Polyester‐tethered near‐infrared fluorophores confined in colloidal nanoparticles: Tunable and thermoresponsive aggregation and biomedical applications

Intricate assembly of multiple molecular chromophores assisted by protein scaffolds is essential in tuning the optical absorption and energy transfer in the light‐harvesting complexes of the photosynthetic systems in nature. However, it remains a challenge to achieve such structural complexity and functionality in synthetic polymer‐chromophore systems. Here, we report a series of polyester‐tethered pyrrolopyrrole cyanine derivatives and their colloidal nanoparticles dispersed in water, which show tunable J‐ or H‐aggregation excitonic coupling and near‐infrared fluorescence by precise control of the polymer chain lengths, composition, and temperature. Moreover, the optimal fluorescence or photothermal effect of the J‐aggregate nanoparticles enables broad applications in fluorescence or photoacoustic bioimaging and phototherapy.


INTRODUCTION
Precise assembly of molecular chromophores assisted by proteins as scaffolds plays an important role in natural lightharvesting complexes in photosynthetic organisms, such as plants and bacteria. [1] Specifically, the intricate supramolecular organization of chromophores (e.g., chlorophyll dyes) on protein scaffolds leads to shift of light absorption to longer wavelengths, a characteristic phenomenon of J-aggregation, and promotes the efficient energy transfer at different size and time scales. Such J-aggregation has also been observed in a series of synthetic molecular chromophores, such as amphiphilic cyanine dyes in aqueous solution, [2][3][4][5][6] as well as derivatives of perylene bisimides, [7] porphyrins, [8,9] Cangjie Yang and Wei Zhang contributed equally to this work.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2022 The Authors. Aggregate published by SCUT, AIEI, and John Wiley & Sons Australia, Ltd. chlorophylls, [10] phthalocyanines, [11,12] BODYPI, [13,14] squaraines, [13,15] and hydroazaacene dicarboximide. [16] In addition, J-aggregates were also observed in the selfassembly of synthetic bacteriochlorins attached to peptides as a type of biohybrid light-harvesting systems. [2,3,7,10,13,[16][17][18][19][20][21][22][23][24] The manner that organic chormophores aggregate (or stack) is determined by multiple factors such as the molecular structure and external environmental conditions (temperature, pressure, [25][26][27] solvent, [28][29][30] etc.). The types of chromophore aggregates explored so far mainly include Jand H-aggregates, [31][32][33] as well as some other types of aggregates [34][35][36] depending on the manner that the chromophore molecules stack. Specifically, J-aggregates are formed by slip-stacking of molecules with a slip angle typically less than 54.7 • . The characteristic optical properties of J-aggregates, which have been extensively investigated by both theoretical and experimental studies, include red shifts of optical absorption and photoluminescence wavelengths versus those of the monomeric chromophores, narrowed emission band with aggregation-enhanced luminescence intensity, and coherence in exciton transport. In contrast to J-aggregates, H-aggregates are often formed by side-by-side stacking of molecular chromophores, resulting in characteristic blue shift of absorption band and significant luminescence quenching, compared to the optical properties of the corresponding monomers. Nevertheless, rational control of intermolecular aggregation of chromophores remains challenging and the efforts toward that goal have still been empirical.
We have previously reported the aggregation behavior in a coil-plate-coil molecule of pyrrolopyrrole cyanine (PPcy) derivative in the co-precipitation with a polymeric surfactant in a mixture of tetrahydrofuran and water. The flash precipitation of PPcy in water resulted in the formation of both H-and J-aggregates together with some monomers. [37] Nevertheless, the effects of chemical structures and molecular weights of synthetic polymers as scaffolds on J-aggregation coupling of molecular chromophores remain poorly understood. [38] Studies to address these issues are important to realize a rational molecular design of synthetic chromophore/polymer complexes for applications such as artificial photosynthesis and biomedical applications.
Herein, we report a series of polyester-tethered molecular chromophores by controllable polymerization from the two ends of a PPcy derivative as the initiator (Scheme 1 and Scheme S1A). The chemical structures and chain-lengths of the polyesters play a critical role in fine-tuning the excitonic coupling among PPcy chromophores within colloidal nanoparticles in water. Moreover, we present a new type of thermo-switchable near-infrared (NIR) emission caused by the reversible transformation between monomeric (M) and J-aggregate states of the fluorophores. We further demonstrate the potential of these thermosensitive materials for applications such as tumor imaging and phototherapy with high efficiency.

