Controllable Vesicular Size and Shape in Polymerization-Induced Self-assembly Aided by Aromatic Interactions

The size and shape of polymeric vesicles have great impact on their physicochemical and biological properties. Polymerization-induced self-assembly (PISA) is an ecient method to fabricate vesicles. In most PISA-cases, the formation of vesicles is driven by the solvophobic interactions which are lack of versatility on nely structural regulation. Herein, controlling vesicular size and shape is realized in PISA aided by aromatic interactions. Aromatic interactions between the membrane-forming blocks contribute to the augments of membrane tension which lead to the formation of smaller vesicles (as small as 70 nm), but overly enhanced aromatic interactions result in vesicle fusion rather than size decreasing. When the membrane tension is dominated by aromatic interactions and meanwhile high enough to overcome the energetic barriers of fusion, the aromatic interactions drive vesicle fusion in a directional manner to form tubular structures. The precise regulation of vesicular size and shape in PISA would pave the way to fabricate vesicles for a series of size/shape-dependent applications.

cases. The weakly intermolecular interactions between the solvophobic blocks lead to enough chain mobility for morphological transition, but sacri ce the membrane tension of the resultant vesicles, which makes forming and maintaining the high surface curvature of small vesicles very di cult. In addition to solvophobic effects, some other non-covalent interactions can also drive/in uence self-assembly of block copolymers. 46-48 PISA aided by hydrogen bonding, 49 static electricity, 50 host-guest complexation, 51 crystallization, 52 and liquid crystal ordering, 53 obviously improve the accessibility of anisotropic nanoobjects. Applying cooperativity of multiple non-covalent interactions generates abundant structures and even functional optimization of the cells and organelles in nature. [12][13][14] Introducing multiple driving forces in PISA syntheses would hopefully facilitate the possibility in complex size/shape regulation of arti cial assemblies.
Herein, synergetic effects of solvophobic and aromatic interactions (or called π-π interactions) on the vesicular size and shape were investigated in PISA system. Vesicles of about 77 ± 26 nm (numberaverage diameter by CMA was selected as the monomer to fabricate the solvophobic block due to the aromaticity of the coumarin units. Strong aromatic stacking and solvophobicity of the membrane-forming poly(7-(2methacryloyloxyethoxy)-4-methylcoumarin) (PCMA) blocks in the vesicles facilitate the formation of sub-100 nm vesicles. The aromatic interactions of the membrane-forming blocks were adjusted by copolymerization of CMA with nonaromatic monomer 2-(diisopropylamino)ethyl methacrylate (DIPEMA), or weakly aromatic monomer benzyl methacrylate (BzMA), or strongly aromatic monomer 2-(methacryloyloxy) ethyl anthracene-9-carboxylate (ACMAE). Variations in aromatic interactions in the vesicular membrane contributed to the discrepancy in membrane-tension which leaded to the formation of vesicles with controllable size. Generally, strong aromatic interactions result in the formation of smaller vesicles, but overly enhanced aromatic interactions lead to vesicle fusion rather than size decreasing of the vesicles. The aromatic interactions drive vesicle fusion in a directional manner to form tubular structures. Our results indicate that su cient frequency of inelastic collision, enough membrane tension to overcome the energetic barriers of vesicle fusion, and aromatic interactions rather than the solvophobic interactions dominated the membrane tension to provide directionality of vesicle fusion, play signi cant roles in the formation of tubular structures.

