Tuning nanoparticle core composition drives orthogonal fluorescence amplification for enhanced tumour imaging

Abstract

Tumour detection with high selectivity and sensitivity is crucial for delineating tumour margins and identifying metastatic foci during image-guided surgery. Optical nanoprobes with preferential tumour accumulation is often limited by inefficient amplification of biological signals. Here, we report the design of a library of hydrophobic core-tunable ultra-pH-sensitive nanoprobes (HUNPs) for orthogonally amplifying tumour microenvironmental signals on multiple tumour models. We find that tuning the hydrophobicity of nanoparticle core composition with non-ionizable monomers can enhance cellular association of HUNPs by more than ten-fold, resulting in a high cellular internalization efficiency of HUNPs with up to 50% in tumours. Combining high tumour accumulation and high cell internalization efficiency, HUNPs show orthogonally amplified fluorescence signals, permitting the precise locating and delineating margins between malignant lesions and normal tissues with high contrast-to-noise ratio and resolution. Our study provides key strategies to design nanomedicines with high intracellular bioavailability for cancer detection, drug/gene delivery, and therapy.

Introduction

As the cornerstone of cancer treatment, surgical resection always aims at the complete removal of the entire tumour bulk while preserving as much healthy tissues as possible¹. Therefore, fluorescence image-guided surgery has been developed rapidly in the past decade², providing dynamic intraoperative visualization of tumour-positive margins³. Several probes, including indocyanine green (ICG)⁴, 5-aminolevulinic acid (5-ALA)⁵, and methylene blue⁶, approved by the United States Food and Drug Administration (FDA)⁷, are employed in the surgical resection of various tumours. Although these small-molecule probes have been integrated into clinical practice, they typically exhibit suboptimal specificity and sensitivity due to their non-specific mechanisms of action⁸. To mitigate these limitations, researchers have developed targeted imaging probes that specifically target cell surface receptors overexpressed in cancer cells⁸. Targeted probes based on fluorophores conjugated to targeting ligands, such as antibodies^9,10,11 (e.g., Cetuximab-IRDye800CW for EGFR^9,10 and Bevacizumab-IRDye800CW for VEGFR¹¹), peptides¹² (e.g., cRGD-ZW800-1 for integrins¹²), and folate^13,14 (e.g., Cytalux for folate receptors^13,14, approved by FDA in 2021), have successfully visualized tumours due to the high specificity. However, the imaging contrast of molecular probes mainly depends on their actual concentration between tumours and nearby healthy tissues. Thus, such Always-ON mode of probe design results in high background signals and poor tumour-imaging outcome⁷.

To further enhance tumour-imaging contrast, the activatable fluorescent probes, also known as ‘turn-ON’ probes have been designed⁸. These imaging probes are fabricated to respond to dysregulated tumour microenvironment (TME), a universal feature of solid tumours, such as overexpression of specific enzymes^15,16, tumour acidosis^17,18,19, and redox stress²⁰. These probes keep quenched fluorescence signals in blood circulation and normal tissues, while fluoresce strongly when triggered by TME signals, thereby achieving high tumour-imaging contrast²¹. Generally, small molecular activatable probes displays poor pharmacokinetics and low target accumulation; in contrast, activatable nanoprobes show prolonged blood circulation and enhanced tumour accumulation through enhanced permeability and retention (EPR) effect^{22,23,24,25,26}. Thus, development of smart nanoprobes with high tumour accumulation and sufficient tumour-specific activation efficiency will enable the accurate identification of the tumours and nearby healthy tissues.

Previously, we designed a smart fluorescent nanoprobe for universal imaging of multiple tumours based on the ultra-pH-sensitive (UPS) nanotechnology^17,27,28,29. The cooperative protonation of tertiary amine-containing groups¹⁷ in the nanoparticle core renders pH-dependent reversible assembly and disassembly of nanoprobes with sharp responsiveness (∆pH_10–90% < 0.25) and tunable pH transition (pH_t) covering physiological pH ranges (4.0–7.4)³⁰. Exploiting the pH-induced exponentially signal amplification^17,18, the product Pegsitacianine, exhibited encouraging outcomes for image-guided tumour resection in clinical trials. The signal amplification of UPS nanoprobe lies on the orthogonal effect of tumour accumulation and fluorescence activation in lysosomes (pH~5.0) upon intracellular internalization. Our recent research has shown that the internalized UPS nanoparticles only account for ~20% of total tumour accumulation³¹, indicating that the majority of tumour-accumulated nanoparticles are distributed in the extracellular milieu and remain silent. Thus, there is an urgent need to enhance the cellular internalization and activation of tumour-accumulated nanoprobes for orthogonally amplified fluorescent signals.

In this work, we report a series of hydrophobic core-tunable UPS nanoprobes (HUNPs) that orthogonally respond to tumour accumulation and cellular internalization for exponential amplification of tumour microenvironment signals. The core composition of HUNPs is precisely tuned by copolymerization of one tertiary amine-containing monomer and a series of non-ionizable monomers with different hydrophobicity. HUNPs present similar physicochemical properties in terms of particle sizes, sharp pH-responsiveness, tunable pH_t, and ON/OFF fluorescence activation as traditional UPS nanoprobes. Serendipitously, we discover that the incorporation of high content of hydrophobic monomers in nanoparticle core dramatically enhanced the cell binding and internalization of nanoprobes, thereby achieving pH-induced fluorescent signal amplification (Fig. 1). We perform a screening and identified iP5B5 nanoprobe bearing 50% butyl methacrylate monomer in the hydrophobic core, demonstrating more than 10-fold cell internalization enhancement over traditional UPS nanoprobes, resulting in strong amplified fluorescent signals. Exploiting iP5B5 nanoprobe, we successfully detect a variety of tumours with high selectivity and sensitivity. This study contributes to the understanding of how to engineer nanomedicines with high cellular bioavailability for enhanced imaging and therapeutic efficacy.

Fig. 1: Design and mechanism of HUNPs for tumour imaging.

a HUNPs (hydrophobic core-tunable ultra-pH-sensitive nanoprobes) are composed of the same PEG shell as traditional UPS (ultra-pH-sensitive nanoprobes) and a hydrophobic core engineered by copolymerization of a tertiary amine-containing monomer with a series of non-ionizable monomers. The fluorescence signal of HUNPs was quenched due to Homo-FRET at micelle state (pH > pH_t), while exponentially illuminated upon the pH-induced disassembly of nanoparticles (pH < pH_t). b Tuning core composition of HUNPs with hydrophobic monomers enables enhanced nano-bio interaction and increased endocytosis for signal-amplified tumour-imaging efficacy. TME: tumour microenvironment. c Compared with traditional UPS nanoprobes that mainly regulate tumour signals by tumour accumulation, the enhanced internalization of HUNPs significantly optimizes the tumour-imaging efficacy. F.I.: fluorescence intensity.

Results

Design and characterization of HUNPs

We synthesized a series of amphiphilic copolymers (mPEG-b-P(R_i-r-R_n)) by our developed copolymerization strategy³². The poly (ethylene glycol) methyl ether (mPEG) acts as a hydrophilic block, and P(R_i-r-R_n) heteropolymeric segment is an ionizable hydrophobic block, where R_i and R_n refer to the tertiary amine-containing and non-ionizable monomers with different hydrophobicity, respectively. As an example, an ionizable tertiary amine-containing monomer, 2-(diisopropylamine) ethyl methacrylate (iDPA-MA) was chosen as a basic building block to generate the HUNPs library. The non-ionizable monomers, including butyl methacrylate (BMA) or hydroxyethyl methacrylate (HEMA) with distinct hydrophobicity were selected and copolymerized with iDPA-MA to tailor the composition of P(R_i-r-R_n) hydrophobic block (Fig. 2a). The total repeating units of R_i and R_n for all the synthetic polymers are 100. We characterized the chemical composition of synthetic copolymers by ¹H NMR spectroscopy. The synthetic copolymers exhibited a tunable chemical composition of hydrophobic block via adjusting the molar percentage of R_n in P(R_i-r-R_n) block from 0% to 50% (Supplementary Fig. 1 and Table 1).

