Construction of fluorescent hyperbranched polysiloxane‐based clusteroluminogens with enhanced quantum yield and efficient cellular lighting

Owing to its designability and intrinsic fluorescence, non‐conjugated hyperbranched polysiloxane (HBPSi) has attracted widespread attention in biological filed, while it is still severely restricted by low fluorescence efficiency. So, in this paper, we introduced disulfide into HBPSi improving their luminescence properties and synthesized different molecular weight HBPSi (P1, P2, and P3). Surprisingly, P1 exhibited ultrahigh quantum yield up to 47.81%. Meanwhile, experiments applied with theoretical calculations were employed to explore the fluorescence mechanism, which is attributed to efficient restricting of non‐radiative decay by clusteroluminogens formed with the cooperation of hyperbranched structure and double hydrogen bonding. In addition, the biocompatibility of P1 was verified by co‐culture with MC3T3‐E1 and P1 lighted up mouse fibroblast cells without fluorescent dyes. This work designed a novel fluorescent polymer with ultrahigh fluorescence quantum yield and cell imaging ability, which is promising in visualization diagnosis and treatment of tumor.


INTRODUCTION
In real cases, fluorescent dyes for cell imaging are often utilized in the aggregated states and tend to reduce fluorescence intensity even in aggregation-caused quenching (ACQ) [1] when placed for a period of time; so, the optical properties of traditional fluorescent dyes including sensitivity, stability, and effectiveness are seriously affected. [2,3] Therefore, creating a fluorescent material with satisfactory optical properties and excellent bioactivity is crucial. Since the debut of the aggregation-induced emission (AIE) concept by Tang et al., [4] it has set off an upsurge in AIE research around the world. [5] AIE materials do not emit light or emit low intensity at low concentrations, while the intensity of luminescence is significantly enhanced when the concentration is increased, which overcomes the ACQ of traditional luminescent materials. Looking at the current AIE polymer materials, they can be roughly classified into two types: conjugated AIE polymers containing large π-electron conjugated systems and non-conjugated AIE polymers containing only electron-rich heteroatom systems. [6,7] Benefiting from its excellent properties such as bright emission, anti-photobleaching, and stable colloidal stability, conjugated AIE polymers are broadly used for in vitro cell tracing, drug release visualizing, and one-photon or two-photon vascular imaging, but the potential biological toxicity from the benzene ring limits their application. [3,8,9] In comparison with conjugated AIE polymers, non-conjugated AIE polymers are similar in chemical structure and composition to biopolymers with potential biocompatibility and environmental friendliness. [10,11] Recent years have witnessed rapid development of AIEgens including the continuous abundance of species and the wide use in different fields, along with the development of many approaches in fabricating AIE polymers. [12] Non-conjugated hyperbranched polysiloxane (HBPSi), an important branch of AIE polymers, combines the advantages of hyperbranched structure and polysiloxane [13][14][15] and the research focus is currently owing to its good designability. [16] Hyperbranched structure is conducive to molecular aggregation to increase fluorescence intensity. [17] The bond angle of Si-O-C (120 • ) in siloxane structure is smaller than that in the traditional bond angle of Si-O-Si (140 • -180 • ) and has good rigidity, which is conducive to the formation of rigid and flexible polymer structure. In addition, the extremely flexible Si-O endows siloxane-like double bond characteristics, [18] which facilitates in building larger electron delocalization system of the molecule as a whole and improves the luminescence performance. [10] Even more fascinating, HBPSi is a class of luminescent materials and easy functional modification. In previous work, some researches have studied the relationship between functional groups, backbone, main chain, and the luminescence properties, and found that the introduction of amine group, epoxy group, hydroxyl group, carbonyl group, vinyl group, and hyperbranched structures can significantly improve the fluorescent performance of HBPSi and endow HBPSi with functional applications [19] and achieved cell imaging and controlled drug release by grafting cyclodextrin, glutamic acid, and oleic acid. [20][21][22] However, HBPSi of biological safety has yet to be verified, and most of the currently synthesized HBPSi has common drawbacks such as low fluorescence quantum yield and weak anti-interference, which restricts its potential application in cancer detection and other fields.
In view of these problems, disulfide bond containing electricity-rich atoms was introduced to HBPSi in this article. On the one hand, the electron-rich sulfur atom is beneficial in forming intermolecular and intramolecular space electronic communication through delocalized electrons of sulfur atom with lone pair of electrons; on the other hand, the dipole moment of the disulfide bond can increase the distortion of the chain segment, which is conducive to molecular aggregation and in line with the design strategy of restricting intramolecular motion to improve quantum yield. [23] Based on the design, it is expected to solve the thorny problem of low quantum yield and weak anti-interference of HBPSi, and greatly broaden its application prospect in biological imaging. Moreover, Si element is a class of bioactive elements that have been shown to effectively promote the proliferation and differentiation of osteoblasts. [24] Hence, HBPSi is expected to has biological safety.
Therefore, in this article, a novel HBPSi namely P1 with disulfide was prepared by the means of transesterification polycondensation to compare with polymers with different disulfide bond content (P2 and P3). Infrared, nuclear magnetic resonance (NMR), and gel permeation chromatography (GPC) were used to verify their structure, and their optical properties were studied by fluorescence spectrum and ultraviolet (UV)−vis absorption spectrum. Meanwhile, theoretical calculations, transmission electron microscopy (TEM), and dynamic light scattering (DLS) were employed to rationalize this luminescence mechanism. Furthermore, the biosafety of P1 was verified and the application of P1 in cell imaging was also successfully realized based on the high quantum yield. It is expected that P1 possessing high quantum yield and biosafety can promote the development of biological applications of HBPSi especially in cancer cell imaging and precision treatment.

