Molecularly precise self-assembly of theranostic nanoprobes within a single-molecular framework for in vivo tracking of tumor-specific chemotherapy

The strategy of molecularly precise self-assembly of theranostic nanoprobes within a single-molecular framework is used to avoid batch-to-batch variability, and concurrently achieving real-time tracking of the in vivo behaviour of prodrugs for the first time.


Introduction
Accurate tracking of in vivo tumor-specic behavior with probes is a perfect strategy for targeted sensing and controlled release of prodrugs. 1-3 However, a signicant limitation of such theranostic nanoprobe design is that multiple components for such a complicated scheme are oen required, inevitably leading to structural heterogeneity, insufficient reproducibility and subsequently huge barriers for clinical translation. 4 Current reporting strategies mainly focus on physical entrapment or chemical modication of drugs with multicomponent nanocarriers, including polymers, 5-15 liposomes 16-23 and inorganic materials. [24][25][26][27][28][29][30][31][32] With the help of these nanovehicles, probes with prolonged blood circulation duration and enhanced permeability and retention (EPR effect) show more effective and specic cancer treatment than free drugs. [33][34][35][36][37][38][39][40][41][42] However, the inevitable leakage and non-uniform loading efficiency based on the physical encapsulation system are insuperable barriers (Fig. 1a). In contrast, polymer-drug conjugates offer other notable benets to reduce premature leakage. [43][44][45][46][47][48][49] But the critical issue with polymer-drug conjugates is polydispersity in both the degree of polymerization and extent of loading attachment with chemical means (Fig. 1a). This inherent structural heterogeneity could cause signicant batch-to-batch variability, which is an impassable obstacle for clinical translation. Even worse, almost all current theranostics suffer from limitations that imaging and therapy are independently performed, rather than in an integrated protocol. 50 Thus, structural heterogeneity and the discrete steps of imaging and therapy make an unpredictable gap between how drugs behave in vitro and in vivo, that is, great difficulties in real-time tracking of drug release and evaluating therapeutic efficacy. To address these hurdles, the design of monodisperse nanomaterials with a single, reproducible entity that possess both in vivo diagnostic and therapeutic competencies is highly in demand.
Herein, we describe the rational design strategy of molecularly precise amphiphilic nanotheranostics by using functional building blocks. The main idea of this approach is to construct a structurally homogenous prodrug in a single, reproducible entity with synergistic targeting ability, which integrates the advantageous features of small molecular theranostics and polymer-drug conjugates (Fig. 1a). It makes full use of the two terminal conjunctions of the bis-condensed dicyanomethylene-4H-pyran (DCM) derivative as the activatable near-infrared (NIR) uorophore: the hydrophobic disulde-bridged anticancer drug camptothecin (CPT) and the hydrophilic PEG oligomer-bridged biotin segment serving as an active targeting unit. We focus on optimizing the hydrophilic fragment length to construct stable, well-dened nanostructured assemblies. Specically, these amphiphilic structures of biotin-PEG n -DCM-S-CPT or BP n -DCM-S-CPT ( Fig. 1) are composed of PEG n -biotin units (n ¼ 0, 5 and 20) as the controllable hydrophilic fragments and the covalently linked hydrophobic DCM-S-CPT moiety as the uorescent reporter. Notably, only BP 20 -DCM-S-CPT could spontaneously form uniform, stable, and reproducible core-shell micellar nanostructures (Fig. 1b). More importantly, the shell surface-graed biotin directly exposed to receptors on cancer cells can markedly facilitate cellular internalization. As demonstrated, these molecularly precise amphiphilic nanoprodrugs possess several striking characteristics: (i) well-dened monodisperse nanostructures with excellent reproducibility, high stability and xed loading efficiency; (ii) real-time tracking of active drug release; (iii) synergistic passive (preferable micelle-based EPR effect), active (biotin receptor-mediated endocytosis) and activatable (endogenous GSH-induced active drug release) targeting ability with extremely high inhibition rates of tumour growth (IRT). As far as we know, BP 20 -DCM-S-CPT is the rst molecularly precise self-assembled nanotheranostics which can be implemented for in situ and in vivo tracking of antitumor chemotherapy in living animal models.