Fig. 2: Synthesis, characterization, and in vivo tumour-imaging of HUNPs.

a Chemical structure of HUNPs. b pH_t of HUNPs with different mole fractions of HEMA and BMA monomers. c Particle size distribution of iP5B5 at pH 7.4 and 4.8 (3 independent experiments with similar results). d ¹H NMR spectra (in D₂O) of methylene protons of PEG-b-P(iDPA₅₀-r-BMA₅₀) copolymer at micelle and unimer states, respectively (3 independent experiments with similar results). e Real-time fluorescence imaging of 4T1 tumour-bearing mice after intravenous injection of HUNPs (n = 3 mice). Insets are rescaled images with individual scale bar. f The fluorescence intensity as a function of time for HUNPs at tumour sites (n = 3 mice, data at 48 h were analyzed using the one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). g The TNR (tumour-to-normal tissue ratio) as a function of time for HUNPs (n = 3 mice, data at 48 h were analyzed using the one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). h The CNR (contrast-to-noise ratio) of HUNPs at 48 h post-injection (n = 3 mice, data were analyzed using one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). i Normalized plasma fluorescence intensity of ICG versus time curves of HUNPs (n = 3 mice). All measurements are presented as mean ± s.d. Source data are provided as a Source Data file.

Fluorescence-labeled copolymers were synthesized through conjugation of fluorophores, including indocyanine green (ICG) and Cy5 to the mPEG-b-P(R_i-r-R_n) copolymers using NHS coupling reaction. HUNPs were cooperatively self-assembled from fluorescent copolymers that present different pH transitions (pH_t) values as shown in Fig. 2b and Supplementary Table 2. For easy description, HUNPs from mPEG-b-P(R_i-r-R_n) copolymers are abbreviated as R_iaR_nb, where a and b represent one-tenth repeating number of corresponding monomers (Supplementary Table 1). For example, HUNPs composed of mPEG₁₁₄-b-P(iDPA₅₀-r-BMA₅₀) refer to iP5B5. As we previously reported, amphiphilic copolymers self-assemble into core-shell nanoparticles through hydrophobic interactions at pH higher than their corresponding pH_t, resulting in fluorescence quenching (OFF state) due to the homo-Förster resonance energy transfer mechanism. As pH drops below the pH_t, pH-triggered disassembly of nanoparticles caused a florescence signal amplification (ON state) with two-orders of magnitude (>100-fold). All the HUNPs exhibited the sharp pH-responsiveness (∆pH_on/off ~ 0.20). Non-ionizable monomers, hydrophilic HEMA (30% mole fraction) and hydrophobic BMA (50% mole fraction) can tune the pH_t of iP10 from 6.42 to 6.76 and 5.22, respectively (Fig. 2b, Supplementary Fig. 2 and Table 2). The pH_t values of iP9B1, iP7B3, and iP5B5 are shifted to 6.20, 5.66, and 5.22, respectively, with an increased mole fraction of BMA in heteropolymeric block. In addition, Cy5-conjugated HUNPs also exhibited more than 100-fold fluorescent signal enhancement upon pH-triggered disassembly of nanoparticles (Supplementary Table 3). Dynamic light scattering analysis showed that nanoparticles present a diameter of 30 nm at pH 7.4, while dissociated into unimers with a diameter of 8 nm at pH 4.8 (Fig. 2c and Supplementary Fig. 3). The stiffness of HUNPs was quantified with atomic force microscopy. The obtained stiffness values ranged from 90 to 130 MPa (Supplementary Fig. 4), indicating that there is no significant difference in the mechanical properties between different HUNPs. HUNPs showed almost completely quenched fluorescent signals in fresh mouse plasma over 24 hours, indicating their good stability in blood circulation in vivo (Supplementary Fig. 5). We further performed ¹H NMR analysis of iP10 and iP5B5 in D₂O to investigate the core-shell structure of HUNPs (Fig. 2d). Results showed that PEG segment keep its typical chemical shift and constant peak intensity at either micelle or unimer states. In contrast, the proton resonance peak of iDPA and BMA/HEMA completely suppressed at micelle states. Thus, the results demonstrated that iP5B5 and iP10 are self-assembled core-shell nanostructure with PEG linear chain as the shell, while both iDPA and non-ionizable monomeric units are located in the micellar core.

HUNPs with higher content of hydrophobic segments exhibited enhanced imaging efficacy

For in vivo tumour-imaging, ICG-labeled HUNPs (20 mg kg⁻¹) were intravenously injected into mice bearing orthotopic 4T1 tumour (~200 mm³). Real-time fluorescence images were then captured using the in vivo imaging system. As shown in Fig. 2e–h, all nanoprobes exhibited a tumour-specific fluorescence signal and a high tumour-to-normal tissue ratio (TNR) ranging from 10.46 ± 1.15 to 27.20 ± 2.00 at 48 h post-administration. iP7HE3-ICG bearing 30% hydrophilic HEMA monomers in the hydrophobic core showed the lowest tumour fluorescence signals and contrast-to noise ratio (CNR, 26.93 ± 6.47) among all the nanoprobes (Fig. 2h). In addition, the fluorescence signals of iP7HE3 within the tumours were almost fully activated at 12 h post-injection, whereas those of other nanoprobes kept increasing over 48 hours. For BMA-bearing HUNPs, tumour fluorescence signals and TNR were significantly enhanced with the increased percentage of hydrophobic BMA monomers in micellar core (Fig. 2f, g). At 48 h post-administration, tumour CNR was greatly improved from 26.93 ± 6.47 for iP7HE3 to 157.67 ± 8.50 for iP5B5 (Fig. 2h), demonstrating segmental hydrophobicity of micellar core plays an important role in the improved tumour-imaging efficacy. Moreover, the biodistribution of nanoprobes in major organs was analyzed at 48 h post-injection of ICG-labeled HUNPs. Despite the fact that the majority of nanoprobes was distributed in liver and spleen, the fluorescence readout of tumours in BMA-bearing HUNPs groups was stronger than that in the liver and spleen, demonstrating a highly efficient signal amplification of HUNPs in tumour tissues over that in normal tissues (Supplementary Fig. 6). To evaluate the intratumoural distribution of HUNPs, 4T1 tumour-bearing mice were intravenously injected with Cy5-conjugated HUNPs, and tumours were excised at 48 h post-injection and cryosectioned for the immunofluorescence staining of LAMP1 (lysosome biomarker) in tumour slices. The results clearly demonstrated the substantial co-localization between the HUNPs nanoprobes and LAMP1-labeled lysosomes (Supplementary Fig. 7). This finding indicates that after tumour accumulation, HUNPs will be internalized by tumoural cells through endocytic trafficking pathways. Furthermore, the toxicity of HUNPs in vitro and in vivo was evaluated (Supplementary Fig. 8). HUNPs showed negligible cytotoxicity towards 4T1 tumour cells at a concentration below 300 μg mL⁻¹. The changes of animal body weight over 10 days post-injection of HUNPs at imaging dose (20 mg kg⁻¹) were monitored and histology of major organs at 24 h post-injection was investigated. The results showed no significant weight loss compared to PBS control and normal histology of major organs, demonstrating the good biosafety of these nanoprobes and facilitating our further research.

Hydrophobic core composition augments cell internalization of HUNPs in tumours

Having validated the enhanced tumour imaging of HUNPs bearing more hydrophobic segments, we proceeded to study the underlying mechanism of signal amplification in tumour tissues. Firstly, we performed the pharmacokinetic studies of ICG-encoded HUNPs after intravenous injection into healthy mice at a dose of 20 mg kg⁻¹. For all five HUNPs, plasma concentration-time curves over 24 h fitted a two-compartment model (Fig. 2i). As compared with iP10, iP7HE3 exhibited a significantly accelerated plasma clearance with β-phase half-lives decreasing from 10.71 ± 2.35 h to 3.69 ± 2.02 h, and the plasma AUC decreased to 34% of that of iP10 (Supplementary Table 4). The fast clearance of iP7HE3 may result in poor imaging readout in tumour tissues (Fig. 2e). However, the pharmacokinetic behavior of BMA-bearing HUNPs did not change significantly with tailoring the percentage of BMA in the micellar core. As compared with iP10, the AUC values of iP9B1, iP7B3, and iP5B5 only presented slight fluctuation from 81% to 124%. The long blood circulation of HUNPs is the prerequisite to high tumour accumulation as shown in Fig. 2e.

