Controlling the Kinetics of Self-Reproducing Micelles by Catalyst Compartmentalization in a Biphasic System

Compartmentalization of reactions is ubiquitous in biochemistry. Self-reproducing lipids are widely studied as chemical models of compartmentalized biological systems. Here, we explore the effect of catalyst location on copper-catalyzed azide–alkyne cycloadditions which drive the self-reproduction of micelles from phase-separated components. Tuning the hydrophilicity of the copper-ligand complex, so that hydro-phobic or -philic catalysts are used in combination with hydro-philic and -phobic coupling partners, provides a wide range of reactivity patterns. Analysis of the kinetic data shows that reactions with a hydrophobic catalyst are faster than with a hydrophilic catalyst. Diffusion-ordered spectroscopy experiments suggest compartmentalization of the hydrophobic catalyst inside micelles while the hydrophilic catalyst remains in the bulk aqueous phase. The autocatalytic effects observed can be tuned by varying reactant structure and coupling a hydrophilic alkyne and hydrophobic azide results in a more pronounced autocatalytic effect. We propose and test a model that rationalizes the observations in terms of the phase behavior of the reaction components and catalysts.


■ INTRODUCTION
Compartmentalized reactions may occur inside nano-or microscale species such as micelles, vesicles, and emulsion droplets. 1 These compartments can serve as microscale reactors and provide a mechanism for controlling the interactions between components of heterogeneous systems. 2 Studying compartmentalized reactions may help us understand biochemical processes and how to control chemical reactivity. 3−5 Living cells are compartmentalized out-of-equilibrium systems which are able to control complex multistep reactions, such as NADPH metabolism, which relies on the ability to separate different components. 6 Similarly, gluconeogenesis likely involves three intracellular compartments performing different reactions. 7,8 Lipid-based compartments have been studied as simple models of cellular membranes 9−11 and display complex aggregation behavior including deformation and division. 12−15 The chemical self-reproduction of synthetic micelles and vesicles has been used to drive the growth and division processes 16−18 and serve as minimal metabolic networks where compartmentalized reactions generate self-organizing components of the compartment. 19,20 In physical autocatalytic processes surfactants are initially slowly formed through the reaction of two phase separated reagents at the interface of a biphasic system ( Figure 1A, step 1). 21−31 Once a critical concentration of the products is reached, supramolecular aggregates form allowing the reaction components to interact by compartmentalization and subsequent solubilization into the opposing phase, facilitating further surfactant formation (step 2). 24 The formation of these aggregates can probably be viewed most simply as an extension of the interface where phase separated reagents are capable of interacting. This concept was first explored by Luisi and co-workers where they reported on the alkaline hydrolysis of ethyl caprylate which yielded sodium caprylate micelles that enhanced the rate of hydrolysis. 22 Later Sugawara and co-workers developed a system of self-reproducing giant vesicles in a biphasic system through a dehydrocondensation reaction between a lipophilic aniline precursor and a amphiphilic aldehyde catalyzed by a amphiphilic acid catalyst. 29 However, the systems studied thus far did not allow for facile control over the compartmentalization of a secondary catalyst which may enable the development of more sophisticated reaction networks. Frećhet and coworkers have used the polarity of compounds to sequester reagents in different phases allowing for the development of a one-pot asymmetric cascade reaction mediated by two catalysts. 32 We recently reported a system where we used coppercatalyzed alkyne−azide cycloadditions (CuAAC) as a secondary catalytic cycle in the self-reproduction of micelles ( Figure  1B). 33 The coupling of an excess of a hydrophobic alkyne that constitutes the organic phase as oil droplets on water with a hydrophilic azide in a biphasic reaction mixture showed complex behavior consistent with autocatalysis.
Here we show that the kinetic behavior of self-reproducing lipid aggregates can be controlled ( Figure 1C−E) by tuning the solubility properties of a Cu-catalyst where the autocatalytic cycle is completely dependent on the Cu-catalyst. We also examine different combinations of hydrophobic and hydrophilic alkyne/azide coupling partners. We rationalize our findings in terms of the system's phase behavior and how the catalyst and the reaction events are localized by compartmentalization. We propose a mechanistic model which explains under which conditions more or less dramatic autocatalytic effects are observed. The generality of this model is tested in a novel system with a phosphocholine polar headgroup.