Synthesis and characterization of pyrrolopyrrole cyanine derivatives and polymers
We first synthesized a series of polymers (denoted as PPC x , where x represents the number-average degree of polymerization of polycaprolactone [PCL] on each side) by grafting biodegradable PCL from a hydroxyl-bearing PPcy initiator (P 0 ) via ring-opening polymerization (Scheme S1). The resulting polymers were characterized by nuclear magnetic resonance (NMR) spectroscopy ( Figure S1), gel permeation chromatography (GPC), and differential scanning calorimetry (DSC) (Figures S2 and S4). As the number-average degree of polymerization (x) increased from 5 to 20, the melting temperature (T m ) of PPC polymers gradually rose from 30.5 • C to 44.3 • C. T m remained nearly constant at 54.8 • C in the PPC  , which was close to that of blank PCL 37 (51.8 • C) without PPcy units ( Figure S4A). Colloidal nanoparticles (NPs) composed of these PPC x (1 mg/mL) were prepared in aqueous media by the typical nanoprecipitation method using PCL 37 -b-POEGMA 60 as a surfactant. [37][38][39] The average diameter of these NPs is approximately 45-65 nm, as measured by dynamic light scattering (DLS) and transmission electron microscopy (TEM) (Figures S5 and  S6).

Influence of chain length and polymer structure on optical properties of NPs
The optical properties of the PPC x NPs were then characterized. First, the UV-Vis-NIR absorption spectrum of PPC 5 NPs shows an intense and sharp J-aggregation band (denoted as J-band) with a maximum absorption wavelength (λ max,abs ) of 788 nm and a full width of half maximum (FWHM) of 46 nm ( Figure 1A). The steady-state fluorescence (FL) emission spectrum of PPC 5 J-aggregate showed a monomodal peak at 810 nm (λ max,em ) and a FWHM of 35 nm, and a FL quantum yield (Φ) (0.65%) 6.5-fold higher than that of the NPs composed of physically blended hydroxyl-bearing PPcy initiator (P 0 ) and PCL ( Figure 1B and Figure S2, Table 1). With the increase of the number-average degree of polymerization (x) from 5 to 25, the J-band in the absorption spectra was gradually attenuated, accompanied with the enhanced absorbance of monomeric PPcy (M-band, Figure 1A). In the FL emission spectra, an approximate 10 nm blue shift of λ max,em was observed with the increase of x from 5 to 20.
Meanwhile, compared to the relatively low Φ (0.65%) of PPC 5 , Φ of PPC 10 , PPC 15 , and PPC 20 were improved to 3.8%-4.6% (Table 1). The FL quantum yield was further improved to 8% for PPC 79 NPs and 24% for PPC 194 NPs. Such enhancement of Φ with an increase of PCL chain length could be attributed to the suppression of homo resonance energy transfer (HRET) due to the increased distance among J-aggregates. [40] This hypothesis was further confirmed by the dynamic fluorescence results ( Figure 2B), where FL lifetime of PPC 5 (0.50 ns) is significantly shorter than those of PPC [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25] (1.14-1.94 ns, λ em = 800 nm). Moreover, the FL lifetime of PPC 79 (0.77 ns, λ em = 760 nm) is much shorter than that of PPC 194 (2.05 ns, λ em = 760 nm), while the excitation wavelength was maintained same as 635 nm ( Table 1). The lifetimes of PPC 5 and PPD 2 NPs (where PCL is replaced with polydecalactone for PPD 2 ) were measured to be 0.50 and 0.59 ns, respectively, which were lower than those of other PPC x NPs (Figure 2A). PPC 25 also showed a dominant J-resonance emission despite only a small proportion of J-aggregates in the NPs, which could be attributed to the intra-particle Förster resonance energy transfer (FRET) from monomeric state to J-aggregates. [37] The apparently complexed trend of the fluorescence lifetimes of PPC 15 , PPC 20 , and PPC 25 NPs might be related to the co-existence of both PPcy monomers and J-aggregates and the resonance energy transfer between them upon photoexcitation at 635 nm. At x = 79 and 194, the J-band completely disappeared and the M-band turned dominant in the absorption spectra. The fluorescence images acquired by in vivo imaging system (IVIS) clearly show the strongest fluorescence of PPC  NPs at 820 nm, compared to the brightest emission of PPC  NPs at 760 nm ( Figure 1C,D). A systematic summary of the optical properties of all the PPC x NPs is presented in Table 1.
