2.
Seeding experiments involving (1+2)-food as well as simultaneous seeding with 16 and 26 were performed similarly to those
involving single seeds. Aliquots of 240 μL were taken and seeded with 24-24 μL of each seed (4.0 mM) or 48 μL buffer (blank).
4.4.3.2. DCLs prepared from 1 and 3
The experiments were performed similarly to those described for the (1+2) system with a few differences: - No co-solvents were used for food preparation.
- Food samples with a volume of 220 μL was seeded with 22 μL of the corresponding seed (or blank). - In seeding experiments involving 1 as food, the libraries were sampled every 170 minutes.
- For seeding experiments involving (1+3) as food, food preparation was performed as follows: Building block 3 (3.98 mg, 4.36 μmol) was dissolved in 1009 μL of phosphate buffer. Building block 1 (1.48 mg, 4.47 μmol) was dissolved in 1118 μL of phosphate buffer. The solution of 4 was oxidized to 80 mol% by adding NaBO3 (82 μL, 40 mM, in phosphate
buffer) in equal portions over 20 minutes. A volume of 660 μL of 1 (4.0 mM, freshly prepared in phosphate buffer) and 660 μL of 4 (4.0 mM, freshly prepared in phosphate buffer) were mixed and left standing for 4 days until UPLC analysis indicated 1.5 mol% thiol content. To this solution, TCEP (40 mM in phosphate buffer 15.72 μL) was added to reach ca. 10 mol% thiol content. This solution was used as food.
136
4.4.4. LC-MS analyses of DCLs
The lowest m/z (exact mass) of each isotopic envelope was selected for comparison of calculated and observed masses.
4.4.4.1. DCLs prepared from 1 and 2
Figure S4. 1. Mass spectrum of 2 (tR =9.52 min) from the LC-MS analysis of a DCL made from 1 and 2.
m/z calculated: 760.35 [M+H]+; m/z observed: 760.20 [M+H]+.
Figure S4. 2. Mass spectrum of 1 (tR =10.67 min) from the LC-MS analysis of a DCL made from 1.
m/z calculated: 332.10 [M+H]+, 354.08 [M+Na]+; m/z observed: 332.23 [M+H]+, 354.20 [M+Na]+.
Figure S4. 3. Mass spectrum of 24 (tR =12.21 min) from the LC-MS analysis of a DCL made from 1 and 2.
137
Figure S4. 4. Mass spectrum of 26 (tR =12.97 min) from the LC-MS analysis of a DCL made from 1 and 2.
m/z calculated: 1137.00 [M+4H]4+, 1515.33 [M+3H]3+; m/z observed: 1136.97 [M+4H]4+, 1515.43 [M+3H]3+.
Figure S4. 5. Mass spectrum of 23 (tR =14.07 min) from the LC-MS analysis of a DCL made from 1 and 2.
m/z calculated: 758.34 [M+3H]3+; m/z observed: 758.16 [M+3H]3+.
Figure S4. 6. Mass spectrum of 1125 (tR =14.25 min) from the LC-MS analysis of a DCL made from 1 and 2.
138
Figure S4. 7. Mass spectrum of 1123 (tR =14.44 min) from the LC-MS analysis of a DCL made from 1 and 2.
m/z calculated: 867.70 [M+3H]3+; m/z observed: 867.73 [M+3H]3+.
Figure S4. 8. Mass spectrum of 1124 (tR =14.64 min) from the LC-MS analysis of a DCL made from 1 and 2.
m/z calculated: 1119.80 [M+3H]3+; m/z observed: 1119.95 [M+3H]3+.
Figure S4. 9. Mass spectrum of 1225 (tR =15.23 min) from the LC-MS analysis of a DCL made from 1 and 2.