■ RESULTS AND DISCUSSION
As previously reported, reaction between hydrophilic maltose azide 1 and hydrophobic aliphatic alkyne 2 via a CuAAC forms surfactant 3 (Figure 2A), which according to DLS and fluorimetry measurements forms micelles of around 7 nm at 2.86 mM ( Figures S21 and S23). These reactions involve "typical" aqueous CuAAC conditions: in situ reduction of CuSO 4 with sodium ascorbate using O-phenylenediamine as a   (Figure 2A; ligand A). Kinetic experiments, after considerable examination of physical parameters such as stirring speed and flask size/shape to give reaction kinetics suitable for study on convenient time scales, reveal a lag period followed by a subsequent period with a faster rate ( Figure 2B; green circles). Seeding the reaction at t = 0 with product 3 at concentrations above the CMC 34 removed the lag period and increased the rate of product formation ( Figure 2C; green circles). These observations are consistent with an autocatalytic mechanism. 35 Controlling Catalyst Compartmentalization by Tuning Ligand Hydrophobicity. Autocatalytic cycles involving a secondary catalyst would be expected to behave differently than cycles driven by spontaneous reactions. As this form of replication involves multiple solution phases, altering the solubility properties of the Cu-catalyst is expected to allow for control over the spatial location of the reaction by, for example, restricting it to the micelle interior. Such catalyst compartmentalization may provide insight into how to control chemical reactivity, develop complex reaction networks, and enable sophisticated models of cells. To this end we varied the hydrophobicity of the copper ligand.
Tristriazole ligands are known 36 to prevent oxidation and disproportionation of Cu(I) and also enhance catalytic activity compared to CuSO 4 /ascorbate alone. Interestingly, these ligands were discovered when autocatalytic effects were observed during mechanistic studies of the CuAAC reaction on polyvalent substrates. 37 The mechanism of this autocatalytic cycle was recently studied in depth. 38 Devaraj and coworkers have used self-reproducing tristriazole compounds to embed copper in a phospholipid bilayer and drive membrane growth and form self-assembling peptide nanomaterials. 30,31 These weakly binding polytriazole ligands are exceptional in water because they prevent the formation of unreactive polynuclear copper acetylides 40−42 and other more strongly binding ligands inhibit the reaction by denying azide access to the metal. 39 Furthermore, simple synthetic modifications to this class of ligand allow easy tuning of solubility without affecting catalytic activity. We chose two known ligands, one strongly hydrophilic (ligand B) and one hydrophobic (ligand C), to compare and contrast with the previously studied ligand A (Figure 2A).
In the unseeded reactions, ligand C accelerates the reaction almost 2-fold compared to ligand A ( Figure 2B; Table S3 for extracted reaction rates). Conversely the reaction with ligand B is slow compared to ligand A with a much longer lag period and gradual initial product formation. There is however a small reproducible amount of initial product formation observed using ligand B, which was further investigated by varying the equivalent of ligand B used ( Figure S44). A correlation between the length of lag period and the ligand used was observed, where 1 equiv led to a significant rate increase and use of 4 equiv of ligand gave no observed product in the first 6 h (after 5 days about 10% conversion was observed).
In the seeded reactions, ligand A shows significant productinduced rate enhancement and is clearly the fastest of the three reactions ( Figure 2C, Table S4 for extracted reaction rates). Ligand B shows an ∼2× rate compared to its unseeded reaction, and a slight lag period. Curiously seeded reactions with B did not show initial formation of a small amount of product as seen with unseeded reactions using B. Ligand C shows only a modest rate increase over the unseeded reaction.
Control Experiments. To eliminate the possibility that C is simply a more effective ligand than B, we performed the reaction in a 1:1 water/tert-butanol mixture to facilitate phase mixing ( Figure S43). Although the kinetic profiles may still appear to have some sigmoidal character, all three ligands gave comparable rates and the reactions went to completion within 2 h. This supports the assumption that phase behavior is responsible for the lag period and slower kinetics and that there is no inherent difference in efficiency between the ligands.
During our previous studies with A we found that 1dodecyne (2b) (likely because of its very low solubility in water) reacted slowly, with a lag period of over 3 h and a reaction time of several days ( Figure 3, blue circles). Here we show that using hydrophobic C gives a much faster reaction with 2b and completely removes the lag period ( Figure 3, red squares).