To confirm the important role of the covalent bonding between PCL and PPcy in the precipitation-induced   J-aggregation, we conducted a control experiment by encapsulating physically blended P 0 and PCL 37 with PCL 37b-POEGMA 60 into NPs in water. The feed ratio between P 0 and PCL 37 is approximately 1:3.3 (by mass), consistent with the theoretical molecular weight ratio between P 0 and PCL block in PPC 20 . A typical absorption feature of the dimeric and monomeric states of PPcy and a bluish green color of dispersion were observed. In contrast to PPC 5 NPs, the NPs composed of P 0 physically blended with blank PCL 37 displayed the typical H-aggregate and monomeric bands (denoted as H-and M-bands, respectively) with λ max,abs at 688 and 748 nm, respectively ( Figure 3A), and weak fluorescence ( Figure S13B). These results suggest that the covalent bonding between PPcy and PCL is crucial in the formation of J-aggregates in the colloidal NPs.
To investigate the influence of the polymer structure on the optical properties of the polymer-tethered PPcy, we synthesized another group of polyesters, denoted as PPL z (z represents the degree of polymerization of PLLA on each side of PPcy), by grafting poly (L-lactic acid) (PLLA) from P 0 via a similar synthetic scheme of ring-opening polymerization (ROP) (Scheme S1). In contrast to the J-band as observed in PPC x NPs, characteristic H-band appeared in the absorption spectra of PPL z NPs ( Figure 3D,E). These H-aggregate NPs showed weak to none fluorescence, a typical feature of H-aggregation. With the decrease of z from 22 to 10, the absorbance of H-band was gradually enhanced. Similar feature of the absorption spectrum can be observed in an amorphous isomer of PPL 10 , that is, poly(D,L-lactic acid), denoted as PP(DL)L 10 ( Figure 3E,F). While the mechanism of the different aggregate structures favorable for PPC and PPL remains unclear and warrants further study, it was also observed in some donor-acceptor merocyanine dyes that subtle change of the substituent groups resulted in dramatically different aggregating states (i.e., H-and J-aggregates) in spin-coated thin films on quartz substrates. [41] We were curious whether the formation of J-aggregation of fluorophores for PPC is due to the semicrystalline behavior of PCL. To verify the hypothesis, the amorphous biodegradable polymer polydecalactone (PDL) was chosen to replace PCL to afford PPD y . Due to the limited reactivity of ɛdecalactone (DL) in ROP, only PPD 2 and PPD 6 with low degree of polymerization of PDL were obtained. One can see that both PPD 2 and PPD 6 NPs showed sharp J-bands in their absorption spectra ( Figure 3B,C). Interestingly, the absorption spectrum of PPD 6 NPs, with an intense J-band at 785 nm and a shoulder peak at 735 nm, is quite similar to that of PPC 15 , although the chain length of PDL in PPD 6 is almost F I G U R E 2 Fluorescence decays of PPC x and PPD 2 nanoparticles (NPs). (A) Fluorescence decays at 760 nm for PPC  upon the excitation at 635 nm. (B) Fluorescence decays at 800 nm for PPC  and PPD 2 upon the excitation at 635 nm F I G U R E 3 UV-Vis-NIR absorption spectra of different NPs. (A) UV-Vis-NIR absorption spectra of P 0 /PCL 37 blending NPs. UV-Vis-NIR absorption spectra of (B) PPD 2 , (C) PPD 6 , (D) PPL 22 , (E) PPL 10 , and (F) PP(DL)L 10 NPs. "D" represents "dimer", "J" represents "J-aggregates", "H" represents "H-aggregates", and "M" represents monomers three-fold shorter than that of PCL in PPC 15 . These results suggest that PDL with amorphous nature can also induce the J-aggregation of PPcy fluorophores.
To explore the effect of different surfactants on PPcy aggregation, we used PEG 45 -b-PCL 105 , PEG 114 -b-PCL 53 , and PEG 66 -b-PLA 43 , in comparison with POEGMA 60b-PCL 37 , to encapsulate PPC 20 aggregates in water. The results are presented in Figure S8. It can be seen that with the same load (50% by weight) of PPC 20 , the dispersions stabilized, with PEG 114 -b-PCL 53 or POEGMA 60 -b-PCL 37 remained transparent, in contrast to the obvious turbidity of the dispersions stabilized with PEG 45 -b-PCL 105 or PEG 66 -b-PLA 43 , respectively. The dispersions remained turbid for the latter two surfactants even when the load of PPC 20 decreased to 33% by weight ( Figure S8B,D). Such turbidity is also reflected from the light-scattering effect in the short-wavelength region of the UV-Vis-NIR absorption spectra. These results suggest the larger loading capacities of PEG 114 -b-PCL 53 and POEGMA 60 -b-PCL 37 compared to those of PEG 45 -b-PCL 105 and PEG 66 -b-PLA 43 , respectively. The UV-Vis-NIR absorption spectra of the dispersions of PPC 20 stabilized by different surfactants all show characteristic absorption peak of J-aggregation. It appears that the formation of J-aggregates was suppressed to some extent in the dispersion stabilized by PEG 45 -b-PCL 105 . The colloidal NPs described below, unless specified otherwise, were all stabilized with POEGMA 60 -b-PCL 37 as the surfactant.