To evaluate the tumour accumulation, we prepared a series of ICG-labeled HUNPs with ‘Always-On’ mode using our previous procedure³¹. The fluorescent signal keeps constant at pH from 4.8 to 7.4 (Supplementary Fig. 9 and Supplementary Table 5), allowing the accurate quantification of nanoparticle accumulation in tumour tissues. iP7HE3 exhibited a significantly lower tumour signal than other four nanoprobes, which is consistent with its poor pharmacokinetic behaviour (Fig. 3a, b, and Supplementary Fig. 10). Intriguingly, iP10 and BMA-bearing HUNPs had a comparable tumour accumulation (Fig. 3a, b, and Supplementary Fig. 10), while iP5B5 presented up to threefold higher tumour fluorescent signal as compared with iP10 group (Fig. 2e, f). Thus, we proposed that the internalization of iP5B5 may be significantly improved after incorporation of the hydrophobic BMA monomers into nanoparticle core.

Fig. 3: In vivo imaging mechanism of HUNPs.

a Fluorescence images of 4T1 tumour-bearing mice with ICG-labeled ‘Always-On’ HUNPs nanoprobes at 48 h post-injection (n = 3 mice for iP7HE3 and iP10, n = 4 mice for iP9B1, iP7B3, iP5B5). b Tumour fluorescence intensity as a function of time for ‘Always-On’ HUNPs nanoprobes (n = 3 mice for iP7HE3 and iP10 group, n = 4 mice for iP9B1, iP7B3, and iP5B5 group, data at 48 h were analyzed using the one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). c Schematic illustration of the binary ratiometric nanoreporter system of HUNPs. d Dual channel fluorescence imaging and the ratiometric images of 4T1 tumour-bearing mice treated with the binary nanoreporter system of HUNPs at 48 h post-injection. The ratiometric channels were generated from the ratio of ICG to Cy5 by ImageJ software (n = 3 mice). e The ratiometric data as a function of time (n = 3 mice, data at 48 h were analyzed using the one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). f The ratiometric results of ex vivo tumours at 48 h post-injection. g Whole-mount imaging analysis of HUNPs internalization in tumours at 24 h post-injection of the binary nanoreporter of iP5B5 (left) and iP10 (right) (3 independent experiments with similar results). The upper right green ‘On-Off’ channel: BDP; the middle right red ‘Always-On’ channel: Cy5; the lower right blue ‘DAPI’ channel: nuclei; the upper left ‘Ratio’ channel: ratio of BDP to Cy5; the lower left ‘Merge’ channel: merge of BDP and Cy5 channels. Scale bar = 50 μm. h The ratiometric images and quantitative results of CT26 tumour-bearing mice with the binary nanoreporter of HUNPs at 24 h post-injection (n = 3 mice). i The ratiometric images and quantitative results of MC38 tumour-bearing mice with the binary nanoreporter of HUNPs at 24 h post-injection (n = 3 mice). All measurements are presented as mean ± s.d. Source data are provided as a Source Data file.

To further interrogate the imaging mechanism, we constructed a binary ratiometric nanoreporter of HUNPs, which can unequivocally distinguish between nanoparticles located inside and outside cells, allowing the accurate quantification of nanoparticle internalization in vivo. The binary system consisted of two fluorescence modules. The ‘On-Off’ micelle module keeps ‘Off’ state in the bloodstream and extracellular space of the tumour (pH_e ~6.5–6.9), and switches to ‘On’ state after endocytosis, while the ‘Always-On’ micelle module keeps signal constant before and after endocytosis (Fig. 3c). The ratiometric images of ‘On-Off’ to ‘Always-On’ channels can be exploited to assess the internalization capacity of nanoparticles. HUNPs labeled with ICG (On-Off) and Cy5 (Always-On) (Supplementary Table 6) in the near-infrared window were selected for in vivo quantitative imaging of internalization, while nanoparticles encoded with BDP (On-Off) (Supplementary Table 7) and Cy5 (Always-On) were utilized as an alternative for the internalization evaluation of tumour slides.

For in vivo quantification of internalization, the cocktail of the ‘On-Off’ (20 mg kg⁻¹) module and ‘Always-On’ module (40 mg kg⁻¹) was intravenously administrated into mice bearing 4T1 breast tumours. As shown in Fig. 3d, e, with the increased proportion of hydrophobic monomers in the nanoprobe, there was no significant difference in tumour accumulation of nanoprobes indicated by the identical fluorescent signal of the ‘Always-On’ system. Meanwhile, the radiance efficiency of the ‘On-Off’ module increased gradually along with the percentage of hydrophobic monomers increased (Supplementary Fig. 11). The ratiometric signal of the ‘On-Off’ versus ‘Always-On’ modules also increased as the proportion of hydrophobic monomer increased. For example, the ratiometric readout of iP5B5 nanoprobe is 3.1-fold higher than that of iP10 (0.90 ± 0.23 versus 2.88 ± 0.13, P = 0.000003). The ratiometric results revealed that as the proportion of hydrophobic monomers in the nanoprobe core increases, the nanoparticle internalization in tumour tissues is greatly enhanced, thereby improving tumour-imaging efficacy. The ratiometric signal of ex vivo tumours also increased from 0.79 ± 0.16 of iP10 to 2.14 ± 0.17 of iP5B5 (Fig. 3f).

We also explored tumour accumulation and internalization of iP5B5 and iP10 by fluorescence imaging of tumour sections. As shown in Fig. 3g, the fluorescent intensity of ‘Always-On’ module was similar between the two groups, while the signal of ‘On-Off’ module in iP5B5 was significantly stronger than that in iP10, resulting in a significantly higher ratiometric readout (Fig. 3g, and Supplementary Fig. 12). The results further demonstrated that the hydrophobic monomer in micellar core plays a critical role in enhanced cell internalization rather than tumour accumulation.

We next evaluated the imaging efficacy of two representative nanoparticles, iP5B5 and iP10, on other two tumour models using binary ratiometric imaging method. The results showed that tumour ratiometric signal output of iP10 and iP5B5 was 1.44 ± 0.10 and 2.77 ± 0.42 on CT26-bearing mice, respectively (Fig. 3h and Supplementary Fig. 13), and 1.35 ± 0.25 and 3.06 ± 0.57 on the MC38-bearing mice, respectively (Fig. 3i and Supplementary Fig. 14). Collectively, our results demonstrated that tuning nanoparticle core composition with more hydrophobic monomers can enhance the cellular internalization in tumour tissues, thereby achieving fluorescence-amplified tumour imaging.

The hydrophobic segment drives cell binding and uptake of HUNPs

Having demonstrated the enhanced cellular internalization of HUNPs bearing high content of hydrophobic monomers in the nanoparticle core in vivo, we further investigated the interaction of HUNPs with different cell types in vitro. We initially investigated the kinetics of nanoparticle internalization, cell trafficking and signal activation processes. A previously reported binary ratio nanoreporter³¹ was used, employing BDP as the ‘On-Off’ module (Supplementary Table 7) and Cy5 as the ‘Always-On’ module (Supplementary Table 6). To quantify the kinetics of internalization, the cell suspension obtained by trypsinization was placed on ice for 10 minutes, followed by centrifugation and resuspension in the binary ratiometric nanoreporter system (100 μg mL⁻¹) at 4 °C. After being pulsed for 10 minutes (cell binding) and washed three times, the samples were chased at 37 °C (internalization) by measuring the fluorescent intensity of each channel for single cells on a flow cytometer. As depicted in Supplementary Fig. 15, the iP10 group showed complete signal activation after 90 minutes, iP9B1 after 105 minutes, iP7B3 after 180 minutes, and iP5B5 after 210 minutes. Additionally, the in vitro cell trafficking and signal activation processes of HUNPs after endocytosis were characterized using A549-RAB5A-GFP and A549-LAMP1-GFP cells (Supplementary Fig. 16). During the process of cellular trafficking, co-localization was observed between HUNPs nanoparticles and RAB5A-labeled early endosomes at early time-points such as 0.5 and 1 hour. Furthermore, at later time-points such as 4 and 6 hours, co-localization was observed between HUNPs and LAMP1-labeled lysosomes, indicating the cell trafficking process through the endosome/lysosome pathway. After nanoparticle internalization, the gradual increase in endosomal acidity resulted in the activation of HUNPs with different pH_t. At 4-hour time-point, the fluorescent signals of all HUNPs can be completely activated in tumour cells. Consequently, an incubation period of 4 hours ensured the full activation of fluorescent signals for different nanoprobes. Therefore, 4T1 tumour and iBMDM macrophages were pulsed with nanoprobes (50 μg mL⁻¹) for 10 min at 4 °C for cell binding and incubated for 4 h at 37 °C for complete signal activation. Flow cytometry of single cell suspension was performed to evaluate the nanoparticle-cell association efficacy. Results showed that the cell binding of HUNPs was enhanced gradually as the percentage of BMA in nanoparticle core composition increased (Fig. 4a and Supplementary Fig. 17). Strikingly, iP5B5 exhibited 10.34 ± 1.56 and 11.37 ± 1.64 -fold higher cell binding than those of iP10 in 4T1 and iBMDM, respectively. There is a positive correlation between the cell binding capacity and the proportion of hydrophobic monomer BMA in the nanoparticle core. Although all HUNPs shared the same PEG shell, an increased proportion of BMA in the hydrophobic core still augmented the nano-bio interaction, revealing the crucial role of core composition of nanoparticles in this enhanced interaction.