We probed the phase behavior of the reaction components using diffusion ordered spectroscopy (DOSY). A lower diffusion coefficient (D) is expected for an aggregated surfactant compared to free surfactant below the CMC. A compound associating with the micelle is expected to diffuse at a similar rate as the surfactant above the CMC, where tetramethylsilane (TMS) is added as a control for the diffusion rate of supramolecular aggregates. Control experiments show that our surfactants aggregate above their CMCs (Table 1, entries A−B), that alkyne 2 is probably fully associated with surfactant 3 above its CMC (entries C−D), and that azide 1 does not associate detectibly with 3 (E−G). Water-soluble complexes Cu-A and Cu-B diffuse slower than the respective free ligands (entries H−I and Table S1, entries N−O) and do not associate with the micelles (entries J−K) while the hydrophobic Cu-C complex binds strongly to the micelles (entries L−M). The diffusion rates of the free small molecules, such as the surfactant below its CMC, the azide, and free B are in agreement with their molecular weights according to Morris'  (Table S1). 43 Using the Stokes−Einstein relationship we calculate a hydrodynamic radius of 1.9 nm for micelles of surfactant 3 at 22 mM, which is in agreement with the data obtained from DLS. 44 Inverting the Polarity of the System. We inverted the polarity of the reaction components by synthesizing a hydrophilic, maltose-based alkyne 4 and a hydrophobic alkyl azide 5 ( Figure 4A) and examined the reactivity of this "inverse" system for ligands B and C. DLS and fluorimetry measurements showed that surfactant 6 forms micelles of around 6 nm at 15 mM ( Figures S22 and S24). Switching the polarity of the coupling partners results in a significantly slower rate and a pronounced lag period for both ligands ( Figure 4B). However, the trend of reactions involving C being much faster than those with B remains (see Tables S5 and S6 for extracted rates). Seeding the C reaction with product entirely eliminated the lag period and led to a remarkably fast reaction ( Figure   4C). Conversely, the reaction with B is slow and reaches completion only after several days. Even when seeded it does not reach completion within 1 day, unlike all the other seeded reactions we have studied.
When the reaction was performed in a 1:1 water/tertbutanol mixture to facilitate phase mixing ( Figure S45) complete conversion was observed within 30 min for both ligands. Interestingly, this is a faster rate than in the original system even though this reaction is significantly slower under biphasic conditions. DOSY experiments aimed at probing aggregation states in the inverse system show similar phenomena as above, in that only hydrophobic components associate with the micelle (Table 2 and SI for extended Table S2). Surfactant 6 aggregates above the CMC (Table 2, entries A−B), azide 5 is fully associated with surfactant 6 above the CMC (entries C−D), and alkyne 4 does not associate detectibly with 6 The diffusion coefficients for all reaction components present in the system during the synthesis of 3. b D values reported in 10 −10 m 2 s −1 . c TMS was added as a control. d ND indicates that the diffusion coefficient could not be extracted because of the limited solubility of the compound in water. e A value below that of 3 and TMS indicates a strong association with the micelle, where the measured value of 3 reflects the presence of monomeric and aggregated surfactant.  (Table S2) and does not associate with the micelles (entries H−I) while the hydrophobic Cu-C complex binds strongly to the micelles (entries J−K). Deuterium Exchange Measurements. The rate of deuterium exchange for alkynes 2 and 4 to yield 2-d and 4-d by copper in D 2 O was measured to probe how readily alkynes coordinate to Cu-L and subsequently interact with water (Scheme 1 and Table 3). These experiments may be relevant to the first step of the CuAAC mechanism where coordination of copper to alkyne is believed to occur. With compound 2 in the absence of copper, no deuterium exchange was observed (entry 1). For 2 with both Cu-B and Cu-C, the rate of H-to D-exchange was slow giving 58% and 32% conversion respectively within a day (entries 2 and 3). It may be that exchange is faster with hydrophilic Cu-B because the complex can more easily interact with water. When H-to D-exchange experiments with 2 were seeded with 3, full conversion to 2-d was observed within 5 h (entries 4 and 5).