Materials
Triethoxyvinylsilane (A-151) was purchased from Jingzhou Jianghan Fine Chemical Co., Ltd. 3,3ʹ-Dithiodipropionic acid (DTDP) was obtained from Shanghai Macklin Biochemical Co., Ltd. All the materials were acquired commercially and used without further purification.

Synthesis of HBPSi
HBPSi was synthesized by transesterification polycondensation reaction. In this reaction, the molar ratio of ethoxy group (-OCH 2 CH 3 ) to carboxyl group (-COOH) is 1:0.8, 1:0.6, and 1:0.4, respectively, and the synthesized products are denoted as P1, P2, and P3. Take the synthesis of P1 as an example. Firstly, at room temperature, a mixture of A-151 (74.1 g, 0.5 mol) and DTDP (84.11 g, 0.4 mol) was stirred in a 250 mL four-necked flask equipped with a mechanical stirrer, a thermograph, a N 2 gas inlet, and a condenser. Then, the mixed system was heated to 90 • C and this temperature was maintained unchanged until the material is dissolved. Thereafter, the reaction temperature was gradually raised to 150 • C, until the distillation temperature dropped above 45 • C and no more distillate came over. Soon after that, vacuum was used for 30 min to remove the residual by-products in the flask, and the crude product was poured into a vial and dialyzed in ethanol to remove low molecular weight products. Finally, the solution was rotary evaporated at 45 • C, and vacuum dried at 65 • C for 6 h to obtain a yellow viscous liquid namely the target polymer P1. The specific synthesis steps of P2 and P3 are the same as P1, except that the reaction raw material DTDP is changed to 0.3 mol (63.08 g) and 0.2 mol (42.07 g), respectively.

Measurements
Fourier transform infrared (FTIR) spectra of polymers and distillates were recorded on NICOLET 5700 FTIR spectrometer. 13 C NMR and 1 H NMR spectra were recorded on a Bruker Avance 400 MHz spectrometer in CDCl 3 or DMSOd 6 solvent. The term degree of branching (DB) was used as a quantitative measure of the branching perfectness for hyperbranched polymers. According to the different chemical environments of -CH 2 -CH 2 -adjacent to the disulfide bond, the peak areas of T, L, and D peaks correspond to the mole fractions of terminal unit, linear unit, and dendritic unit. Then, the DB can be expressed as DB = (T + D)/(T + L + D). Waters 1515 was used to obtain the molecular weights and molecular weight distributions taking tetrahydrofuran (THF) as the eluent. UV-vis absorption spectra for ethanol solutions of HBPSi were measured by a Shimadzu UV-2500 spectrophotometer. Fluorescent excitation/emission spectra for ethanol solutions of HBPSi were measured on a Hitachi F-4600 fluorescence spectrophotometer. Fluorescence lifetimes and absolute quantum yields for pure HBPSi were measured on a steady/transient-state fluorescence spectrometer coupled with an integrating sphere (FLS980, Edinburgh Instruments). DLS spectra were achieved on Malvern Zetasizer Nano ZS. TEM images were obtained via an FEI Tecnai G2 F20 microscope. The geometry optimization was performed using density functional theory (DFT) at the B3LYP-D3/6-31G(d) level for the ground states of molecule. [25]