Results and discussion
Rational design of molecularly precise theranostic nanoprobes within a single-molecular framework by tuning PEG segments In our system, all intrinsic building blocks are considered as both functionally active and structurally guiding units. The functional PEG n -biotin unit as a hydrophilic shell is not only used to stabilize the micelles and prolong the blood circulation time but also acts as an active targeting ligand resulting in receptor-mediated endocytosis 51 and enhanced uptake into tumor cells. 52,53 Bis-condensed dicyanomethylene-4H-pyran (DCM) is employed as the uorescent reporter owing to its attractive features such as controllable emission wavelength in Fig. 1 Amphiphilic self-assembled nanotheranostic core-shell systems. (a) Schematic illustration of general prodrug loading strategies: (I) physically encapsulated drug delivery system, physical encapsulation of the drug into the amphiphilic block copolymer, (II) polymer-drug conjugates, grafting the drug onto the amphiphilic polymer chain via covalent bonding, and (III) molecularly precise self-assembled amphiphilic theranostics, integrating the advantageous features of small molecular theranostics and polymer-drug conjugates. (b) Tumor specific molecularly precise self-assembled nanotheranostics for BP 20 -DCM-S-CPT with enhanced synergistic targeting including (i) passive targeting from the preferable micelle-based EPR effect, (ii) active targeting from shell surface-grafted biotin directly exposed to receptors on cancer cells for markedly facilitating cellular internalization via receptor-mediated endocytosis; and (iii) activatable targeting from endogenous GSH-induced active drug release in cancer cells.
the NIR region, large Stokes shi, high photostability ( Fig. S1a †), and particularly making full use of the two terminal reactive conjunctions. 54,55 The synthetic route to BP n -DCM-S-CPT is depicted in Scheme 1. A-DCM-NH 2 was initially reacted with the key intermediate CPT-S-OH in the presence of triphosgene at room temperature. Finally, BP n -DCM-S-CPT (n ¼ 5 and 20) prodrugs were obtained by the reaction of DCM-S-CPT with the azido-PEG n -biotin moiety via a typical 'click' reaction in a yield of 30%. Specically, in the 1 H NMR spectra (Scheme 1b), the propargyl proton (d ¼ 2.57 ppm) of DCM-S-CPT disappeared during the formation of BP n -DCM-S-CPT, while new protons (d ¼ 3.56-3.68 ppm) corresponding to PEG chains (-CH 2 -) were observed. Moreover, the two peaks at m/z 1504.5049 (corresponding to [BP 5 -DCM-S-CPT + H] + ) and 2164.8990 (corresponding to [BP 20 -DCM-S-CPT + H] + ) clearly further identify these molecularly precise structures from their individual high resolution mass spectra (HRMS, Scheme 1b). All the detailed procedures and characterizations are shown in the ESI. † As is well known, conventional carrier-based nanodrugs suffer from variable drug contents and release capacities from batch-to-batch. In contrast, molecularly precise self-delivery prodrugs have signicant advantages of xed and higher drug loadings. In our case, the xed drug content of BP n -DCM-S-CPT (n ¼ 0, 5 and 20) was directly calculated from its molecular structure, corresponding to the respective drug loadings of 36%, 23% and 16% (Table S1 †). These well-dened structural prodrugs can feature constant drug release capacities. The excellent reproducibility, structural homogeneity and xed loading efficiency of BP n -DCM-S-CPT completely avoid the inherent structural drawback of polymer-drug conjugates (Fig. 1a).
Critical effect of PEG segment lengths on forming highly stable amphiphilic micelle-based nanotheranostics As a unifying rule, when the ratio of hydrophilic moiety to total mass f hl in amphiphilic molecules is over 0.45, they can be expected to form stable micelles. The inherent amphiphilicity and suitable f hl (0.56) of BP 20 -DCM-S-CPT provide itself an opportunity to self-assemble into nanoparticles in aqueous solution. 56 To determine the size, morphology and stability of the selfassembled nanoparticles, a dimethyl sulfoxide (DMSO) solution of BP 20 -DCM-S-CPT was added dropwise into water, followed by dialysis against water to remove DMSO. A stable solution with a nal BP 20 -DCM-S-CPT conjugate concentration of 1.0 mg mL À1 was obtained. Notably, we observed that the assemblies of BP 20 -DCM-S-CPT had an average size of 87 nm measured by dynamic laser scattering (DLS) with a PDI (polydispersion index) of 0.25 (Fig. 2c). The size and morphology of the self-assembled nanostructures were further conrmed by transmission electron microscopy (TEM). As shown in Fig. 2e, TEM images reveal that the BP 20 -DCM-S-CPT micelles have uniform spherical shapes with an average size of approximately 70 nm. This size corresponds to that measured by DLS but only slightly smaller due to the drying stage during the TEM sample preparation.