With hydrophilic alkyne 4, H-to D-exchange also did not occur in the absence of copper (entry 6). Using the combination of 4 and ligand B exchange was so fast that full conversion is observed in less than a minute (entry 7). However, C required 3 h for full D-exchange (entry 8). The hydrophilic Cu-B complex and hydrophilic alkyne combination results in much faster exchange than phase separated components. Surprisingly, seeding Cu-C mediated exchange of 4 to 4-d with product 6 did not significantly increase the reaction rate (entry 9). This might indicate that these CuAAC reactions benefit more from solubilization of hydrophobic azide 5 than from bringing hydrophobic Cu-C into the aqueous phase or interface.
Overall copper is required for the deuterium exchange, and it appears that phase separation between the Cu-L-complex and alkyne impacts the rate of exchange. Addition of product leads to acceleration of exchange for alkyne 2 while 4 is minimally impacted. This may indicate that the first step in the CuAAC reaction here is accelerated by product formation for alkyne 2 while this is not the case for alkyne 4.
Mechanistic Rationalization. The mechanism at play in these systems is reminiscent of phase transfer catalysis, where two substances located in different immiscible phases are able to react via a catalyst. This is possible because the catalyst is The diffusion coefficients for reaction components present in the system during the synthesis of 6. b D values reported in 10 −10 m 2 s −1 . c There may be an association between monomers of 6 and [Cu-B] leading to the observed D of 2.3 and not the expected D of ∼3.3. The D of TMS suggests that no surfactant aggregation/encapsulation is taking place. 6 with ligand B (and no copper) diffuses at the expected rate (see SI Table S2).

Scheme 1. Deuterium Exchange Control Experiment for Alkyne 2 and 4
Standard reaction conditions were applied in D 2 O in the absence of azides to prevent CuAAC. b Different rates of deuterium exchange were observed when ligand B or C was applied. c Seeding the system with 3 (for 2) or 6 (for 4) increased the rate of D-exchange in all cases. d Reactions monitored and conversion determined by 1 H NMR spectroscopy. e Control reaction in the absence of copper.

The Journal of Organic Chemistry
Article able to transfer a substrate to its opposing phase. Although phase-transfer catalysis is well developed, some mechanistic aspects still remain unclear, 45−47 at least partially because of problems arising from monitoring biphasic systems and the many parameters involved. Mechanistic studies have been performed on micellar autocatalysis; 22,48,49 however, these models never involved a secondary catalyst such as reported here.
Our rationalization for the kinetics observed in the original system is shown in Figure 5. We think of the system as consisting of three phases: an organic phase (alkyne), an aqueous phase, and an interfacial phase where the two mix well. The micelles are seen as an extension of the interface. The mechanism of CuAAC is complex, but there are likely two key steps that involve interaction of components that would be phase separated here, namely the interaction of copper with alkyne and subsequently the interaction of this complex with azide. 50−57 Although initial coordination of azide to copper might occur in a phase separated case, we assume the generally accepted sequence of events occurs here and believe that if the sequence were inverted the overall conclusions drawn from the model (vide infra) would not change. We assume other steps, such as formation of the copper-ligand complex and demetalation of the triazole, occur rapidly relative to the phase separated steps and are therefore not incorporated into the model nor is the mono-or dinuclear nature of the catalyst complex.
The deuterium exchange experiments have given some insight into the rates of interaction of alkyne with copper, and from this, the rates of the second step can be inferred. The model assumes that the phase behavior of complexes and intermediates are similar to those of the individual reaction components observed during the DOSY experiments.
In steps where the hydrophilic reagent is present at high concentrations in the aqueous phase, we expect efficient phase transfer catalysis by the micelle and these steps to be promoted by product. For steps where the hydrophilic reagent is present at low concentrations, we expect less efficient phase transfer catalysis, and therefore they are weakly promoted by product (weakly autocatalytic). Colocalization, solubilization of two hydrophobic reagents within the micelle will lead to a strongly promoted step. Whenever it is stated that a step is faster, promoted, or more efficient, it is important to note that this is due to a change in the rate of mass transport, not a change in catalytic activity of copper.