Co-culture of MC3T3 cells and imaging
The mouse osteoblast cell (MC3T3-E1) was provided by the Shanghai Institute of Biochemistry and Cell Biology and cultured in a humidified atmosphere of 5% CO 2 at 37 • C with α-minimum essential medium (Hyclone) containing 10% fetal bovine serum and 1% penicillin/streptomycin (Solarbio). MC3TC-E1 cells were seeded on different concentration P1 coated cTi and the polished cTi samples (as the control group) at a density of about 4 × 10 4 cells per well. Adhesion and spreading of MC3T3-E1 cells were observed using scanning electron microscopy (SEM) and fluorescence microscopy for 1, 3, and 5 days. Cytoskeletal actin fibers and nuclei were dyed with fluorescein isothiocyanate (5 μg/mL, Solarbio) and Dulbecco's phosphate buffered saline (10 μg/mL, Solarbio), respectively. Cell Counting Kits (CCK-8, Dojindo) were used to quantify cell proliferation based on measurement of mitochondrial activity.

Synthesis and characterization of HBPSi
P1, P2, and P3 were synthesized by transesterification polycondensation reaction, as illustrated in Figure 1 (Scheme 1); to gain more insight, their structures were verified by FTIR, NMR spectra, and GPC.

1 H NMR and 13 C NMR study
As shown in Figure 1A, the proton peaks at 1.26 and 3.74 ppm correlate to the methyl group (Si−O−CH 2 −CH 3 ) and the methylene groups (Si−O−CH 2 −CH 3 ) of P1, respectively. The proton peaks of P1 are much weaker than those of A-151. Meanwhile, the proton peak at 12.38 ppm pertaining to the carboxy group (−COOH) of DTDP disappears in P1. This is consistent with the observation from the 13 C NMR spectrum ( Figure 1B), where the peaks related to primary carbon (CH 3 −CH 2 −O−Si) at 14.22 ppm and the methylene groups (CH 3 −CH 2 −O−Si) at 60.72 ppm of P1 are also significantly reduced in comparison with those of A-151. The change trend of NMR spectra of P2 and P3 is consistent with that of P1, as shown in Figures S1 and S2. According to Figure S4B, there is only a weak T peak at P2 and we got DB P1 = 0.76 and DB P2 = 0.12 by calculation. The DB of the HBPSi has been reported to vary from 26% to 74% over a wide range. P1 synthesized in this study has a higher DB, while DB of P2 is lower, but this may be due to the larger steric hindrance, which hinders the further condensation between silane monomer.

FTIR study
As shown in Figure 1C, compared with the peak of C=O of DTDP at 1698 cm −1 , the peaks of C=O of P1, P2, and P3 move to 1736 cm −1 , indicating the formation of ester groups.
As with A-151, the typical absorption bands for vinyl are also found in P1, P2, and P3 at 1600 cm −1 . The spectra of the distillate from the synthetic process of P1, P2, and P3 are almost identical to that of standard ethanol ( Figure S3), indicating that the distillate is ethanol.

GPC study
The molecular weights and related parameters of P1, P2, and P3 were tested by GPC using THF as the mobile phase. As shown in Figure S4A and Table S1, P1 has the largest average molecular weight and three peaks are 21,100, 11,400, and 700 Da, and molecular weights of P2 and P3 are 8000 and 700 Da. Moreover, the polydispersity index of P1, P2, and P3 are around 1, and polymer dispersion is relatively uniform. Above results indicated that transesterification polycondensation reaction was successfully completed and P1, P2, and P3 were successfully synthesized. And according to the calculated DB and molecular weight, P2 and P3 are also oligomer. However, since molecular weight of P3 is too small, its optical properties will not be studied in depth, and only be used to compare the effect of molecular weight on luminescence.