Notably, the DLS measurements at different time intervals during 32 days demonstrated that these assemblies exhibited extremely high stability for long storage (Fig. S1b †). Moreover, the surface charge of the BP 20 -DCM-S-CPT solution was also investigated. The results show that the value of zeta potential is À3.9 mV, a lower negative surface charge, thus causing less repulsion between NPs and the cell membrane in aqueous solution. Most importantly, these well-dened BP 20 -DCM-S-CPT self-assemblies maintain their good initial stability over 96 h in fresh human serum at 37 C ( Fig. S2 and S3 †). To gain further insight into its self-assembly behavior in aqueous solution, the critical micelle concentration (CAC) of BP 20 -DCM-S-CPT was measured by using pyrene as a uorescent probe, and the results are shown in Fig. 2f. Notably, the CAC value of BP 20 -DCM-S-CPT is calculated to be only about 3.3 mg mL À1 (1.52 mM), which is much lower than those of reported amphiphilic self-assembled systems. 57 In contrast, when decreasing the PEG oligomer chain from 20 to 0 or 5, it became difficult for the resulting DCM-S-CPT and BP 5 -DCM-S-CPT to form stable self-assembled micelles, which was further conrmed by TEM (Fig. S4a †). These observations clearly indicated that the regulation of the suitable hydrophilic fragment length of PEG n -biotin units in BP n -DCM-S-CPT is very important to form stable, well-dened nanostructured assemblies. Denitely, it is of critical importance to balance hydrophobicity and hydrophilicity for the formation of well-dened nanostructures. In this case, we successfully build the strategy of molecularly precise amphiphilic self-assembly of prodrugs, wherein all the building blocks including the hydrophobic DCM-S-CPT moiety and hydrophilic PEG n -biotin units are vital structurally guiding elements rather than just functional units.
Activatable targeting: GSH-induced active drug release in synchronism with NIR uorescence signals Owing to its D-p-A structure, DCM-S-CPT exhibits a typical broad ICT absorption band at 455 nm with fairly weak emission at 529 nm in a mixed DMSO/PBS buffer solution (40/60, v/v, pH ¼ 7.4, 10 mM, Fig. 2a). On the other hand, A-DCM-NH 2 has strong broad uorescence under the same conditions. The distinct difference between A-DCM-NH 2 and DCM-S-CPT in the spectroscopic properties can be attributed to the masking of the electron-withdrawing amide bond in DCM-S-CPT, along with disruption of the electron donating ability of the nitrogen atom. 55,58 Notably, the emission band of A-DCM-NH 2 is broad so that A-DCM-NH 2 still has a very strong uorescence signal over 650 nm, which guarantees A-DCM-NH 2 as a promising uorophore for in vivo imaging.
First of all, to test our system containing an activatable disulde linker, the spectral properties of the prodrug DCM-S-CPT upon reaction with GSH were investigated in a mixed DMSO/PBS solution (40/60, v/v, pH ¼ 7.4, 10 mM). As expected, the prodrug DCM-S-CPT produced uorescence spectral changes as shown in Fig. 2a. Upon treatment with 2.5 mM GSH, signicant NIR uorescence was activated, which increased by about 10-fold upon excitation at 480 nm. Subsequently, the anticipated release of CPT as an active cancer drug was further proven by ESI-MS analyses. In the presence of GSH, the peaks of DCM-S-CPT at m/z 453.1 (corresponding to [CPT-SH + H] + ), 522.2 (corresponding to [A-DCM-SH + H] + ) and 418.2 (corresponding to [A-DCM-NH 2 + H] + ) were simultaneously observed in the HRMS (Fig. S5 †). Clearly, it was indicated that the active CPT can be released from the prodrug DCM-S-CPT upon exposure to GSH, with concomitant generation of the NIR uorescent reporter A-DCM-NH 2 by a two-step reaction (cleavage of the disulde bond and then intramolecular cyclization, Scheme S2 †). 59 Correspondingly, similar results of BP n -DCM-S-CPT (n ¼ 5 and 20) were also observed in the emission spectra (Fig. S6b †). All these pieces of evidence clearly indicated that these prodrugs could be specically activated by abundant GSH in aqueous solution.