Results And Discussion
2.1 PEO 45 -CPADB mediated RAFT dispersion polymerization of CMA.
The monomer CMA was successfully synthesized using the reagents 7-hydroxy-4-methylcoumarin, 2bromoethanol, and methacryloyl chloride, the structure of CMA and the intermediate were con rmed by 1 H and 13 C NMR spectra ( Figure S3, S4 and S5). Afterwards, PEO 45 Figure S7). Homogeneous polymerization conducted in the rst stage of polymerization within 1 h, the monomer was consumed slowly and the formed copolymer might be molecularly dissolved in the ethanol/water mixture (7/3, w/w). Spherical micelles were formed at 1.25 h of polymerization with monomer conversion getting 18 % at this time ( Figure 1A, S6 and S7), then heterogeneous polymerization was conducted. The apparent enhancement of polymerization rate after phase separation might be induced by the relatively high local monomer concentration in the nanoparticles. 54,55 The GPC traces of PEO 45 -b-PCMA obtained at different polymerization time are summarized in Figure S8, which shows narrow molecular weight distribution (M w /M n ≤ 1.10) for all samples.
The morphologies of PEO 45 -b-PCMA assemblies obtained at different polymerization time were traced by TEM (Figure 1), which go through an evolution from spherical micelles, to nanoworms, to lamellae and semi-closed  Figure   2A and S6). Besides, uorescence emission spectrum can also be used to examine the aromatic interactions because the emission maximum is also in uenced by the strength of the aromatic interactions. 19 When the concentration of CMA increases, the uorescence emission spectra show gradual red shift as a result of the formation of a low-energy excimer state due to stronger π-π electronic overlap between coumarin units ( Figure S11). Compared to the uorescence emission spectrum of CMA monomer in CH 2 Cl 2 ( Figure 2B), the uorescence emission spectra of diblock copolymer PEO  in CH 2 Cl 2 also show a red shift as a result of the stronger aromatic interactions between coumarin units ( Figure 2B), which is consistent with the results of 1 H NMR spectra (Figure 2A and S6). Greater red shift in the uorescence emission spectrum of PEO 45 -b-PCMA 80 vesicles was observed ( Figure 2B), indicating a more pronounced aromatic interactions in the assemblies than that in CH 2 Cl 2 . This phenomenon is reasonable, because inter-chain interactions between coumarin units are greatly enhanced in vesicles, while that between coumarin units in CH 2 Cl 2 is greatly reduced due to the molecularly dissolving state of the PEO 45 -b-PCMA 80 block copolymer. The above results demonstrate that there are strong aromatic interactions in the membrane forming PCMA blocks of the PEO 45 -b-PCMA 80 vesicles, and the strength of the aromatic interactions can be characterized by 1 H NMR spectrum and uorescence emission spectrum. In general, stronger aromatic interactions result in lower chemical shifts of the aromatic coumarin protons as well as red shifts of the uorescent emission maximum of coumarin.

Controlling vesicular size
Combining the fact that membrane tension has great in uence on the vesicular size and that aromatic interactions in vesicular membrane signi cantly impact the membrane tension, we proposed that adjustment of aromatic interactions may achieve the controllability of vesicular size. In order to adjust the aromatic interactions in vesicular membrane, RAFT dispersion copolymerization of CMA and nonaromatic monomer DIPEMA was performed ( Figure 3A, Table S1 and S2). Insertion of DIPEMA units into the solvophobic PCMA block is expected to reduce the aromatic interactions between CMA units to some extent, and the overall aromatic interactions between the solvophobic blocks decrease accordingly.