Optical properties of HBPSi
Synthetic P1 and P2 are light yellow viscous liquids under sunlight ( Figure S5), while P3 is a brown viscous liquid, and the viscosity increases with the increase in molecular weight from P3 to P1. Surprisingly, they emit bright blue-green  Figure S5). We also studied the steadystate excitation and emission spectra, fluorescence lifetime (τ), and absolute quantum yield of pure P1, P2, and P3. It is obvious that the excitation and emission wavelengths of pure P1, P2, and P3 are 376 and 450 nm, respectively ( Figure 2A). More surprisingly, Figure S6A shows that the quantum yield of P3, P2, and P1 increases from 35.68%, 40.57% to 47.81% with increase in molecular weight, to best of our knowledge, which is currently the highest among non-conjugated polysiloxanes. [17] Hence, its intrinsic mechanism should be profoundly studied and analyzed to provide theoretical guidance for the subsequent development of non-conjugated fluorescent polymers with high quantum yield. Generally, the fluorescence lifetimes of P3, P2, and P1 are shown in Figure S6B, which are 1.19, 6.97, and 6.40 ns, respectively. It is at the same level as the fluorescence lifetimes of non-conjugated fluorescent polymers reported so far. [19] To gain more insight, subsequently, we systematically studied the UV−vis absorption and fluorescence spectra of P1, P2, and P3 ethanol solutions. Figures 2B and S7A,B depict that the strong absorption peark1 of P1 red shifts from 205 to 228 nm with concentration increasing from 1 to 5 mg/mL, and the absorption of peark1 of P2 and P3 have position overlap to P1. This may be attributed to the n-δ* electronic transition of the disulfide bond. In essence, P3 to P1 is the increase of the number of disulfide bonds in the system, so that more lone pair electrons in the sulfur atom could be excited to transition. It is worth noting that there is another absorption peak at 246 nm, namely Peark2, and the peak intensity is almost negligible when the concentration is very low. With the increase in concentration, the peak intensity increases sharply, which may be attributed to the π-π*, n-π* electronic transitions from C=C, C=O, and Si−O. Photoluminescence spectra of P1 ( Figure 2D) shows that P1 exhibits an aggregation-enhanced emission (AEE) feature from 5 to 100 mg/mL at 450 nm (λ Ex = 376 nm) in the ethanol solution. Figure 2C shows the emission spectra of the 10 mg/mL P1 ethanol solution are obtained at different exciting wavelengths. P1 has no excitation-dependent red shift, in which emission spectra are concentrated in 450 nm with the gradual increase in the excitation wavelengths ranging from 330 to 410 nm, implying excellent stability. [26] By adding dilute hydrochloric acid or sodium hydroxide solution to adjust pH of P1 solution, the fluorescence of P1 at different pH were recorded. As shown in Figures 2E and S8, P1 solution presents obvious pH-dependent emission characteristics in the pH range of 12.0-3.0. Under the pH value of 12.0-6.0 or 3.0-6.0, the fluorescence intensity of the P1 solution increases continuously, and the emission peak is almost the same. The fluorescence of P1 shows good stability under different pH values, and the initial measured pH value of the ethanol solution of P1 is 6.0 because the assembly remains more stable in its own pH environment. In essence, the H + or OHin the solution can weaken the interaction between molecules by coordinating with the carbonyl group, destroy the stability of the supramolecular assembly, so as to relax the conformation and weaken the fluorescence through non-radiative relaxation. [27] Different metal ion sensitivities of P1 are validated in Figure 2F (ΔI = I 0 − I; I 0 is the emission intensity of P1 in water−ethanol solution without any metal ions; I is the emission intensity of P1 in water−ethanol solution with different ions). The fluorescence intensity of P1 decreased after adding Co 2+ , and it was apparent that the fluorescence spectrum showed a significant quenching effect after adding Fe 3+ . Whereas the P1 fluorescence intensity increases after the addition of Al 3+ , Ba 2+ , Na + , Cd 2+ , K + , Zn 2+ , Cu 2+ , Ca 2+ , and Hg 2+ . Such behavior could account for that Fe 3+ (3d 5 ) of semi-filled 3d orbital has strong orbital interactions and coordination ability with molecules, and Fe 3+ has strong oxidizing properties. Hence, it is easier to obtain electrons from the coordinating molecule system, and led to the charge transfer quench inside the molecule. It is worth highlighting that the strong quenching effect of Fe 3+ appears the specificity of the selection of polymer materials. [28] It is worth noting that the positions of the excitation and emission peaks in Figure 2A peaks changes, and the position of the emission peaks almost has no red shift or blue shift ( Figure 2E). That is, to say they have strong anti-interference ability, we speculate it is due to the stability of the hyperbranched structure, regardless of the DB. [29] The excellent performance combined with high quantum yield can effectively exploit the application of HBPSi in the field of bioimaging.