Subsequently, the effect of self-assembled micellar nanostructure formation on GSH-induced active drug release was systematically studied. In our strategy, the consistent supply of bioactive CPT from the self-assembled nanostructures in the presence of GSH is anticipated to be directly visualized using uorescent reporter signals. As expected, in its molecular dissolution state, the time-course experiments of BP n -DCM-S-CPT (n ¼ 5 and 20) in the presence of GSH were found to correlate well with the continuous increase in uorescence intensity at 650 nm (arising from the activated DCM moiety in Fig. S6 †). Notably, in PBS solution (pH ¼ 7.4), the well-dened micellar assemblies of BP 20 -DCM-S-CPT exhibited slower responses towards GSH than BP 5 -DCM-S-CPT (Fig. 2b). Furthermore, TEM of BP 20 -DCM-S-CPT in aqueous solution demonstrated the formation of micelles in the absence of GSH (Fig. 2e) and complete micelle dissociation in the presence of GSH (Fig. S4b †), suggesting the active drug release induced by GSH. This observation suggested that the PEG 20 -biotin unit of BP 20 -DCM-S-CPT as the hydrophilic shell could reduce permeation of GSH into the activatable disulde-bridged hydrophobic core, which might diminish premature release side-effects.
Either in the self-assembled micellar state or a molecular dissolution state, neither CPT release by HPLC nor the uorescence enhancement of BP 20 -DCM-S-CPT was observed in the absence of GSH. As shown in Fig. S7, † upon addition of GSH, both the CPT release and uorescence enhancement of BP 20 -DCM-S-CPT in the molecular dissolution state reached a plateau within 30 min. In contrast, it took more than 35 min to reach the same plateau in the self-assembled state of BP 20 -DCM-S-CPT (Fig. 2d). Most importantly, we found that upon GSH-triggering, the consistent CPT release of self-assembled BP 20 -DCM-S-CPT (Fig. S8 †) is always synchronized with its continuous NIR uorescence enhancement in PBS solution. This could be attributed to molecularly precise BP 20 -DCM-S-CPT with a single, reproducible entity. All these pieces of evidence conrmed that this molecularly precise BP 20 -DCM-S-CPT has the key characteristics of the CPT release in synchronism with NIR uorescence enhancement. Thus, the turn-on NIR uorescence signal from the selfassembly of BP 20 -DCM-S-CPT with a single, reproducible entity could be used to precisely monitor the active CPT release in the presence of GSH, performing the activatable targeting function.