The effect of temperature on the optical properties of NPs
To characterize the thermoresponsive properties of the NPs as described above, we first applied thermal treatment to PPC 15 NPs, which were heated from 25 • C. Upon heating, the intensity of the J-absorption band at 789 nm decreased until disappearance at 55 • C, accompanied by the enhanced absorbance of the M-band at 740 nm ( Figure 4A,B). After cooling down and maintaining at 25 • C for 30 min, the J-absorption band was recovered nearly completely ( Figure 4C). The continuous attenuation of J-band with gradual increase of temperature ( Figure 4G,H) can be attributed to the heating-induced decrease of Flory-Huggins interaction parameter χ, that is, the incompatibility between PCL and chromophore. [42] In addition, the increased vibrational and/or translational mobility of both fluorophores and polymers upon heating may also facilitate the dissociation of J-aggregates and even order-to-disorder transition within NPs. [23] We further measured the absorption spectra over five heating-cooling cycles and found similarly periodic variations of the absorbance at 677, 740, and 789 nm, respectively, indicating the good reversibility of J-to-M transition ( Figure 4I).
Moreover, little change of the average diameter (56 nm) of NPs ( Figure S10) was observed by DLS after the heating process, suggesting good colloidal stability and integrity of the NPs. In the emission spectra of PPC 15 NPs at 55 • C, the FL intensity at 806 nm became 10-fold lower than the initial FL intensity at 25 • C. Notably, a full recovery of FL intensity was also observed after cooling down, further supported by the IVIS images ( Figure 4D-F). The dramatic attenuation of emission at 55 • C may be due to the reabsorption phenomena of monomers confined in NPs. [43,44] Similar phenomenon of aggregation-enhanced FL intensity was also observed in J-aggregates of other molecular fluorophores, [4] resembling nature's strategy of circumventing concentration quenching in photosynthetic light-harvesting systems. [1] We observed similar reversibility in the thermally induced J-to-M transformation for PPC 25 , PPC 20 , PPC 10 , and PPD 6 NPs (Figure S11 and S12; Figure 5A,B). Nevertheless, the critical annealing temperature (T c ) required to complete the J-to-M transition is higher for the PPC x NPs with shorter PCL chains. Specifically, T c of PPC 25 , PPC 20 , PPC 15 , and PPC 10 NPs increases in the order of 45 • C, 53 • C, 55 • C, and 65 • C, respectively. In addition, the J-band of PPC 5 NPs remains stable even at 70 • C ( Figure S12E). Similarly, no spectral change  Figure 5A and Figure S12). Moreover, some extent of H-to-M transformation was observed for PPL 10 NPs upon heating at 70 • C ( Figure 5C-E). Nevertheless, PPL 10 NPs remained almost nonfluorescent before and after heating. These results indicate that both the composition and the chain length of the tethering polymers could affect the phase transition temperature of the NPs, which was further studied by X-ray scattering spectroscopy in the following section.