Fig. 4: Enhanced nano-cellular interaction of HUNPs with tunable hydrophobicity of core composition.

a The relative-binding amount of HUNPs to 4T1 cell membrane (relative to iP10) (n = 4 independent experiments, one-way ANOVA post-Brown-Forsythe test, followed by Tukey’s multiple comparisons test). b Correlation between the relative binding amount of HUNPs and the molar percentage of BMA in micelle core (n = 4 independent experiments). R value: derived using a linear regression model. Error band in grey: 95% confidence intervals of the fitted line. Centre of the error band: the predicted values from the linear regression model. P value: slope significantly non-zero. c The uptake of HUNPs by 4T1 tumour cells. The four peaks from left to right are as follows: blue for iP10, purple for iP9B1, green for iP7B3, and red for iP5B5. d The relative uptake of HUNPs by 4T1 cells (relative to iP10) (n = 3 independent experiments, one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). e Correlation between the relative in vitro uptake and the relative in vivo tumour internalization as shown in Fig. 3d. R value: derived using a linear regression model. Error band: 95% confidence intervals of the fitted line. The centre: predicted values from the linear regression model. P value: slope significantly non-zero. f Intracellular activation of iP5B5-Cy5 or iP10-Cy5 in 4T1 tumour cells. Scale bar = 10 μm. g Quantification results of nanoparticle activation in lysosomes of tumour cells. (n = 20 independent cells per group, two-sided unpaired t test). The dashed line in the middle of the violin plots: the median; upper and lower dashed lines: the first and third quartiles. The width of the violin plot: density of the data at different values. F.I.: fluorescence intensity. h The relative cellular uptake of HUNPs in different tumour cell lines and macrophage cell lines. The cell uptake for each group was normalized to that of iP10 (n = 3 independent experiments). Independent experiments mentioned above are derived from biologically different cell samples. All measurements are presented as mean ± s.d. Source data are provided as a Source Data file.

The cellular uptake of HUNPs was also analyzed. After 4 h incubation with nanoprobes, 4T1 cells were analyzed by flow cytometry. The fluorescent signals of cells were significantly increased with the enhanced proportion of BMA (Fig. 4c, d). The fluorescence signal of iP5B5 was 8.73 ± 0.44 -fold higher than that of iP10. We also found a good correlation between the relative cellular internalization of HUNPs in 4T1 tumour tissues and the relative uptake of HUNPs by 4T1 cells (Fig. 4e). The enhanced cell uptake of iP5B5 was further demonstrated by confocal microscopy (Fig. 4f). Quantitative analysis of confocal images was performed on 20 cells for each nanoprobe as shown in Fig. 4g. The average fluorescence intensity of iP5B5 was up to 22.25-fold higher than that of iP10. The internalization pathways of different HUNPs were investigated in 4T1 tumour cells. The impact of different inhibitors on cellular uptake of nanoparticles was explored, including chlorpromazine inhibiting clathrin-mediated endocytosis, dynasore inhibiting dynamin-dependent endocytosis pathway, EIPA for macropinocytosis, nystatin for caveolae-mediated endocytosis, and genistein as a protein tyrosine kinase inhibitor. The results (Supplementary Fig. 18) demonstrated that iP10 enters cells through multiple pathways, including clathrin-mediated endocytosis, caveolae-mediated endocytosis, and macropinocytosis. As the percentage of hydrophobic monomer increases, the macropinocytosis pathway is gradually blocked, and the nanoparticles rely more on the clathrin-mediated and caveolae-mediated pathways for cellular internalization.

To further demonstrate the potential of HUNPs on enhanced cell internalization, we performed the uptake experiment in various cell lines, including five tumour and four macrophage cell lines. In all tumour and macrophage cell lines, the cell uptake of HUNPs was significantly enhanced as the increased percentage of BMA in nanoparticle core (Fig. 4h and Supplementary Fig. 19). Different cell types exhibited varying degrees of increased uptake of the nanoprobes. For example, compared to iP10, PANC02 cells showed only 2.6-fold uptake of iP5B5, while CT26 and BMDM exhibited up to 8-fold uptake of iP5B5. Therefore, tuning core composition with more fraction of hydrophobic monomers led to enhanced nano-cell interaction, which resulted in enhanced uptake of nanoprobes and fluorescence-amplified tumour-imaging efficacy in vivo.

Hydrophobicity of non-ionizable side-chain rather than pH_t drives the cell internalization of HUNPs

As shown in Fig. 2b, copolymerization of hydrophobic non-ionizable monomer results in pH_t decrease of HUNPs. Our previous research demonstrated that the nanoparticles with lower pH_t have a higher tumour-imaging efficacy³³. Thus, we next explored whether the enhanced imaging effect of HUNPs is caused by its decreased pH_t. We chose three fluorescent nanoprobes, iP10, iP5B5 and DB10, with the latter two probes having the same pH_t value for tumour imaging. DB10 copolymer was synthesized by copolymerization of tertiary amine-containing monomers, 2-(dibutyl amino) ethyl methacrylate (DBA-MA) with 100 repeating units (Fig. 5a and Supplementary Fig. 20). All three nanoprobes have more than 100-fold fluorescence activation upon disassembly at pH lower than its pH_t. Compared to iP10, the pH_t of iP5B5 and DB10 decreased to around 5.2. Mice bearing 4T1 tumours were intravenously injected with ICG-conjugated iP10, iP5B5, and DB10 (20 mg kg⁻¹). After injection, the fluorescence images were captured at 48 h post-injection. As shown in Fig. 5b, c, iP5B5 exhibited a significantly stronger imaging efficacy compared to DB10. Tumour fluorescent signals of DB10 and iP5B5 groups were 2.1- and 3.5-fold of iP10 group, respectively. Therefore, although the reduced pH_t has a certain contribution to the enhanced imaging efficacy, the micellar core composition modified by non-ionizable monomers and the resulting enhanced nano-cell interaction may make a greater contribution.

Fig. 5: Optimization of hydrophobic core composition of HUNPs for efficient tumour imaging.

a Schematic illustration of different imaging mechanism of UPS and HUNPs. b Images of 4T1 tumour-bearing mice at 48 h post-injection of ICG-labeled iP10, iP5B5, and DB10 ‘On-Off’ nanoprobes (n = 4 mice). c Tumour fluorescence intensity from b as a function of the pH_t of nanoprobes (n = 4 mice). Radiant efficiency data were analyzed by one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test. d Images of 4T1 tumour-bearing mice with ICG-labeled iP10, iP5B5, and DB10 ‘Always-On’ nanoprobes at 48 h post-injection (n = 3 mice). e Fluorescence intensity from d as a function of pH_t (n = 3 mice). Radiant efficiency data were analyzed by one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test. f Relative endocytosis of each group normalized to that of iP10 (n = 3 mice, one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). g Images of 4T1 tumour-bearing mice at 48 h post-injection of ICG-labeled nP10, nP5B5 (n = 3 mice). h Fluorescence intensity from g and from 48 h time-point of Supplymentary Fig. 20c (n = 3 mice for nDPA group, n = 4 mice for EPA group, two-sided unpaired t test for each group). i Images of 4T1 tumour-bearing mice at 48 h post-injection of ICG-labeled nanoprobes comprised of 50% of different hydrophobic monomers in micellar core (n = 3 mice). j Fluorescence intensity from i as a function of alkyl carbon atoms number in the hydrophobic monomers (n = 3 mice, one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons test). The number 0 refers to iP10-ICG. k Images of 4T1 tumour-bearing mice with iP5B5-ICG or free ICG (n = 4 mice). l Images of splayed CT26, MC38, MCF-7, and KPC tumours during the mock surgery process at 24 h post-injection of iP5B5-ICG (n = 5 mice for CT26 and MCF-7, n = 4 mice for MC38, n = 3 mice for KPC). All measurements are presented as mean ± s.d. Source data are provided as a Source Data file.