Having conrmed the GSH-driven NIR response of BP 20 -DCM-S-CPT in buffer solution, we then assessed whether these nanoassemblies can be used in real biological systems. The in vitro release behaviour of BP 20 -DCM-S-CPT with other biologically relevant amino acids, enzymes and serum markers was investigated (Fig. S9 †). Upon exposure to 1,4-dithiothreitol (DTT), cysteine (Cys), and homocysteine (Hcy), a similar spectroscopic response of DCM-S-CPT to GSH could be observed due to its thiol-containing structure. On the other hand, no appreciable uorescence enhancement could be induced by treatment with other non-thiol amino acids, enzymes and serum markers, conrming the specic cleavage of the disulphide bond elicited by thiol-containing species. Actually, the potential interference of DTT, Cys, and Hcy can be neglected due to their relatively low concentration in contrast to that with a high physiological concentration of GSH in the cytoplasm. Meanwhile, the BP 20 -DCM-S-CPT nanostructures exhibited enough high stability in an aqueous solution or fresh human serum with uniform size, which is ideal for tumor targeting by means of the EPR effect. All these results indicated that BP 20 -DCM-S-CPT can sustain in the inactive form under normal physiological conditions but is capable of consistently releasing active CPT in synchronism with turn-on NIR uorescence signals under the GSH-overexpressed physiological conditions. Active targeting: shell surface-graed biotin directly exposed to receptors on cancer cells for facilitating cellular internalization and visualizing drug release in living cells The in vitro toxicity of our designed molecularly precise prodrugs BP n -DCM-S-CPT (n ¼ 0, 5 and 20) was assessed by using the standard MTT assay. A number of cell lines including normal cells (QSG-7701) and cancer cells (SMMC-7721 and HeLa) were chosen to incubate with BP n -DCM-S-CPT. As depicted in Fig. 3a and S10, † all BP n -DCM-S-CPT prodrugs exhibit no cytotoxic effects on normal cells at the studied concentrations (0-20 mM). In contrast, the remarkably higher cytotoxicity of BP n -DCM-S-CPT in the presence of GSH was observed with cancer tumor cells, indicating that the released bioactive CPT by GSH is dominant and mainly responsible for the observed cellular toxicity (Fig. 3b). Meanwhile, all these prodrugs BP n -DCM-S-CPT exhibited signicantly higher cytotoxicity for cancer cells (SMMC-7721 and HeLa) because of abundant intracellular GSH in cancer cells (Fig. 3c and d). [60][61][62] Furthermore, upon addition of extra GSH, a further enhancement of cytotoxicity was found in cancer cells (Fig. S11 †). In conjunction with the above results, we can conclude that the observed toxicity mainly results from the GSH-triggered released active CPT inducing cell apoptosis.
In fact, the recognition and binding efficiency of biotin and biotin receptors is the key factor that signicantly inuences cancer cell seeking and cellular internalization. 63 As can be imagined, the shell surface-graed biotin of BP 20 -DCM-S-CPT directly exposed to the receptors on cancer cells can markedly facilitate cellular internalization via biotin receptor-mediated endocytosis. Thus, the effect of self-assembled nanostructure formation on the toxicity of prodrugs was also investigated. Notably, the general trends of BP n -DCM-S-CPT were clearly observed in cancer cells (SMMC-7721 and HeLa cells) that the sharply enhanced cytotoxicity was found with varying concentrations from 1 to 5 mM (Fig. 3c and d). Specically, the BP 20 -DCM-S-CPT assemblies exhibited the highest cytotoxicity in cancer cells, while BP 5 -DCM-S-CPT showed higher cytotoxicity in contrast to DCM-S-CPT. This could be mainly attributed to the more exposed biotin-graed surface targeting units with the core-shell micellar nature of BP 20 -DCM-S-CPT nanostructures (Fig. 1b). Taken together, these results validated that a higher cytotoxicity of BP 20 -DCM-S-CPT assemblies resulted from the more effective uptake by cancer cells. Cancer cell specic uptake is one of the important factors affecting therapeutic efficacy in tumor therapy. To understand whether the internalization of our self-assembled prodrugs might be improved, the properties of BP n -DCM-S-CPT were further evaluated in HeLa cancer cells by ow cytometry analysis. In fact, it is well-known that biotin receptors are overexpressed on many cancer cell surfaces, for example HeLa cells. 64 As shown in Fig. 4a, it took 12 h for the cellular uptake ratios of DCM-S-CPT to increase from 9.7% to 82.7%. In contrast, within only 3 h, the uptake ratios of BP 5 -DCM-S-CPT and BP 20 -DCM-S-CPT increased to 72.2% and 96.8%, respectively, meaning that the biotin units of these two prodrugs effectively increased the uptake efficacy of cancer cells (Fig. 4b and c).
To further verify shell surface-graed biotin facilitating cellular internalization, HeLa cells were pretreated with free biotin so that the overexpressed biotin receptor on the HeLa cell surface was mostly bound. Subsequently, we observed the slower uptake ratios of BP 5 -DCM-S-CPT and BP 20 -DCM-S-CPT (15.6% and 43.6%) within 3 h, respectively (Fig. 4d and e). All these ndings are thus fully consistent with our design concept of BP n -DCM-S-CPT, that is, incorporation with biotin unit can be uptake into biotin receptor-positive cancer cells with high efficiency. In particular, the shell surface-graed biotin on the BP 20 -DCM-S-CPT self-assemblies markedly facilitates cellular internalization of cancer cells.