Optical properties and phase separation of polymer films
To investigate the relationship between the optical properties and the phase separation of PPC x in solid states, we prepared thin films by drop-casting polymer solution (1 mg/mL in tetrahydrofuran) on glass substrates, followed by drying of the films under ambient condition. First, J-band was observed in the absorption spectra of PPC 15 film. At 70 • C, J-band disappeared completely ( Figure 6A). Subsequent cooling down to 25 • C enabled recovery of the J-band. The IVIS images of PPC 15 films showed the highest FL intensity at 820 nm ( Figure 6B-D). At 70 • C, the strongest FL intensity was observed at 780 nm. With temperature decreased to 25 • C, the maximal FL intensity was observed at 820 nm again ( Figure S13).
The phase separation of PPC x in films is closely related to the chain lengths of the polyesters. Figure 6E shows the patterns of small-angle X-ray scattering (SAXS) from PPC x films at room temperature. The SAXS profile of PPC 5 film at 25 • C shows multiple sharp peaks in the region of q = 0.05-0.40 Å -1 ( Figure 6E). The scattering pattern most resembles that of multiple continuous cubic phases as revealed in the phase separation of BAB-type block copolymers. [45] The SAXS profiles of both PPC 10 and PPC 15 films show three diffraction peaks at q = 0.15, 0.23, and 0.26 Å -1 with a characteristic ratio of 1: √ 3 ( Figure 6F and Figure S15A), suggesting presence of body-centered cubic (BCC) phase separation. In contrast to PPC 5 , PPC 10 Figure 6E). The disappearance of the diffraction peaks at q = 0.15, 0.23, and 0.26 Å -1 when x ≥ 25, suggests the domination of PCL phase and minimal phase separation between PPcy and PCL. These SAXS results suggest that shorter chain length of PCL leads to enhanced phase separation between PCL and PPcy in PPC x films. Figure 6G shows the wide-angle X-ray scattering (WAXS) profiles of PPC x in comparison to blank PCL 37 films at room temperature. Multiple diffraction peaks can be seen with maximal scattering intensities at q = 1.52 and 1.68 Å -1 , indexed as (110) and (200) intrinsic to a orthorhombic phase of crystalline PCL, in blank PCL 37 and all the PPC x films except x = 5 or 10. The WAXS profile of PPC 10 films shows obvious reduced intensities of the characteristic diffraction peaks of PCL, and emergence of two relatively broad peaks around q = 0.8 and 3.6 Å -1 , respectively. Nevertheless, these two peaks, together with other characteristic peaks of crystalline PCL, all disappear in the WAXS profile of PPC 5 films, with only a broad halo remaining in the region of q = 1.0-4.0 Å -1 .
We further monitored the evolution of phase transition of PPC 15 , PPC 10 , and PPC 5 films by both SAXS and WAXS in situ during the heating-cooling process. Figure 6F shows the SAXS pattern of PPC 15 film in a heating-cooling cycle from 30 • C to 55 • C and backwards. Figure 6F shows little change of diffraction peaks at q = 0.15, 0.23, and 0.26 Å -1 during the increase of the temperature from 30 • C to 50 • C. But all the three peaks disappeared when the film was heated to 55 • C. After being cooled down to 30 • C, the SAXS pattern recovered.
Compared to PPC 15 , PPC 10 film exhibited a similar temperature-dependent variation in SAXS profiles ( Figure  S15), but slightly higher thermal stability of the phase separation. The SAXS patterns, corresponding to BCC phase, of PPC 10 film at 45 • C and 55 • C are essentially similar to that at 30 • C. The three diffraction peaks at q = 0.15, 0.23, and 0.26 Å -1 disappeared after the film was further heated to 65 • C ( Figure S15A). The WAXS profiles of PPC 10 films at different temperatures ( Figure S15B) show that the intensity of the multiple scattering spikes corresponding to lamellae of crystalline PCL became obviously weaker at 45 • C and completely disappeared at 55 • C and 65 • C at which the polymer completely melted.
The SAXS profiles of PPC 5 films at different temperatures ( Figure S16A,C) show that there is no obvious change of the phase separation until 70 • C. The scattering intensity of the peaks is obviously reduced at 80 • C. The persistence of phase separation in PPC 5 film even at 80 • C is consistent with the limited attenuation of J-band absorbance of the NPs upon heating. A clear phase transition can be observed in PPC 5 film after being heated from 80 • C to 90 • C. At 90 • C, the SAXS pattern of PPC 5 showed three peaks at q = 0.11, 0.21, and 0.32 Å -1 , respectively, with a ratio of 1:2:3 that corresponds to the lamellae structure. The WAXS patterns of PPC 5 films remain the same essentially at different temperatures from 30 • C to 90 • C ( Figure S16B).
The SAXS and WAXS results described above indicate relatively rich phase-separation behavior in PPC 5 films in the temperature range of 25 • C-90 • C. With the increase of PCL chain length to x = 10 and 15, BCC phase separation, which is mainly contributed by the incompatibility between PPcy and PCL, remains in the polymer films in the temperature range of 25 • C-50 • C for PPC 15 and 25 • C-65 • C for PPC 10 . When the PCL chain is longer than x = 25, no obvious phase separation between PPcy and PCL could be observed and the polymer films shows X-ray scattering patterns intrinsic to that of blank PCL.