Fixing DP of solvophobic block P(CMA-co-DIPEMA) at 80, a series of vesicles were prepared with varying feed molar ratio of DIPEMA/(CMA+DIPEMA) from 0 to 0.40 ( Figure 3B 1  The conversions of CMA and comonomer DIPEMA by 1 H NMR spectra of all the samples were above 82% as shown in Table S2, and the molar contents of comonomer DIPEMA (x) were roughly identical to the targeted ones. The corresponding diblock terpolymers were denoted as PEO 45 80 , wherein x represents molar content of DIPEMA in the solvophobic block, and x = 0.05, 0.10, 0.15, 0.25, and 0.40, respectively. 1 H NMR spectra were used to study the variations of aromatic interaction strength for a series of PEO 45 -b-P(CMA 1-x -co-DIPEMA x ) 80 diblock terpolymers by examining the changes in chemical shift of coumarin protons ( Figure 3C and S12). Insertion of DIPEMA units in solvophobic PCMA block resulted in deshielding of the protons on the coumarin ring to some extent. The chemical shift of coumarin protons H b and H d gradually move to higher δ values with increasing molar content of DIPEMA units ( Figure 3C and S12), suggesting the aromatic interactions are gradually weakened. Fluorescence spectra of the vesicles in ethanol/water (7/3, w/w) were used to study the effect of DIPEMA content on the variation of aromatic interactions of the solvophobic blocks P(CMA 1-x -co-DIPEMA x ) 80 in the vesicles ( Figure 3D). Gradual blue shifts of the emission maximum from 401 nm (x = 0) to 390 nm (x = 0.40) for PEO 45 -b-P(CMA 1-x -co-DIPEMA x ) 80 vesicles reveal that the aromatic interactions are more disturbed at higher content of DIPEMA. Ultra-sensitive differential scanning calorimetry (US-DSC) analysis was used to study the variations of the solvated T g (de ned as T sg herein) of P(CMA 1-x -co-DIPEMA x ) 80 in the polymerization medium ethanol/water (7/3, w/w). The dispersions of PEO 45 -b-P(CMA 1-x -co-DIPEMA x ) 80 vesicles (50 mg/g) in ethanol/water (7/3, w/w) were directly used for US-DSC analysis. PEO is soluble in ethanol/water and shows no signal, so US-DSC curves only exhibit the T sg of the membrane-forming block P(CMA 1-x -co-DIPEMA x ) 80 . The same DP (80) of P(CMA 1-x -co-DIPEMA x ) 80 in these samples eliminates the impact of the molecular weight on the T sg value. Figure 3E shows that the T sg values of P(CMA 1-x -co-DIPEMA x ) 80 at x = 0, x = 0.10, and x = 0.25 were determined to be 62.3, 58.1 and 49.0 o C, respectively. The decreased T sg with increasing of the x values indicates that the membrane-forming blocks P(CMA 1-x -co-DIPEMA x ) 80 are much easier to be plasticized by the solvent, which illustrates that the exibility of the membrane-forming blocks increase in these conditions. As the insertion of DIPEMA units in PCMA blocks greatly weakened the aromatic interactions in vesicular membrane, the stacking of polymer chains is supposed to be less compact in this situation which accordingly increases the mobility of membrane-forming blocks. In addition to the decrease in aromatic interactions strength, reduction in solvophobicity of P(CMA 1-x -co-DIPEMA x ) 80 blocks may partly account for the rising exibility. The  RAFT dispersion copolymerization of CMA and the weakly aromatic monomer BzMA was also carried out to fabricate a series of PEO 45 -b-P(CMA 1-x -co-BzMA x ) 80 vesicles ( Figure 4A, 4B 1 -B 5 , Table S3 and S4). The conversions of CMA and comonomer BzMA of all the samples were above 82% (Table S4) Table S4). By contrasting vesicles prepared from copolymerization of CMA with BzMA and DIPEMA, it can be identi ed that PEO 45 -b-P(CMA 1-x -co-DIPEMA x ) 80 vesicles generally have larger size than that of the PEO 45 -b-P(CMA 1-x -co-BzMA x ) 80 vesicles at the same x value (Table S2 and S4).
Concurrent reduction of the aromatic interactions and solvophobicity was induced by copolymerization of CMA with DIPEMA, while only the solvophobicity was reduced in copolymerization of CMA with BzMA, the former generated lower membrane tension to facilitate the formation of larger vesicles.