Mechanism study of HBPSi
To deeply investigate the mechanism of high quantum yield and spectral stability of HBPSi, theoretical calculation aided experiment was performed to explain this phenomenon. DLS was employed to obtain the particle size distribution of ethanol solutions with different concentrations of P1 and P2. As shown in Figures 3A and S9, the self-assembly ruler of P1 and P2 gradually increases as the concentration increases from 2 to 100 mg/mL. At low concentrations, about 2-10 mg/mL, the difference in particle size distribution between P1 and P2 is not obvious, distributing in range 3-30 nm.
As the concentration increases, the particle size of P1 not only increases to 6000 nm, but also has a wide distribution in the large size 100-6000 nm. The maximum particle size of P2 is only 900 nm even at 100 mg/mL, and most of them are around 100 nm. This is consistent with the observation from TEM ( Figure 3B). The microscopic morphology of different concentration P1 ethanol solution indicates that P1 assembles into irregular spheres with different sizes (the average diameter of about 200 nm-1 μm) with the increase in concentration. This all indicates that the increase in concentration and molecular weight will aggravate the aggregation of molecules. At the same time, theoretical calculation was carried out to rationalize the mechanism by using the geometry optimization and molecular orbital analysis. Figure S10A shows the optimized configuration of the first generation of molecular aggregates. It is obvious that as the number of molecules increases from one to three, the molecules aggregate more tightly. Similarly, as the generation increases from one to three, the molecules aggregate closer, as shown in Figure S10B. In particular, its molecular aggregation driving force was further studied and shown in Figure 4A, and an aggregate of three generations of one HBPSi molecule was taken as an example. Obviously, the molecules interact and self-assemble to form a supramolecular topology through the double hydrogen bonds, that is H⋅⋅⋅O and H⋅⋅⋅S (2.439-2.972Å). Driven by double hydrogen bonds, the aggregate is more compact compared with the previous single H⋅⋅⋅O drive, [17] thus effectively confining non-radiative decay. At last, we also used the time-dependent DFT method to optimize the conformation of the excited state of HBPSi at the B3LYP-D3/6-31G(d) level and the excited state optimized conformation for an aggregate of three generations of one HBPSi molecule, as shown in Table S2. The excitation energy of the first generation of three molecular aggregate is specially analyzed with oscillator strength (f) of 0.0151, as shown in Figure 4B excited to -S-S-on orbit of 887, and so on. Among them, the excitations of 883 → 895 and 883 → 896 account for the largest number, which are close to the actual peak position of the absorption spectrum, and the UV−vis spectrum is mainly due to electronic transitions, corresponding to atoms and functional groups in different orbital transitions in the particular excited state. In addition, the experimental results are in good agreement with the calculation results, which proves the reliability of the calculation method.
Generally, quantum yield, that is the utilization rate of photons, is the competitive result of radiative decay and nonradiative decay. [30] Restricting the dissipation of energy by intramolecular rotation and vibrational motion can effectively improve the quantum yield. [31,32] And luminescent polymers based on AIE luminogens of luminescent behavior are quite dependent on their structures that their luminescence efficiency significantly decreased with increasing spacer length, and DFT calculation shows that the reduction of spacer length enables the electron communication inside the structure and thus results in the extended π-π conjugated structure and the improvement of the luminescent efficiency. [33,34] It can be reasonably concluded that the quantum yield of HBPSi is due to weakening of non-radiative decay. On the basis of above fundings, we speculate: (1) increasing the concentration or the content of disulfide bonds favors the increase in size of the aggregates, and thus enhances the quantum yield. That is, the more compact structure of large-sized aggregates can effectively stiffen the conformation, confine non-radiative decay, and thus enhance the quantum yield. At the same time from P1 to P3, the quantum yield increases with the increase in molecular weight; (2) in addition, the hyperbranched structure has more branch points than the linear structure, and the molecule has a compact structure similar to a sphere, which can effectively rigidify the conformation, limit the non-radiative decay, and thus improve the fluorescence intensity and quantum yield. The quantum yield of P1 is higher than that of P2 partly due to the higher DB of