(Step 1) Hydrophobic alkyne and Cu-L interact. Since the [C-Cu(I)]-complex and alkyne are both strongly hydrophobic, they can react freely and quickly in the organic phase ( Figure  5, step 1a and 1b). This step will therefore not benefit from any product formation. The [B-Cu(I)]-complex on the other hand is mainly present in the aqueous phase and has to encounter the alkyne at the interface (step 1c). Autocatalysis can accelerate this step, but it will be limited by Cu-B complex concentration leading to a weak promotion of this step. ( Step 2) Aqueous azide coordinates to Cu-L-alkyne complex. The azide is hydrophilic, and with both B and C this should be a biphasic process ( Figure 5, step 2a and 2b) that is promoted by the presence of product. We assume that the Cu-B-alkyne complex will be in the interface and B is not sufficiently polar to solubilize it completely into the aqueous phase (slow H-to-D exchange is observed here, Table 3, entry 2), so that step 2 with both ligands will occur at the interface and behave similarly. Subsequently product 3 accumulates at the interface, while Cu-L re-enters the catalytic cycle. Once the CMC is reached and micelles form, the product will start to catalyze any biphasic steps.
This qualitatively accounts for the observed differences between B and C ( Figure 2). For B, step 1c is slow since it is biphasic and we observe a significant lag-period (up to about 7 h). For C, step 1a/b will be fast as it is a single-phase reaction and we observe a short lag period of 30 min.
Step 2 presumably has significant biphasic character in either case.
The model also qualitatively accounts for differences observed between seeded and nonseeded reactions. With B seeding with product all but eliminates the lag period and doubles the maximum rate, likely by facilitating the mixing of hydrophilic catalyst and hydrophobic azide in step 1. With ligand C there is not a significant difference in maximum rate between seeded and nonseeded reactions (see Tables S3 and  S4 for extracted rates). Overall it appears that the ligand polarity can effectively make step 1 fast or slow: with ligand C, step 1 is rapid, while, with ligand B, step 1 becomes ratelimiting. The Journal of Organic Chemistry Article A modified model ( Figure 6) is required to explain the kinetics of the inverse system which considers the phase behavior of the different coupling partners.
(Step 1) Aqueous alkyne and Cu-L interact. Association of the [C-Cu(I)]-complex with alkyne is initially slow ( Figure 6, step 1a) but only weakly promoted once micelles of 6 are formed. Although the complex can associate with the micelle (Table 2, entries J and K), 4 does not readily associate with micelles (Table 2, entries E, F, and G). And as judged by H-to D-exchange experiments ( Table 3, entries 8 and 9) micelles do not have very much of an effect on the way the complex and 4 encounter each other.
Step 1b for the [B-Cu(I)]-complex is fast, as alkyne and Cu-B are both soluble in water (Table 3,  entry 7). ( Step 2) Hydrophobic azide coordinates to Cu-L-alkyne complex. Cu-C-alkyne is likely hydrophobic enough to colocalize with 5 inside the micelle. This will strongly promote step 2a upon formation of micelles. For B step 2b will be weakly promoted: the organic azide needs to encounter Cu-Balkyne complex assembled from two hydrophilic components. While micelles should increase the concentration of azide that can encounter Cu-B-alkyne, the concentration of the latter in micelles is probably low. 6 will accumulate at the interface and, once the CMC is reached, catalyze any biphasic steps.
Overall the kinetics of this system are slower, with both ligands showing pronounced lag periods ( Figure 4B). Water/ tert-butanol experiments suggest this is not due to inherent substrate reactivity ( Figures S43 and S45). This may be due to rate limiting step 2 where the Cu-L-alkyne complex has to interact with the hydrophobic substrate. Since this complex will be present in the low concentrations, this could be the cause for the slower kinetics observed in the absence of micelles.
When seeded ( Figure 4C) reactions with C showed a significant rate increase, but with B the reaction is only minimally promoted by product. Co-localization of the Cu-Calkyne complex and 5 in the micelles should lead to higher local concentrations and faster reactions for C, while the strongly hydrophilic Cu-B-alkyne complex does not as easily interact with the micelles. Overall it appears that step 2 is rate limiting here and this step can be strongly promoted using product 6 with ligand C but not ligand B.