The above results indicate that controlling vesicular size can be realized by RAFT dispersion copolymerization of CMA and DIPEMA or BzMA to adjust the interactions between the membrane-forming blocks, and the diameter of the resultant vesicles increased due to the reducing of membrane tension. In order to demonstrate whether controlling vesicular size with a decreasing trend can be realized by increasing the membrane tension, RAFT dispersion copolymerization of CMA with a strongly aromatic monomer ACMAE ( Figure S16 and 17) were carried out to fabricate a series of PEO 45 -b-P(CMA 1-x -co-ACMAE x ) 80 vesicles ( Figure 5A, 5B 1 -B 5 , Table S5 and S6). The conversions of CMA and comonomer ACMAE of all the samples were above 93 % as shown in Table S6 and the molar contents of ACMAE units (x) were roughly identical to the targeted ones. Although the interplanar distance between the coumarin units was enlarged as a result of the insertion of ACMAE in solvophobic P(CMA 1-x -co-ACMAE x ) 80 blocks, anthracene rings of ACMAE, which took the  Figure 5C and S18), which indicate that the aromatic interactions were enhanced upon insertion of ACMAE in the solvophobic block. Red shifts of the emission maximum in uorescence spectra of the PEO 45 -b-P(CMA 1-x -co-ACMAE x ) 80 vesicles ( Figure 5D) are observed as the molar content of ACMAE increase. The great red shift of the emission maximum in uorescence spectra from 401 nm at x = 0 to 450 nm at x = 0.05 is mainly due to the uorescence resonance energy transfer (FRET) effect, 58 resulting from the signi cant spectral overlap between the emission spectrum of CMA and the absorption spectrum of ACMAE ( Figure S19) as well as the close distance between the two uorophores in vesicular membrane. However, the red shift of the emission maximum in uorescence spectra from 450 nm at x = 0.05 to 464 nm at x = 0.40 ( Figure 5D) further indicated that the aromatic interactions in the solvophobic blocks were gradually enhanced as the molar content of ACMAE increase. Moreover, US-DSC results ( Figure 5E) show that the T sg values of P(CMA 1-x -co-ACMAE x ) 80 Figure 5F, S20, and Table S6), respectively. The possible reason is that the membrane tension at this condition is too high to maintain the thermodynamic stabilization of the vesicles, which may result in part of vesicle fusion rather than size decreasing of the vesicles. Continuous increasing the molar content of strongly aromatic ACMAE units in the membraneforming blocks P(CMA 1-x -co-ACMAE x ) 80 to x = 0.6 results in further increase of the aromatic interactions as demonstrated by 1 H NMR spectra ( Figure S18) and uorescence spectra ( Figure S21), and DLS results reveal further increases in number-/volume-/intensity-average diameter at this condition ( Figure 5F, S20, and Table S6). Obvious vesicle fusion to form necklace-like structures can be observed from the TEM image ( Figure S22), which indicates that the aromatic interactions drive vesicle fusion in a directional manner. However, the fusion of the vesicles is incomplete and tubular structures were not formed at this condition ( Figure S22), which may be due to the insu cient enhancement of membrane tension by only introducing strongly aromatic units (ACMAE).
According to the above results, both the membrane tension and aromatic interactions can be weakened by insertion of DIPEMA units into PCMA blocks, weakened membrane tension but without signi cant in uence on the aromatic interactions is induced by insertion of BzMA units into PCMA blocks, both the membrane tension and aromatic interactions can be enhanced by insertion of ACMAE units into PCMA blocks. The weakened membrane tension can not maintain the high curvature of the small vesicles and then lead to the formation of bigger vesicles. The enhanced membrane tension results in size decreasing of the resultant vesicles, but excessive enhancement of membrane tension leads to vesicle fusion rather than size decreasing. Moreover, the aromatic interactions seem to drive the vesicle fusion in a directional manner ( Figure S22).

Controlling the directionality of vesicle fusion.