Based on the tumor-specic intracellular uptake, the GSHinduced disulde linkage cleavage in concomitance with turnon NIR uorescence in BP n -DCM-S-CPT offers an opportunity for directly visualizing the in vitro drug release by confocal laser scanning microscopy. Aer 3 h incubation with cells at 37 C, only for cancer cells, the turn-on NIR uorescence from BP n -DCM-S-CPT was found in the cytoplasm (Fig. 5g-i), suggesting that the endogenous GSH triggers the active CPT release. Notably, of the three prodrugs, BP 20 -DCM-S-CPT exhibited the strongest turn-on uorescence signal in cancer cells. It is rmly evident that the BP 20 -DCM-S-CPT self-assemblies showed the most effective cellular uptake and quickly converted into emissive BP 20 -DCM-NH 2 and active CPT by endogenous GSH (Fig. 5i). Consistent with ow cytometry and the in vitro cytotoxicity by the MTT assay, all these uorescence measurements provided solid evidence that the BP 20 -DCM-S-CPT micellar assemblies are intrinsically suited for tumor targeted delivery and controlled release of CPT to tumor cells. We can attribute that the shell surface-graed biotin guaranteed the highly efficient cellular internalization of cancer cells (Fig. 1b). Thus, our strategy of molecularly precise self-assembly of theranostics within a single, reproducible entity provides great opportunities to gain insight into the mechanisms of their dynamic assemblies and targeted therapies.
Synergistic targeting: in situ behavior of self-assembled amphiphilic prodrugs in living animals The promising results in living cells such as specically seeking cancer cells and facilitating cellular internalization inspired us to further explore the feasibility of BP n -DCM-S-CPT as an in vivo NIR uorescence-tracking and synergistic targeting drug delivery system. The in vivo drug delivery performance and biodistribution of BP 20 -DCM-S-CPT, BP 5 -DCM-S-CPT and DCM-S-CPT were investigated with HeLa tumor-bearing mice aer intravenous injection at different time intervals, respectively. The nude mice were inoculated with HeLa cells on their right anks by injecting 10 6 cells subcutaneously. As shown in Fig. 6a, the biodistribution proles show that only a small amount of uorescence of DCM-S-CPT was located at the tumor site, while there was strong uorescence in the liver. In contrast, BP 5 -DCM-S-CPT with active targeting biotin exhibited an  enhanced amount of tumor accumulation (Fig. 6b). In particular, due to the specic self-assembled micelle with shell-graed biotin, BP 20 -DCM-S-CPT possessed both passive targeting from the EPR effect and active targeting, exhibiting a signicantly enhanced tumor accumulation aer 24 h injection (Fig. 6c).
The ex vivo uorescence images of excised tumors further conrmed the highest accumulation of BP 20 -DCM-S-CPT assemblies, with much weaker uorescence in the liver and no uorescence in the heart, spleen, lung and kidney (Fig. 6f). In contrast, DCM-S-CPT and BP 5 -DCM-S-CPT exhibited strong uorescence signals in the liver region, suggestive of a continuous CPT release triggered by the liver metabolism function towards small molecular prodrugs (Fig. 6d and e). In the case of BP 20 -DCM-S-CPT, the rational hydrophilic PEG length is very critical to form micelles, which can greatly facilitate passive EPR targeting and active targeting via the more exposed biotin-graed surface. Here all the synergistic targeting (passive targeting from the EPR effect, active targeting from shell graed biotin, and activatable endogenous GSH-driven targeting) makes BP 20 -DCM-S-CPT achieve high carrying efficiency, good targeting properties, and sustained and tumor-specic release in living mice.