The four different systems investigated and the degree to which surfactant products should promote each step are summarized in Figure 7. Steps where both reagents are present in the same phase will be fast, while steps where substrates are phase separated are slow. Micelles will promote phase separated steps, but if this involves a low concentration of aqueous substrate the step will only be weakly promoted.  Overall 'systems 2 and 3' (Figure 7) should see the more dramatic autocatalytic effects since the rate limiting steps in these two systems are promoted by product formation. A new system was developed to investigate if biphasic autocatalytic CuAAC reactions are reasonably general and see if the model could be used to predict a reaction with an autocatalytic effect. Alkyne 7 containing a zwitterionic phosphocholine headgroup ( Figure 8A) was coupled to azide 5 using ligand C to yield surfactant 8 (conditions of system 3, Figure 7). A CuAAC reaction with this new polar alkyne should show efficient autocatalysis since according to the model the main prerequisite for this effect is phase separation. This system shows a lag period, and seeding with product leads to a disappearance of the lag period and a pronounced rate acceleration ( Figure 8B). 58 These results support the idea that the model can be used to predict an autocatalytic effect.

■ CONCLUSIONS
This study has shown that a secondary catalyst promoting CuAAC reactions can be used to tune the kinetics of a system that is autocatalytic as a whole by virtue of phase behavior. Coupling a secondary catalytic cycle to an autocatalytic cycle enables control of the reaction kinetics through rational variations in catalyst structure. Changing the hydrophilicity of the ligand, and therefore the location of the active catalyst, provides a handle to vary the rate limiting step of the reaction. Our observations can be rationalized in terms of the phase behavior of the reaction components. A predictive model was obtained that has shown that a hydrophobic ligand and azide combined with a hydrophilic alkyne can be used to obtain a pronounced autocatalytic effect. We anticipate that these studies might be useful in understanding how to gain control over the spatial organization of compartmentalized reactions.

■ EXPERIMENTAL SECTION
General Information. Procedures using oxygen-and/or moisture-sensitive materials were performed with anhydrous solvents under an atmosphere of anhydrous argon in flame-dried flasks, using standard Schlenk techniques. Analytical TLC was performed on precoated aluminum-backed plates (Silica Gel 60 F254; Merck),and visualized using aqueous ceric ammonium molybdate (CAM), aqueous basic potassium permanganate, or ninhydrin stains. Flash column chromatography was carried out using Merck Geduran Si 60 (40−63 μm) silica gel. The compound was loaded on to the columns with Chemtube Hydromatrix from Agilent Technologies. Pressure was applied at the column head via a flow of nitrogen with the solvent system used in parentheses.
Cooling of reaction mixtures to 0°C was achieved using an ice− water bath. Cooling to −10°C was achieved using a salt−ice bath. Cooling to −78°C was achieved using a dry ice−acetone bath.
Chemicals. All chemicals were purchased from Sigma-Aldrich or Fluorochem Scientific and used without further purification. Dry CHCl 3 , THF, CH 2 Cl 2 , Et 2 O, toluene, benzene, hexane, pentane, DMF, and acetonitrile were collected fresh from an mBraun SPS-5 solvent purification system having been passed through anhydrous alumina columns. All other solvents were used as purchased from Sigma-Aldrich, Honeywell, or Fisher Scientific.
Equipment. All NMR spectra were recorded at room temperature. 1 H NMR and 13 C NMR spectra were recorded using Bruker AVIII HD 400 (400/101 MHz) and AVIII HD 500 (500/126 MHz) spectrometers. Chemical shifts are reported in ppm from the residual solvent peak. Chemical shifts (δ) are given in ppm, and coupling constants (J) are quoted in hertz (Hz). Resonances are described as s (singlet), d (doublet), t (triplet), q (quartet), and m (multiplet). Assignments were made with the assistance of 2D COSY and HSQC NMR experiments.
DOSY NMR measurements were performed using a Bruker AVIII HD 500 equipped with a TFI probehead at 298 K using the 2D sequence for diffusion measurement using double stimulated echo for convection compensation and longitudinal eddy current delay, using bipolar gradient pulses for diffusion, and using three spoil gradients (Bruker terminology: dstebpgp35) pulse sequence. The samples were thoroughly mixed using a Vortex Genie 2 mixer (Scientific Industries) and were then clarified using a hand centrifuge (Hettich, model 1011) and then measured. Samples containing saturated alkyne consequently had a small layer of neat alkyne above the D 2 O layer; sufficient D 2 O was used to ensure that the alkyne layer was not detectible by the NMR probe. Experiments were performed in two stages: initially 1Dedited DOSY experiments were used to optimize the diffusion period to Δ = 100 ms. The 2D dstebpgp35 sequence was then used, based on the optimized Δf rom the previous procedure and with δ = 4 ms, with the gradient amplitude ranging from 2% to 85% with 16 points in between. Data were analyzed using the T 1 T 2 module in TOPSPIN 3.2, and plots were generated using the eddosy module.