In vivo optical imaging and phototherapy of tumor models
Considering the relatively high fluorescence quantum yields (Table 1) Figure 7C). To assess the pharmacokinetics of PPC 15 and PPC 20 NPs, the same dose of NPs was injected into mice, from which the blood samples were collected and imaged at several time points ( Figure S17). Blood circulation half-time of PPC 15 and PPC 20 NPs were both 4.5 h.
Due to the relatively weak fluorescence and high fraction of PPcy, PPD 2 and PPC 5 NPs are more suitable for photoacoustic (PA) imaging and photothermal therapy ( Figure 8B; Figures S23 and S24). The photothermal efficiency of PPC 10 , PPC 5 , and PPD 2 NPs was examined by NIR laser irradiation at 808 nm with a power density of 1.77 W/cm 2 for 15 min ( Figure S23). The temperature of PPD 2 NPs (0.1 mg/mL) increased to 60 • C and maintained constant. In contrast to the photothermal behavior of PPD 2 NPs, the temperature of PPC 5 and PPC 10 NPs slightly decreased after reaching the maximum (53.4 • C and 50.8 • C, respectively), which could be due to the photothermally reduced absorbance at 808 nm of PPC 5 and PPC 10 NPs (Figure S12C,E).
We further characterized the temperature dependence of the PA spectra of PPC 10 , PPC 15 , and PPD 2 NPs as temperature-responsive PA probes for potential bioimaging applications ( Figure 8C-E). The PA spectrum of PPC 15 NPs shows two peaks at 710 and 770 nm, corresponding to monomeric and J-aggregate states, respectively ( Figure 8E). Upon heating at 45 • C for 10 min, the intensity of the PA signal at 770 nm decreased by 3.5-fold due to the disappearance of J-band. Similar results were obtained with PPC 10 NPs. In contrast, PPD 2 NPs exhibit no change in PA signal upon heating under the similar conditions. These results are consistent with the absence of phase transition of PPD 2 NPs in the range of 25 • C-50 • C.
Given the fact that the photoacoustic contrast of chromophores is proportional to their photothermal conversion efficiency, [46,47] PPD 2 NPs, due to their relatively high photothermal conversion efficiency as described above, were chosen for in vivo photoacoustic imaging of tumor models under pulsed laser excitation at 770 nm. Before the photothermal therapy, the biodistribution of PPD 2 NPs in tumor (HepG2) and other organs of mice was studied by both IVIS fluorescence ( Figure S18) and photoacoustic imaging ( Figure 8F). The results ( Figure S18) show obvious accumulation of the NPs in the tumor, liver, and spleen at 12 h. We further monitored the PA intensity at 770 nm at tumor site at different time points after administration of PPD 2 NPs ( Figure 8F). A weak PA signal in the tumor region was observed at 0 h (i.e., immediately after the tail-vein injection of the NPs) and enhanced obviously at 12 h, indicating that NPs gradually accumulated in the tumor ( Figure 8F). Similar accumulation in tumor and liver was also observed in PA imaging of the same mouse model administered with PPC 5 NPs ( Figure S24). For the photothermal therapy of the tumor model (HepG2), an 808-nm laser with a power density of 1.77 W/cm 2 was used to irradiate the tumors in xenografted tumor-bearing mice once at 12 h after tail-vein injection (dose of 6.7 μg/g) of the NPs.
The status of the tumor after the laser irradiation was monitored by IVIS fluorescence imaging system every 2 days (up to 10 days). Immediately after the laser irradiation for 10 min, some parts of the tumor in the PPD 2 -treated mice turned dark brown. Interestingly, the fluorescence of the PPD 2 NPs disappeared in the tumor region after the laser irradiation. After 2 days, the tumor region developed eschars and turned black in the whole tumor region ( Figure 8G). No obvious change was observed after 4 days, except for some slight recovery of the fluorescence in the tumor region. Some shrinkage of the tumor was observed after 8 and 10 days, respectively. Similar results, except for some recurrence of the tumor of some mice, were observed in the group of mice treated with PPC 5 NPs and the laser irradiation under the same condition as described above ( Figure S25). The comparison suggests (C-E) Photoacoustic spectra of PPD 2 , PPC 10 , and PPC 15 NPs at different temperatures. "J" represents "J-aggregates" and "M" represents "monomers." (F) In vivo photoacoustic images of mice tumors immediately (0 h) and 12 h after administration of PPD 2 NPs, respectively. (G) Images of mice treated with PPD 2 NPs under ambient light and the in vivo imaging system after irradiation by an 808-nm laser with a power density of 1.77 W/cm 2 for 15 min. The white arrows label the tumor site. (H) The evolution of tumor volume of mice treated with PPD 2 , PPC 5 , and PBS buffer after laser irradiation (n = 3 per group) that the phototherapeutic effect of PPC 5 NPs is not as good as that of PPD 2 ones, which is consistent with the superior photothermal effect of the latter as shown in Figure S23. In contrast to the obvious phototherapeutic effect of PPD 2 and PPC 5 NPs, the control group of mice treated with phosphate buffer saline (PBS) and the laser irradiation under the same condition did not show any phototherapeutic effect at all, and obvious growth of the tumor was observed over 10 days ( Figure S25). In addition, both of two NPs showed low cytotoxicity ( Figure S26) and excellent blood compatibility in vitro ( Figure S27). Negligible changes were observed in the time dependence of bodyweight of mice ( Figure S28), and no noticeable histopathological abnormalities were found in livers, kidneys, and spleens ( Figure S29), suggesting the good biocompatibility of these NPs.