The ‘Always-On’ systems of these nanoparticles (Supplementary Table 5) were further established to quantify their tumour accumulation. Results showed that the tumour accumulation of DB10 was 2.2-fold of iP10 and iP5B5 (Fig. 5d, e, and Supplementary Fig. 21), while there was no significant difference of tumour accumulation between iP5B5 and iP10 groups. The ratiometric readouts of fluorescent activation signal to absolute tumour accumulation of nanoparticle was calculated to evaluate the efficiency of nanoparticle internalization within tumours (Fig. 5f). Results showed that the endocytosis of iP5B5 was more than 3.5-fold higher than that of iP10 and DB10, while the latter two exhibited a comparable cell internalization in vivo. Moreover, we evaluated the cellular uptake of DB10 and iP5B5 in vitro using flow cytometry. iP5B5 presented a 15.5-fold higher cell uptake than DB10, while DB10 showed slightly lower cellular uptake than iP10 (Supplementary Fig. 22). Thus, for UPS nanoprobes with 100% tertiary amine-containing hydrophobic segment, the pHt is a dominant parameter for tumor accumulation and imaging efficacy (e.g., iP10 versus DB10). In contrast, for HUNPs with hydrophobic heteropolymeric segment, the hydrophobic non-ionizable side-chain elicits enhanced nano-bio interaction and fluorescence-amplified tumour-imaging efficacy (e.g. iP10 versus iP5B5). Taken together, hydrophobicity of non-ionizable side-chain rather than pH_t of nanoparticle driven the enhanced tumour-imaging efficacy.

To validate the universality of hydrophobicity tuning on the enhanced cell uptake of nanoparticles, two tertiary amine-containing monomers, 2-(dipropylamino) ethyl methacrylate (nDPA-MA), and 2-(ethyl propylamine) ethyl methacrylate (EPA-MA) were copolymerized with BMA. The synthesized nP5B5 and EP5B5 had a pH_t of 5.15 and 5.83, respectively (Fig. 5g, h, and Supplementary Figs. 23 and 24). Results showed that the fluorescence signals of nP5B5 and EP5B5 were significantly improved in comparison to nP10 and EP10, respectively.

We next screened the chemical structures of non-ionizable monomers on the tumour-imaging efficacy. Various non-ionizable monomers with different number of alkyl carbon, including ethyl methacrylate (EMA), propyl methacrylate (PMA), butyl methacrylate (BMA), hexyl methacrylate (HMA), and octyl methacrylate (OMA) were copolymerized with iDPA monomer at a mole fraction of 50:50. The pH_t of the synthetic HUNPs decreased as the hydrophobicity of non-ionizable monomers increased (Supplementary Table 2). Fluorescence images of 4T1 tumour-bearing mice were captured at pre-designated time-points of post-injection of HUNPs at a dose of 20 mg kg⁻¹ (Supplementary Fig. 25). As shown in Fig. 5i, j, the activated fluorescent signals in tumour tissues exhibited a bell-shaped curve as the number of alkyl carbon in hydrophobic monomers increases. Tumour-imaging signals of HUNPs were enhanced gradually as the number of alkyl carbon in hydrophobic monomers increases from 0 to 4. However, iP5H5 and iP5O5 with pH_t values of 4.89 and 4.34 exhibited lower fluorescent signals than iP5B5. An ‘Always-On’ system (Supplementary Table 5) was further utilized to investigate the tumour accumulation of these two nanoprobes. We confirmed that iP5H5 displayed slightly higher tumour accumulation than iP5B5, while iP5O5 exhibited approximately twice the tumour accumulation of iP5B5 (Supplementary Fig. 26). Nevertheless, they still demonstrated reduced fluorescent signals compared to iP5B5, which probably due to the fact that pH_t values of these two nanoprobes are lower than the pH of most lysosomes in tumour cells, resulting in incomplete cellular activation. Therefore, iP5B5 was screened as the optimized nanoprobe for tumour-imaging applications.

The toxicity of iP5B5 was preliminarily evaluated in healthy mice. The mice were intravenously injected with iP5B5 at a dose of 300 mg kg⁻¹. We recorded the alterations in animal body weight during the 10-day period post-injection and performed histological examination of major organs at 1, 7, 11, 18 days post-administration. The results showed no weight loss after the injection of iP5B5 and the histology of the major organs was normal, which demonstrated the biosafety of iP5B5 (Supplementary Fig. 27).

For tumour-imaging application, we performed a parallel comparison between iP5B5-ICG and free ICG, which is extensive utilized in clinics for over sixty years. Mice bearing 4T1 tumours were injected with iP5B5-ICG and free ICG at an equivalent dose of 1 mg kg⁻¹ ICG. iP5B5-ICG exhibited strong and persistent tumour-specific signals over 48 h (Fig. 5k). In contrast, free ICG presented a poor tumour signal due to the rapid blood clearance, which only accounted for 1.1% of that in iP5B5-ICG group at 48 h post-injection (Supplementary Fig. 28). The TNR value was 23.80 ± 4.83 for iP5B5-ICG in tumours, which is 9.4-fold higher than that of free ICG (TNR = 2.54 ± 0.19), demonstrating its excellent imaging efficacy.

To investigate the distribution and internalization of iP5B5 in normal tissue and organs, we fabricated a binary ratiometric nanoreporter imaging system, which comprises of iP5B5-Cy5 and iP5B5-Cy7.5 (molar ratio, 9.4: 0.6) (Supplementary Table 8) as previously reported³¹. Imaging results showed that although the accumulation assessed by the ‘Always-On’ system indicated a higher level in the liver compared to the tumour, the internalization of nanoparticles in the tumour was more pronounced (Supplementary Fig. 29). This enhanced internalization in the tumour contributes to a high TNR contrast. To further understand this observation, we conducted cellular uptake experiments. At pH 7.4, the uptake of nanoparticles by 4T1 tumour cells was slightly higher than that of normal 3T3 cells, however, when we lowered the pH of culture media to 6.8, which mimics the acidic tumour microenvironment, there was a significant enhancement in nanoparticle uptake by tumour cells (Supplementary Fig. 30). Therefore, the long circulation of nanoparticles, the enhanced permeability and retention (EPR) effect in tumours, and the improved uptake of nanoparticles by cells in the tumour microenvironment collectively contribute to the high TNR contrast of iP5B5.

iP5B5-ICG effectively imaged and guided the resection of multiple tumours

Orthotopic 4T1 tumour-bearing mice model was used to evaluate the imaging time-window of iP5B5-ICG. The tumours were inoculated in the fifth mammary pad. After nanoprobe injection, fluorescent images of tumour-bearing mice were captured at pre-designated time-points over 72 h. Results showed that CNR in the mock surgical procedure during 12–72 h was greater than 100, demonstrating the long time-window for image-guided surgery (Supplementary Fig. 31). Furthermore, tumours with different sizes (30, 80, 150, 500 mm³) can be precisely visualized by iP5B5-ICG with CNR higher than 100 (Supplementary Fig. 32). We also constructed 4T1 tumour-bearing mice model at different positions, including the second mammary pad, subcutaneous tissue, and thigh muscle of mice (Supplementary Fig. 33). iP5B5-ICG effectively imaged all the tumours at different locations and guided the surgical removal of tumours (CNR: 99.67 ± 21.88 - 415.56 ± 70.33).

The imaging efficacy of iP5B5-ICG was also evaluated in 4 tumour-bearing mice models, including orthotopic MCF-7 breast cancer and KPC pancreatic cancer, subcutaneous CT26 and MC38 colorectal cancers. Bright tumour illumination was observed across all tumour types with high CNR (64.43 ± 7.12 −156.00 ± 39.59) during surgery at 24 h post-injection (Fig. 5l, and Supplementary Fig. 34). Tumour to muscle signal ratio in all the 4 tumour models were over 18.6, demonstrating the imaging capability of a broad range of tumours.