Application: biological safety and cell imaging of P1
In order to explore the biocompatibility and cell imaging of P1, MC3T3-E1 cells co-cultured on P1 of different concentrations coated cTi (cTi as control group, Figure 11 shows the surface is basically flat after P1 coating), and SEM, confocal microscopy, and CCK-8 are utilized to observe cells ( Figure 6). Figures 6A and S12B present the adhesion and spreading morphology of MC3T3-E1 cells on the surface of cTi and P1 coatings with different concentrations under SEM for 1 and 3 days. Obviously, by 1 day, most of the cells have begun to attach and spread to the cTi and P1 coatings, but they are not yet fully spread and are in the initial stage of adhesion. The whole is fusiform, and the number of stress fibers and plate pseudopods is small. The coating supports fast and tight cell adhesion and good spreading at 3 days. Stress fibers continually assembled in the following days, and finally, a cytoskeleton network developed to distribute on the whole surface after 3 days, suggesting that P1 coating did not disrupt the assembly of the cytoskeleton.
The same phenomenon is observed using fluorescence microscopy observation (Figure 6B), the cells on the 1.0 mg/mL P1 coating are sparsely distributed, and the cells on the 0.05 mg/mL P1 coating are equivalent to cTi. Quantitative data ( Figure 6C,D) compared with cTi display that the number of cells on the 0.5 mg/mL coating surface increased by 16.6%, and the cell spreading area is equivalent in each group. It indicates that P1 is beneficial to cell attachment and proliferation, which may be related to good biological activity of silicon element. The cell proliferation after 1, 3, and 5 days of culturing was also investigated via CCK-8 as shown in Figure 6E. At 1 day, that is, at the initial stage of cell proliferation, different concentrations of P1 coatings have the same biosafety as of cTi, and at 3 days, the low concentration of P1 coating is slightly higher than that of cTi. This is because (1) the HBPSi segment composition is similar to biological macromolecules and does not contain toxic aromatic structures such as benzene rings; [17] (2) Si element is a class of bioactive elements that have been shown to effectively promote the proliferation and differentiation of osteoblasts. [24] However, the proliferation activity of all P1 coating cells is lower than that of cTi at 5 days. Even so, the proliferation activity of P1 coating cells shows an increasing trend, and there is no significant difference between the proliferation activity of P1 coating cells and cTi at low concentration at late stage, and the cell activity decreased by less than 30%. In conclusion, the biosafety of P1 can be guaranteed at a reasonable concentration range, so the P1 coating can be considered non-toxic to the cells and has similar biosafety as cTi. It must also be mentioned that overcoming the traditional quenching of fluorescence aggregation is a hot spot in fluorescent dyes. The P1 coatings successfully illuminated the cells alone, and compared with conventional fluorescent dyes, the imaging efficiency does not decrease and the fluorescence intensity does not decrease after 1 day placement ( Figures 6B  and S12B). This is due to the traditional aggregation-induced quenching challenge that AIE materials overcome. But what needs to be improved is the continued development of dyes that can precisely light up the nucleus or cell membrane. In conclusion, this AIE P1 coating with both high quantum yield and spectral stability has excellent biocompatibility and broad application prospects in the field of cell imaging.

CONCLUSIONS
In summary, a novel non-conjugated HBPSi (P1) with a high fluorescence quantum yield (47.81%) and spectral stability has been successfully synthesized. Theory calculations and experiments explain that the excellent properties may attribute to formation of clusteroluminogens limiting nonradiative decay and stability of hyperbranched topologies. Most importantly, P1 not only has excellent biocompatibility, but also successfully illuminates mouse osteoblasts. Therefore, this study provides an effective method for designing HBPSi with high quantum yield for biological applications such as cell imaging.

C O N F L I C T O F I N T E R E S T
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

D ATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.