CONCLUSION
We have presented synthesis, optical properties, and biomedical applications of a series of polyester-tethered fluorophores of PPcy derivatives and their colloidal nanoparticles dispersed in water. We found that both the chemical composition and chain length of polyesters affect the intermolecular aggregation among PPcy fluorophores and the optical properties of the nanoparticles. Relatively short chains of polycaprolactone or polydecalactone tethered from PPcy promote the formation of J-aggregates when co-precipitating in water in the presence of amphiphilic block copolymers (e.g.,

Characterization
1 H NMR measurements were conducted in chloroform-D using a Bruker AV300 MHz NMR spectrometer. Molecular weights of polymers were measured using gel permeating chromatography (GPC) (Agilent 1260, USA). The eluent was THF at a flow rate of 1.0 mL/min. A series of polystyrene standards with narrow polydispersities were employed for the GPC calibration. Fluorescence emission and excitation spec-tra were measured by fluorescence spectrophotometry on a PerkinElmer LS-55 at 25 • C. UV-Vis-NIR absorption spectra of the samples were measured on a SHIMADZU UV-2450 spectrophotometer. Fluorescence lifetime measurements were performed on a Horiba Fluolog 3 spectrofluorometer. TEM measurements were performed with a TEM Carl Zeiss Libra 120 Plus at an acceleration voltage of 120 kV. A 5 μl droplet of diluted samples was directly dropped onto a copper grid (300 mesh) coated with a carbon film, followed by drying at room temperature. The size distribution of resulting nanoparticles was determined by DLS using a BI-200SM (Brookhaven, USA) with a detection angle at 90 • . Samples for small-angle/wide-angle X-ray scattering characterization were prepared on Kapton tape with the dimension of 1 × 1 cm 2 . Before test, the bulk PPC polymers were placed in a vacuum oven at 100 • C for 2 h, and then left to gradually cool down to room temperature. SAXS/WAXS experiments were performed at Nanoinxider Xenocs at microsource, 40 μm, 30 W. The X-ray wavelength was 1.687 Å. The heating treatment on samples was conducted with heating rate of 2 • C/min, temperature increment of 10 • C, and stabilization time for 10 min at each temperature. The SAXS/WAXS curves were plotted as intensities (I) versus q, where q = (4π/λ)sin(θ), λ is the wavelength of the incident X-ray beam, and 2θ is the scattering angle. Fluorescence images of NPs dispersions were acquired with an IVIS Spectrum CT imaging system.

Synthesis of PPC and PPL [38]
Taking the synthesis of PPC 25 as an example, a mixture of Compound P 0 (0.02 mmol, 27.8 mg), ε-caprolactone (1 g), and SnOct 2 (3 mg) was placed in a Schlenk tube and stirred for 24 h in 3 mL anhydrous toluene at 120 • C under N 2 atmosphere. After cooling, the solution was poured into cold methanol. The product was obtained after further purification by precipitation in methanol three times and dried at 50 • C under vacuum. Other PPC and PPL polymers were synthesized by varying the feed ratios between P 0 and monomers as shown in Table S1.

Synthesis of PPD
A mixture of P 0 (6 mg), ε-decalactone (60 mg), and SnOct 2 (3 mg) was placed in Schlenk tube and stirred for 48 h in 1 mL of anhydrous toluene at 120 • C under N 2 atmosphere. After cooling, the solution was concentrated under reduced pressure and poured into methanol. We obtained 5.9 mg of PPD 2 after drying under vacuum at 80 • C. This product PPD 2 was further dissolved in 0.3 mL of toluene and mixed with 200 mg of ε-decalactone. The extension polymerization was conducted at 120 • C under N 2 atmosphere. After 48 h, the mixture was poured into excess methanol to obtain the precipitates after filtration. PPD6 was obtained (8.8 mg) after being dried under vacuum at 80 • C.

Preparation of nanoparticles
Polymers (5 mg) and amphiphilic copolymer PCL 37 -b-POEGMA 60 (5 mg) were dissolved in 0.5 mL of THF. These solutions were injected into 5 mL of water rapidly under ultrasonication. After evaporation of excess THF, the welldispersed solutions were obtained and the concentration was facilely adjusted by 10 × PBS to 1 mg/mL. Other PPC, PPL, and PPD-based NPs were prepared by using similar method.

Cytotoxicity study
The cytotoxicity of PPD 2 and PPC 20 NPs against HepG2 and HeLa cancer cells was evaluated by PrestoBlue (PB) assay. Briefly, HepG2 and HeLa cells were seeded in 96-well plates (Costar, IL, USA) at a density of 1 × 10 4 cells/mL. After 24 h incubation, the cells were exposed to a series of doses of NPs at 37 • C. After the designated time intervals, the wells were washed twice with 1 × PBS buffer, and 100 μl of freshly prepared PB solution in culture medium was added into each well. The PB medium solution was carefully removed after 1 h of incubation in the incubator. The absorbance of PB at 570 and 600 nm was monitored by a microplate reader (Genios Tecan). Cell viability was expressed as the ratio of the percentage PB reduction of the cells incubated with PPD 2 or PPC 20 suspension to that of the cells incubated with culture medium only.