Incomplete surgical resection is one of the important reasons for tumour recurrence. Figure 6a, b exhibited the process of 4T1 tumour from incomplete resection to complete clearance of residual tumours under the guidance of iP5B5-ICG. The residual tumour tissue, a tiny spot on the muscle line, was difficult to distinguish by white light surgery, but is clearly visible as a bright spot under the IVIS system, based on which the complete removal of the residual tumour was achieved. H&E staining verified that the bright spot observed during in vivo imaging was tumour tissue, the size of which was <1 mm × 1 mm.

Fig. 6: iP5B5-ICG for image-guided tumour surgery.

a iP5B5-ICG guided complete resection of residual 4T1 tumour 24 h post-injection. b The histological analysis of residual tumours and adjacent tissues. Scale bar in the rightmost panel = 1 mm, in remaining three left panels = 100 µm. The zoom-in images of the insets 1, 2, and 3 in the rightmost panel are displayed in the three left panels. c Relative fluorescence intensity of MCF-7-GFP tumours and the adjacent muscles (n = 8 regions for tumour sites, and n = 8 regions adjacent to tumour sites for normal tissue sites, two-sided paired t tests). The solid line in the violin plots: mean value; thick dashed line: the median; thin dashed lines: the first and third quartiles. d Images of the resection process of the MCF-7-GFP tumours. e The relative fluorescence intensity of iP5B5-Cy5 of ten random areas of tumours and muscles in the slides of three samples (n = 10 random areas, two-sided unpaired t tests). The solid line in the violin plots: mean value; thick dashed line: the median; thin dashed lines: the first and third quartiles. f Three adjacent slides demonstrate the co-localization of iP5B5 with macrophages to outline the tumour margin. Scale bar in row 1 = 2.5 mm, in row 2 = 250 μm, in row 3 = 150 μm, in row 4 = 500 μm. The dashed line in the first row indicates the boundary of muscles. The zoom-in images of the insets 1, 2, and 3 in row 1 are shown in other three rows below. g In vivo imaging of mouse peritoneal metastatic tumours, yellow arrow indicates the colocalized signals of ICG and Luc. h Imaging of ex vivo tumours from g. i Relative fluorescence intensity of tumours and adjacent normal intestine tissues. The relative signal is normalized to the mean signal of the normal intestine tissues (n = 10 regions for tumour sites, and n = 10 regions adjacent to tumour sites for normal tissue sites, two-sided paired t-tests). All measurements are presented as mean ± s.d. Source data are provided as a Source Data file.

The MCF-7-GFP tumour model was established through intramuscular inoculation in thigh of mouse to investigate the capability of iP5B5 for the delineation of tumour margins. As shown in Fig. 6d, ICG channel exhibited a good co-localization with the GFP channel tumours. The signal intensity of the muscle tissues adjacent to the tumours in both channels was significantly lower than the tumour tissue (Fig. 6c). The signal intensity of the excised tumour was 6.1 ± 1.9-fold of the adjacent muscle, demonstrating the excellent capacity of iP5B5 to delineate tumour boundary (Supplementary Fig. 35). Tumours from mice injected with Cy5-conjugated iP5B5 were excised and three consecutive slides (10 μm) were dissected, one for H&E staining to determine the tumour margin, one for macrophages staining with anti-F4/80 antibody, and another for analyzing the distribution of iP5B5 in tumour tissues. The fluorescent signal of iP5B5 at the tumour margin displayed a good co-localization with the F4/80-labeled macrophages (Fig. 6f). Moreover, the fluorescent signal in the tumour area was much higher than that in the muscle area (Fig. 6e, f).

Finally, we evaluated the imaging efficacy of iP5B5 for the precise imaging of peritoneal disseminated tumours, which was established through intraperitoneal injection of 4T1-Luc tumour cells (1 × 10⁶) and allowing to growth for 5 days. After intravenous injection of iP5B5, mice were imaged under the IVIS system. The peritoneal metastatic tumours, including tiny lesions that are hard to recognize with the naked eyes, can be accurately identified through fluorescence imaging, which was consistent with the tumour locations mapped by bioluminescence imaging, thus demonstrating the high imaging sensitivity to tumours (Fig. 6g–i).

Discussion

The sensitive and accurate visualization of tumour margins is the prerequisite for effective tumour therapy by image-guided surgery. Currently, most molecular probes convert the concentration differences of biomarkers (e.g., HER2 and EGFR) between tumour and normal tissues into signal contrast, which is a linear imaging strategy^34,35. These targets are expressed in a small subset of tumours with high heterogeneity within and between diverse cancer types¹⁷. Thus, achieving broad imaging of diverse tumour types with high sensitivity and accuracy remains a big challenge.

For cancer visualization, the imaging efficacy of activatable nanoprobes mainly relies on the tumour accumulation and activation efficiency in cancerous tissues. We have demonstrated that UPS nanoprobes have a high accumulation of ~3% ID/g in tumour tissues¹⁷. Moreover, our recent work revealed that the first generation of UPS nanoprobes exhibit an internalization rates of ~20% in tumour regions³¹. The poor cellular internalization of UPS nanoprobes hinders the complete amplification of tumour signals. Moreover, inadequate intracellular internalization is a common reason why many nanomedicines fail to achieve enhanced therapeutic response³⁶. Currently, numerous reports have focused on the modifications of nanoparticle surface with antibodies, peptides, or other motifs that specifically recognize cell surface receptors, resulting in an enhanced cell uptake^37,38. Although, the surface chemistry of nanoparticles and their impact on nano-bio interactions have been widely recognized, our understanding on the contribution of core composition of nanoparticles to nanoparticle-cell interactions is still limited.

In our current work, we report the next generation of ultra-pH-sensitive nanoparticles (HUNPs) by introducing variable proportion of different non-ionizable monomers into the hydrophobic core of the nanoparticles, enabling precise modulation of their core composition and pH transitions. Surprisingly, we discovered that the increase in proportion of hydrophobic BMA monomer in the micellar core resulted in a significant enhancement in the cellular uptake of nanoparticles, which was caused by an order of magnitude increase in cell membrane binding of nanoparticles. Compared to iP10 that has a cellular internalization of ~15% in 4T1 tumours³¹, iP5B5 exhibited a more than 3-fold tumour internalization in vivo, which is up to 50%. We also found that the imaging efficacy of HUNPs exhibits a parabolic-shaped curve as the number (n = 0–8) of carbon atoms on the hydrophobic monomer R_n increases. The incorporation of BMA with four carbon atoms achieved the optimal tumour signal amplification via enhancing intracellular internalization. Very recently, Hammond et al. have reported that cancer cells distinguish nanoparticles based on the hydrophobic core composition by massively parallel pooled screening³⁹. Thus, the imaging efficacy of HUNPs was significantly amplified through orthogonal response to tumour accumulation and cellular internalization. Recently, optical probes in NIR-II region have shown great potential in biosensing and bioimaging applications⁴⁰. We will adapt the HUNPs nanotechnology to NIR-II dyes for precise imaging of solid tumours in the future.

In summary, we reported a series of hydrophobic core-tunable nanoprobes that achieve exponential signal amplification through the orthogonal effect of tumour accumulation and cellular internalization. Based on this strategy, the iP5B5-ICG nanoprobe has been screened and selected for the broad imaging of various tumour models and image-guided surgical resection. Our research enhances understanding of nanoparticle-cell interaction and highlights the potential of the second generation of UPS nanoparticles for drug/gene delivery.

Methods

Ethical statement

This research complies with all relevant ethical regulations. All animal studies were conducted in accordance with the National Institute Guide for the Care and Use of Laboratory Animals. The experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Peking University (Accreditation number: LA 2019039).