High-resolution mass spectra (EI and ESI) were recorded using a Bruker MicroTOF spectrometer by the internal service at the University of Oxford. Low-resolution mass spectra were recorded using a Walters LCT premier XE.
Infrared measurements (thin film) were carried out using a Bruker Tensor 27 FTIR with internal calibration in the range 4000−600 cm −1 .
Optical rotations were recorded using a PerkinElmer 241 polarimeter at 25°C in a 10 cm cell in the stated solvent.
Fluorimetry was performed using Edinburgh Instruments Spectrofluorometer FS5 model with Fluoracle software. The slit width for both excitation and emission was set at 1 nm. DLS measurements were recorded using a Malvern Zetasizer Nano ZS DLS instrument and analyzed with Zetasizer software. All samples were prepared in ultrapure Milli-Q water and filtered through 0.2 μm PTFE filters before measuring. The unseeded reaction (red diamonds) has a lag period followed by an increased rate of product formation. When the system is seeded with product 8 (30 mM, gray triangles), the lag period is eliminated and a significant rate increase is observed. Points are the mean of three independent experiments, and the error bars are the standard deviation.  59 To a stirred suspension of CuI (0.02 equiv) in degassed CH 2 Cl 2 was added protected maltose (1 eq, 160 mM), DIPEA (0.04 equiv), AcOH (0.04 equiv) and hydrophobic coupling partner (1.4 equiv). The resulting solution was stirred for 18 h. The reaction mixture was concentrated in vacuo, and the crude was purified with flash column chromatography. The column was eluted with EtOAc:hexane (1:1) to yield the product.
General Procedure 2: Acetyl Deprotection. Synthesis according to Mahon et al. 60 To a stirred suspension of protected sugar (150 mM) in MeOH was added sodium methoxide (0.1 equiv). Upon dissolution of the solid and concurrent disappearance of protected sugar (TLC control) the solution was neutralized using Amberlyst 15 resin (H + form). The resin was filtered off and washed with MeOH and the filtrate was concentrated in vacuo. The residue was dried under high vacuum to give a deprotected sugar as a tacky, hygroscopic white foam.
General Procedure 3: Setup of Kinetic Experiments. A solution of hydrophilic substrate (0.408 mmol, 1 eq, 1.5 mL of 272 mM standard solution in D 2 O), CuSO 4 ·5H 2 O (6 mg, 0.024 mmol, 0.06 eq, 1 mL of 6 mg/mL standard solution in D 2 O) (and deprotected surfactant in the reported concentrations for the seeded reactions) were added to D 2 O (2 mL) to give a total volume of 4.5 mL in a round-bottom flask (25 mL) with a stirrer bar of an identical size (4.5 mm diameter and 12 mm length) for each experiment. Subsequently ligand (0.12 equiv), hydrophobic substrate (2 equiv) were added and the flask was capped with a septum and the solution was degassed by bubbling argon through it for 30 min. Afterward sodium ascorbate (16.2 mg, 0.082 mmol, 0.2 equiv) was added to initiate the reaction. The reaction mixture was stirred at 200 rpm under a continuous flow of argon. Samples for analysis during the kinetic experiments were prepared at regular intervals by diluting 0.1 mL of the reaction mixture in 0.4 mL of D 2 O and immediately taking a 1 H NMR measurement.
Note: These reactions are very sensitive to physical parameters, so changing the flask size or stirrer bar or adding compounds at different time points during the degassing process might affect the overall kinetics of the reaction.
Note 2: The deuterium exchange experiments were performed in the absence of azide 1 or 5. Alkyne 2 was monitored by extracting 0.1 mL of the reaction mixture into 0.4 mL of CDCl 3 and immediately taking a 1 H NMR measurement.

Notes
The authors declare no competing financial interest.

■ ACKNOWLEDGMENTS
The EPSRC (EP/M0025241/1) and the ERC (Consolidator Grant, Autocat, 681491) are gratefully acknowledged for financial support. We wish to thank Andrew Bissette for his insightful comments.