Hemolysis analysis
Red blood cells (RBCs) were used to perform the hemolysis assay. The erythrocytes were collected from the whole blood of mice by centrifugation at 1500 rpm for 15 min and washed five times by saline. A stock of RBC dispersion was obtained by mixing 3 mL of centrifuged erythrocytes with 11 mL of saline. One-hundred microliter of RBC stock dispersion was added into 1 mL of PPD 2 , PPC 15 , or PPC 20 dispersions to give different concentrations of 50, 100, and 200 μg/mL, and the mixtures were incubated at 37 • C for 3 h. Saline and ultra DI waters were used as the negative and positive controls, respectively. After centrifugation at 12,000 rpm for 15 min, the UV-Vis absorbance of the supernatants at 788 nm was measured to determine the hemolysis percentage, which was calculated according to the following formula: hemolysis (%) = (A s − A n )/(A p − A n ) × 100%, where, A s is the absorbance of supernatant after incubating PPD or PPC solution with the erythrocyte suspension, A n is the absorbance of supernatant after adding the saline in the erythrocyte suspension, and A p is the absorbance following the addition of deionized water into the erythrocyte.

Tumor mouse model
The care and use of laboratory animals were performed according to the approved protocols of the Institutional Animal Care and Use Committee (IACUC) at Xiamen University (XMULAC20190146) and South University of Science and Technology (SUSTC-JY2017078). To establish 4T1 tumor models in BALB/c mice (female, 4-5 weeks old, Charles River, USA), 1 × 10 7 4T1 cells (breast cancer cell line, American Type Culture Collection) in 150 μl of 1 × PBS were injected subcutaneously in the right flank region of the mouse. To establish HepG2 tumor models in nude mice (female, 4-5 weeks old, Charles River, USA), 1 × 10 7 HepG2 cells in 100 μl of 1 × PBS were injected subcutaneously in the right flank region of the mouse. Tumors were allowed to grow to 150 ± 20 mm 3 before being used for in vivo imaging experiments.

In vivo and ex vivo optical imaging
In the experiments of long-term monitoring of biodistribution, after the nude mice were anesthetized using 2% isoflurane in oxygen, a series of nanoparticles (0.1 mL of 1 mg/mL solution) was systematically injected through the tail vein using a microsyringe. Fluorescence whole animal imaging was performed using in vivo IVIS Spectrum (PerkinElmer, USA). The fluorescence distribution was monitored at 1, 3, 6, 9, and 12 h after NPs administration using the in vivo imaging system with appropriate wavelength (λ ex = 710 nm, λ em = 760 ± 10, 780 ± 10, 800 ± 10, 820 ± 10, 840 ± 10 nm). After administration by a series of NPs, mice were sacrificed by cervical dislocation under deep isoflurane anesthesia at 12 h post injection (n = 7). The liver, tumor, spleen, kidney, heart, and lung were harvested for ex vivo fluorescence imaging to estimate the tissue distribution of the NPs. All data were analyzed by IVIS system software (Living Image 4.5.5).

4.11
In vivo PA imaging: PPD 2 and PPC 5 An aliquot of PPD 2 or PPC 5 NPs (0.1 mL of 1 mg/mL solution) was injected into the tail veins of nude mice that bore xenograft HepG2 tumors using a microinjector (n = 3). After 12 h of the injection, PA images of tumor and liver were acquired at 770 nm using a PA imaging system (Endra Nexus 128 scanner).

Pharmacokinetic study of NPs
An aliquot (0.1 mL) of nanoparticles (1 mg/mL in pure water) was systematically injected into 4-week-old female BALB/c mice (five per group) through the tail vein, then 30 μl blood was collected at 0, 1, 3, 6, 9, and 12 h through the blood vessels of tail after the injection and added with 1 mg/mL EDTA. The fluorescence intensity of blood samples was measured and calculated by IVIS imaging system.

Histological study
We dissected the mice and harvested the tumors, hearts, livers, lungs, spleens, and kidneys. Tumors and organs were treated with paraformaldehyde solution (4%), further paraffined and sliced them into 5 μm thickness. A fluorescence microscope (Olympus, IX73) was used to observe and take the images.

A C K N O W L E D G M E N T S
Mingfeng Wang acknowledges the financial support by Nanyang Assistant Professorship from Nanyang Technological University, AcRF Tier 2 (ARC 36/13), AcRF Tier 1 (2016-T1-001-214; 2018-T1-001-173) from the Ministry of Education, Singapore, and the University Development Fund (UDF01001806) from the Chinese University of Hong Kong, Shenzhen. We thank Dr. Jia Niu for helpful comments.

C O N F L I C T O F I N T E R E S T
The authors declare they have no conflicts of interest.