Materials

The non-ionizable monomers, including ethyl methacrylate (EMA) and butyl methacrylate (BMA) were purchased from Shanghai Macklin Biochemical Co., Ltd. Propyl methacrylate (PMA), hexyl methacrylate (HMA), octyl methacrylate (OMA), and HEMA were purchased from Alfa Aesar. The tertiary amine-containing monomers, including 2-(dipropylamino) ethyl methacrylate (nDPA-MA), 2-(ethylpropylamino) ethyl methacrylate (EPA-MA), 2-(dibutyl amino) ethyl methacrylate (DBA-MA) and PEG macroinitiator, mPEG5k-Br were synthesized in our lab^27,32. Monomers 2-(diisopropyl amino) ethyl methacrylate (iDPA-MA) and 2-aminoethyl methacrylate (AMA) were obtained from Polyscience Company. The N-hydroxysuccinimidyl (NHS) esters of fluorescent dyes were purchased as follows: Bodipy FL-NHS and Cy5-NHS were from Lumiprobe Corporation; ICG-NHS was purchased from AAT Bioquest. All the solvents and reagents were used as received from J&K Scientific. Hoechst 33342 was obtained from Invitrogen Inc. The anti-F4/80 antibody (Cat. NO. ab6640, clone number: CI:A3-1, Dilution: 1:200), anti-LAMP1 antibody (Cat. NO. ab208943, clone number: EPR21026, Dilution: 1:400), AF-488-conjugated Goat anti-Rabbit (Cat. No. ab150077, Dilution:1:1000) secondary antibody and AF-594-conjugated Goat anti-Rat (Cat. No. ab150160, Dilution: 1:1000) secondary antibody were purchased from Abcam (Shanghai, China).

Cell culture

4T1 (3101MOUSCSP5056) and MCF-7 breast cancer cells (3101HUMSCSP531), CT26 (1101MOU-PUMC000275), PANC02 pancreatic cancer cell (CRL-2553), and A549 lung carcinoma (1101HUM-PUMC000002) were obtained from the National Infrastructure of Cell Line Resource. KPC pancreatic cell, was a gift from Dr. Feng Qian of School of Pharmaceutical Sciences, Tsinghua University. ID8 epithelial ovarian cancer cell, was a gift from Dr. Yi Li of Peking University People’s Hospital. MC38 colorectal cancer cells were obtained from Jinming Gao lab. The aforementioned cell lines were kept in-complete media (RPMI 1640 or DMEM with 10% FBS and antibiotics) in a humidified atmosphere that contained 5% CO₂ at a temperature of 37 °C. RAW264.7 macrophage (TIB-71) was obtained from the National Infrastructure of Cell Line Resource. The immortalized bone marrow-derived macrophage (iBMDM) was a gift from Dr. Feng Shao of National Institute of Biological Sciences, Beijing. These cell lines were kept in a heat-inactivated culture medium (DMEM with 10% heat-inactivated FBS and antibiotics) in a humidified atmosphere that contained 5% CO₂ at a temperature of 37 °C. In addition, murine bone marrow-derived macrophages (BMDM) were obtained by isolating bone marrow cells from surgically resected femur and tibia of C57BL/6 female mice (6–8 weeks). M0-like BMDMs were cultured in the complete heat-inactivated DMEM medium containing recombinant murine macrophage CSF (m-CSF, 20 ng mL⁻¹, PeproTech) for 5 days. M0-like BMDMs were stimulated with IL-4 (20 ng mL⁻¹, BIOplastics) for 24 h to obtain M1-like BMDMs or M2-like phenotypes, respectively. Short tandem repeat (STR) DNA finger printing or morphological features were applied to authenticate and identify cell lines used in this study. All cell lines were tested for mycoplasma contamination routinely. No mycoplasma contamination was found.

Animal models

All treatments and care for animals were carried out in accordance with ethical standards and with the approval of the Animal Ethics Committee of Peking University. Female BALB/c (6–8 weeks, 18–22 g), C57BL/6 (6–8 weeks, 18–22 g), nu/nu nude (18–22 g) and NOD-SCID (18–22 g) mice were purchased from Vital River Laboratory Animal Centre (Beijing, China). The animals were housed in a humidity-controlled environment at 25 °C with free access to food and water in a 12 h dark/light cycle. 4T1 tumour-bearing BALB/c mice model was established by subcutaneous injection of tumour cell suspension (1 × 10⁶ cells/animal) into the fifth mammary fat pad. CT26 and MC38 colon tumour models were created to verify the universality of HUNPs for enhanced imaging efficacy. Mice bearing 4T1 tumours at different locations through intramuscular, subcutaneous, and second right mammary pad inoculations were established to validate the capacity of iP5B5 for precise tumour imaging. In addition, an orthotopic KPC pancreatic cancer model in C57BL/6 mice, an MCF-7 breast cancer model at the fifth pair of mammary pads in NOD-SCID mice, a subcutaneous CT26 model in BALB/c mice, and an MC38 model in C57BL/6 mice were also established to verify the broad application of iP5B5 for tumour imaging. MCF-7-GFP tumour model was established to verify the ability of iP5B5 for the delineation of tumour margins. To establish an intraperitoneal disseminated tumour model, 1 × 10⁶ 4T1-Luc cells were injected into the abdominal cavities of BALB/c mice. In the above model, tumors with volumes reaching 150–200 mm³ were used for cancer imaging studies. When exploring the imaging effects of HUNPs on tumors of different volumes, the largest tumor volume was 500–600 mm³. For tumour-imaging studies, the maximal tumour size of 1500 mm³ was permitted by IACUC of Peking University. We confirm that the maximal tumour size was not exceeded in any of the experiments.

Syntheses of HUNPs copolymers

All HUNPs copolymers were synthesized by atom transfer radical polymerization (ATRP) method³². Take the copolymer mPEG₁₁₄-b-P(DPA50-BMA50) as an example, which was synthesized by the following procedure: a schlenk flask was charged with mPEG₁₁₄-Br (0.1 mmol), iDPA-MA (5 mmol), BMA-MA (5 mmol), AMA (0.3 mmol), and PMDETA (0.1 mmol). Then, the 2-propanol (2 mL) and N’N-dimethylformamide (DMF, 2 mL) mixed solvents were added into the flask to dissolve the aforementioned compounds. After three cycles of freeze-pump-thaw to remove oxygen, CuBr (0.1 mmol) was introduced into the reaction flask under a nitrogen environment, and the reaction was maintained under vacuum for 12 h at 40 °C. Tetrahydrofuran (10 mL) was then added to the reaction mixture, and a neutral aluminum oxide column was used to remove the catalyst. Most of the eluent was removed by rotovap. The residue was dialyzed against distilled water for 24 h and then lyophilized to obtain white powder. The resultant polymers were characterized by ¹H NMR (Bruker, 400 MHz).

Synthesis of dye-conjugated copolymers

In 2 mL of anhydrous DMF, 10 mg of AMA-containing copolymers were dissolved. The dye-NHS esters were then added to the reaction tube. The ratios of dyes to AMA were listed in Supplementary Tables 2, 3, 5–8. To completely remove free dyes, the copolymers were purified by dialysis (MWCO = 3500) against DMF after 24 hours, followed by dialysis against Milli-Q water for DMF removal. The resultant copolymers with dye conjugations were lyophilized and stored at −20 °C for additional research. UV-Vis spectrophotometry was used to calculate the degree of dye conjugation to each polymer.

Preparation and characterization of nanoparticles

The nanoparticles have been prepared using a reported method³². As an illustration, 1 mL of methanol was used to dissolve 5 mg of ICG-conjugated mPEG₁₁₄-b-P(iDPA50-BMA50) polymer. Then, the solution was added dropwise into 4 mL of Milli-Q water under sonication. Using a micro-ultrafiltration unit, the mixture was ultra-centrifuged four times to remove the organic solvent (MWCO = 100 kDa). The resultant polymeric nanoparticle was then adjusted to 5 mg mL⁻¹ with Milli-Q water.

The hydrodynamic diameter of nanoparticles (1 mg mL⁻¹) at pH 7.4 and 4.8 was measured by Zetasizer NanoZSP (Malvern Instruments) at room temperature. The stiffness of nanoparticles was measured by atomic force microscope (Bruker). To prove the core-shell structure of HUNPs, mPEG₁₁₄-b-P(iDPA50-BMA50) and mPEG₁₁₄-b-iPDPA100 copolymers (20 mg) were first dissolved in 1.0 mL of 50 mM deuterated chloride (DCl) solution. pH titration was performed by adding 1 M sodium deuteroxide (NaOD) solution dropwise under stirring. Solutions of copolymers with protonation degrees of 0 and 100% were obtained for ¹H NMR analysis on a 400 MHz NMR Spectrometer at room temperature.