In vivo anticancer activity with high tumor growth inhibition rates
To evaluate whether precise targeting and improved biodistribution result in the enhancement of therapeutic efficacy, the in vivo anticancer activities were compared using the HeLa xenogra tumor model in nude mice. The mice were randomly divided into six groups (n ¼ 6) in order to minimize weight and tumor-size differences. Mice bearing the tumors were intravenously injected with CPT, DCM-S-CPT, BP 5 -DCM-S-CPT and BP 20 -DCM-S-CPT at a CPT-equivalent dose of 10 mg kg À1 and phosphate buffer solution (PBS) as a control via the tail vein. The tumor volume and body weight of HeLa tumorbearing mice were monitored every 3 days for 21 days (Fig. 7a). CPT, DCM-S-CPT, BP 5 -DCM-S-CPT and BP 20 -DCM-S-CPT could inhibit HeLa tumor growth compared with the blank control group. As shown in Fig. 7b, compared with that of the PBS group, the inhibition rates of tumor growth (IRT) of CPT, DCM-S-CPT, BP 5 -DCM-S-CPT and BP 20 -DCM-S-CPT on HeLa tumors were 58.7%, 85.8%, 94.5%, and 99.7%, respectively. All BP n -DCM-S-CPT show low toxicity with no signicant weight loss during the treatment throughout the experiments, indicating that our designed prodrugs DCM-S-CPT, BP 5 -DCM-S-CPT and BP 20 -DCM-S-CPT groups did not cause severe systematic side effects, whereas CPT could lead to toxicity in mice (Fig. S13 †).
Obviously, BP 20 -DCM-S-CPT exhibited the highest antitumor efficacy, revealing the priority of passive targeting (EPR), active targeting and activatable targeting to achieve the highest tumor accumulation and carrying efficiency. In addition, BP 5 -DCM-S-CPT group treatment led to signicant inhibition of tumor growth compared to DCM-S-CPT (Fig. 7c), illustrating a better drug delivery efficacy via active targeting biotin. It's worth noting that BP 20 -DCM-S-CPT nearly eradicates the tumor ( Fig. 7c and d), suggesting that BP 20 -DCM-S-CPT almost cures the mice with cancer bearing tumors. These data verify that the specic nanoassemblies of BP 20 -DCM-S-CPT exerted excellent therapeutic activity in synchronism with in vivo turn-on NIR uorescence biodistribution via synergistic targeting in living animals.

Conclusions
In summary, we described the rational design strategy of molecularly precise self-assembled nanotheranostics for in situ and in vivo tracking of antitumor chemotherapy, in which PEG nbiotin units are utilized as the tunable hydrophilic fragments and the hydrophobic DCM-S-CPT moiety as an activatable NIR uorescent reporter. It is found that the hydrophilic PEG length is very critical to form the stable micellar nanoassemblies. Upon changing different PEG oligomers from 0 to 5 and 20, only BP 20 -DCM-S-CPT can simultaneously self-assemble into uniform micelles (ca. 80 nm) with a low CAC value, which displays high stability in fresh human serum. As demonstrated, the molecularly precise self-assembled nanotheranostics BP 20 -DCM-S-CPT can not only overcome the inevitable drug leakage and nonuniform drug payload based on the physically encapsulated drug delivery system, but also avoid the polydispersity in both the degree of polymerization and extent of drug loading in the polymer-drug conjugate system.
The well-dened monodisperse nanoassemblies of BP 20 -DCM-S-CPT possess the unique feature of real-time tracking of active CPT release in synchronism with turn-on NIR uorescence signals. BP 20 -DCM-S-CPT displays excellent tumor site-specic delivery in HeLa tumor-bearing nude mice via synergistic targeting, including passive targeting from the EPR effect, active biotin targeting (shell surface-graed biotin directly exposed to receptors on cancer cells markedly facilitates cellular internalization), and endogenous GSH-induced specic cleavage. As a result, BP 20 -DCM-S-CPT displays high tumor growth inhibition rates and in vivo anticancer activity and nearly eradicates the tumor (IRT ¼ 99.7%). In particular, the in situ synergistic targeting behavior of self-assembled amphiphilic prodrugs in living animals can overcome the limitations in the current theranostics, that is, imaging and therapy are independently performed, rather than in an integrated protocol. These in vivo and in situ behaviors with molecularly precise selfassembled theranostics make signicant insight into understanding amphiphilic self-delivery nanotheranostics, presenting new opportunities to drug-loading nanostructures.

Conflicts of interest
There are no conicts to declare.