Alternatively, increasing the DP of the solvophobic blocks is a more direct strategy to enhance the membrane tension of the vesicles. 24,43 When the membrane tension is high enough to overcome the energetic barriers of vesicle fusion, vesicle fusion occurs to reduce the total surface area of the vesicles and simultaneously releases part of the membrane tension to minimize the total free energy in the system. Besides su cient membrane tension, enough frequency of inelastic collision is also very important to induce vesicle fusion. 43 RAFT dispersion polymerizations of CMA with varying DP PCMA of 100, 120, 150 and 200 were carried out at 5% and 10 % solid  Table S7). The formation of anisotropic structures by vesicle fusion was not observed at 5 % solid content from DP PCMA = 100 to DP PCMA = 200 ( Figure 6A 1 -A 4 ). However, at 10 % solid content, nanotubes mixed with vesicles were formed at DP PCMA = 100 ( Figure 6B 1 ), and most of the vesicles transformed to nanotubes at DP PCMA = 120 ( Figure 6B 2 ). What is surprising, the proportion of the nanotubes in the mixture of vesicles and nanotubes obviously decreased at DP PCMA of 150 and 200 ( Figure 6B 3 and 6B 4 ). The width of the nanotubes gradually decreased from DP PCMA 120 to 200 ( Figure   S24-S26), which illustrated the membrane tension was still obviously enhanced during the growth of DP PCMA from 120 to 200. But why the enhancement of membrane tension did not induce the continuous fusion of the vesicles? Although the solid content kept at 10 % for the samples of DP PCMA = 100, 120, 150 and 200, the concentration of the polymer chains obviously decreased in turn for these samples ( Figure  6C and Table S7). In this regard, xing the chain concentration during the adjustment of DP PCMA might be a better strategy to provide high frequency of inelastic collision than xing the solid content. Then the samples of DP PCMA = 150 and 200 with a xed chain concentration at 3.01 × 10 -3 mmol/g (the chain concentration is equal to the samples of DP PCMA = 120 in Figure 6B 2 ) were prepared ( Figure 6D 1 , 6D 2 , S27, S28, and Table S7) to demonstrate the above speculation, and nanotubes with negligible residual of vesicles were formed at these conditions. The above results indicate that su cient membrane tension and high frequency of inelastic collision are the prerequisites for vesicles fusion.
In order to demonstrate the importance of aromatic interactions in providing directionality of vesicle fusion to form the tubular structures, RAFT dispersion The above results indicate that enough membrane tension to overcome the energetic barriers of vesicles fusion, aromatic interactions rather than the solvophobic interactions dominated the membrane tension, and su cient frequency of inelastic collision, are three prerequisites to drive directional fusion of vesicles to form tubular structures. In order to further demonstrate the above conclusions, RAFT dispersion copolymerizations of CMA and DIPEMA/BzMA/ACMAE with the molar content of CMA remaining at 60 % and the molar content of the comonomer (DIPEMA or BzMA or ACMAE) keeping at 40 % in the solvophobic blocks, were also carried out ( Figure S35-S41, and Table S9). Increasing the DP of the solvophobic blocks to induce vesicle fusion, compound vesicles of PEO 45

Conclusions
In summary, controlling vesicular size to the minimum of about 70 nm and directionality of vesicle fusion to form nanotubes is realized in PISA aided by aromatic interactions. The strong membrane tension generated by the great aromatic and solvophobic interactions of the membrane-forming block is evidenced to be a major factor for the formation of such small vesicles, but overly enhanced membrane tension results in vesicle fusion rather than size decreasing. The aromatic interactions play a crucial role in driving the vesicle fusion in a directional manner to form tubular structures. There are three prerequisites for the formation of tubular vesicles: (1) su cient frequency of inelastic collision to induce vesicles fusion, (2) enough membrane tension to overcome the energetic barriers of vesicles fusion, (3) aromatic interactions rather than solvophobic interactions dominated the membrane tension to drive vesicle fusion in a directional manner. In consideration of the aromaticity of many drug molecules and uorophores, the reported approach herein with the distinct mechanism in precise regulation of vesicular size and shape, and the intrinsic scalability of PISA, paves the way to fabricate vesicles for a series of size/shape-dependent applications including drug delivery and biological in vivo imaging.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. supportinginformation.docx