Nomenclature of nanoparticles

For ease of description, the polymers are named using abbreviations of tertiary amine-containing monomers/non-ionized monomers and their proportions. The naming sequence is as follows: the abbreviation of tertiary amine-containing monomers/non-ionized monomers + the repeating units of those monomers divided by ten. For example, mPEG₁₁₄-b-iPDPA100 is named iP10; mPEG₁₁₄-b-P(iDPA90-BMA10) is named iP9B1; mPEG₁₁₄-b-P(iDPA70-HEMA30) is named iP7HE3; mPEG₁₁₄-b-PEPA100 is named EP10; PEG-PDBA100 is named DB10.

pH-responsiveness of nanoparticles

For each nanoprobe, the sample (5 mg ml⁻¹) was prepared in Milli-Q water. Then, the solution was diluted in 0.1 M phosphate-buffered saline (PBS) buffer with different pH values to 50 μg mL⁻¹. The nanoprobe conjugated with ICG was excited at 780 nm, and the emission spectra were collected from 795 to 900 nm. The emission intensity at 810 nm was used to quantify the signal amplification. To quantify the pH sensitivity, the concept of “Normalized fluorescence intensity” (NFI) is introduced, and its calculation formula is as follows:

$${{{\rm{NFI}}}}=\frac{F-{F}_{\min }}{{F}_{\max }-{F}_{\min }}$$

(1)

where F is the fluorescence intensity measured at any pH value, F_min and F_max are the minimum and maximum fluorescence intensity under OFF and ON conditions, respectively.

NFI values as a function of pH values curves were plotted and fitted using Origin software. On the fitted curve, the pH range corresponding to the increase of NFI from 10% to 90% is defined as ΔpH_10–90%. The transition pH (pH_t) of nanoprobes were calculated as the pH value at which the pH-NFI curve reaches its maximum derivative.

Plasma stability of HUNPs

Fresh plasma was firstly collected from mice. To evaluate the stability of the nanoprobes, iP5B5-ICG was selected an example. The iP5B5-ICG nanoprobe (4 mg mL⁻¹) with 10 μL was incubated with 390 μL of PBS (pH 7.4 or 4.8), or plasma at 37 °C in a humidified chamber for 24 hours. At each time-point, the solutions were immediately measured on a Hitachi fluorometer (F-7000 model, Tokyo, Japan) to evaluate the changes of fluorescent signals.

Preparation and characterization of the ‘Always-ON’ nanoprobes

The Always-On nanoprobes were prepared using our reported procedure³¹. When the number of fluorescent molecules in the nanoparticle core decreases to a certain extent, the distance between dye molecules increases, the homo-FRET quenching between the dye molecules abolished, and the fluorescence signal can be effectively emitted. Therefore, maintaining a low number of dye molecules in the core of nanoparticles is crucial for achieving ‘Always-On’ nanoprobe. To achieve this, we synthesized dye-conjugated copolymers with an average of 0.5 probe per polymer. Additionally, we have meticulously balanced the mixture of the blank copolymer and the Cy5-labeled copolymer in specific proportions (90–95% for the blank copolymer and 5–10% for the Cy5-labeled copolymer) to achieve an ‘Always-On’ goal. Take 90% blank copolymer as an example, the dye-conjugated polymers and corresponding blank polymers at a molar ratio of 1 to 9 were dissolved in methanol, and the mixture was added into Milli-Q water dropwise. Then, the solvent was removed by ultrafiltration. The prepared micelles were diluted to a concentration of 100 μg mL⁻¹ in 0.1 M phosphate-buffered saline (PBS) buffer with pH values of 7.4 and 4.8. The emission spectra of the nanoprobes from 645 to 800 nm were collected at an excitation of 630 nm. The F_On/F_Off ratio should be within 0.8–1.25.

Cell trafficking analysis of HUNPs

To investigate the cell trafficking and signal activation of nanoprobes, the lysosome was labeled with RAB5A-GFP (a specific early endosome membrane-bound protein) or LAMP1-GFP (a specific lysosomal membrane-bound protein). The A549 cells were genetically transfected with RAB5A-GFP or LAMP1-GFP. The cells were precooled on ice for 10 min, pulsed with HUNPs solution (100 μg mL⁻¹) at 4 °C for 10 min, washed thrice with ice-cold PBS. And the cells were placed in a cell culture incubator. Images were captured using a confocal microscope at 30 minutes, 1 h, 2 h, 4 h and 6 h time-points. The analysis involved assessing the signal intensity of fluorescent nanoprobes and their subcellular localization within organelles.

Kinetics study of HUNPs internalization in vitro

To investigate the kinetics of HUNPs internalization, we utilized a binary ratiometric nanoreporter system composed of a BDP-labeled ‘On-Off’ module and a Cy5-labeled ‘Always-On’ module (molar ratio: 9.5:0.5) to investigate the kinetics of the nanoparticle internalization process. To study the kinetics of internalization, the cell suspension obtained by trypsinization was placed on ice for 10 minutes, followed by centrifugation and resuspension in the binary ratiometric nanoreporter system (100 μg mL⁻¹) at 4 °C. After being pulsed for 10 minutes (cell binding) and washed three times, the samples were chased at 37 °C (internalization) by measuring the fluorescent intensity of each channel for single cells on a flow cytometer.

Nanoparticle binding on the cell membranes

4T1 cells were selected as an example. 4T1 cells were seeded at a density of 1 × 10⁴ cells/well in a 24-well plate and cultured until the confluence reached 60–70%. The cells were placed on ice for 5 minutes and then treated with Cy5-labeled nanoprobes at a concentration of 50 μg mL⁻¹ for 10 min on ice. The nanoprobes were removed and cells were cultured for another 4 hours for complete signal activation. After trypsinization with 0.25% EDTA, the cells were transferred to a 1.5 mL EP tube, centrifuged at 1000 × g for 3 minutes, and the supernatant was removed. The cells were washed with serum-free culture medium without phenol red. The cells were then resuspended in 200 μL PBS, and the fluorescence intensity in the APC channel was detected by flow cytometry. 5000–10000 cells were collected, and the data were analyzed using FlowJo or Cytexpert software.

Cellular uptake of different nanoprobes

4T1 cells were chosen as an example. 4T1 cells were seeded at a density of 1 × 10⁴ cells/well in a 24-well plate and cultured until the confluence reached 60–70%. After removal of the culture medium, 200 μL of nanoprobes solution (50 μg mL⁻¹) diluted with cell culture medium was added. After 4 hours incubation, the culture medium was replaced with fresh complete medium, and the cells were further incubated for 4 hours for full activation of fluorescent signal. After cell digestion and resuspension, the fluorescence intensity in the APC channel was detected using a flow cytometer. 5000–10,000 cells were collected, and the data were analyzed using FlowJo or Cytexpert software. The treatment for other cells was performed using the same procedure described as above.

To image the nanoparticle signal in lysosomes, tumour cells were treated with Cy5-labeled nanoprobes (50 μg/mL) for 4 hours. Then, the culture medium was replaced with fresh complete medium, and the cells were incubated for another 4 hours. Lysotracker green was used to label lysosomes before imaging. Confocal pictures were captured with a Nikon confocal microscope (AXR, Nikon, Japan) equipped with a ×60 lens. The signals were captured as the following conditions: Hoechst 33342 (λ_ex/em: 405/466 nm), GFP (λ_ex/em: 488/515 nm), and Cy5 (λ_ex/em: 647/650 nm). The images were analyzed using ImageJ software.

Analysis of cellular uptake pathway of HUNPs

4T1 cells were seeded at a density of 1 × 10⁴ cells/well in a 24-well plate and cultured until the confluence reached 60–70%. Prior to co-incubation with the inhibitor containing HUNPs-Cy5 nanoprobe solution, the cells were pre-incubated for 1 hour with cell culture medium containing specific inhibitors, including chlorpromazine (3 mM) for clathrin-mediated endocytosis, dynasore (8 mM) for dynamin-dependent pathway, EIPA (4 mM) for macropinocytosis, nystatin (3 mM) for caveolae-mediated endocytosis, and genistein (10 mM) as a protein tyrosine kinase inhibitor. The HUNPs-Cy5 nanoprobe solution was introduced to the cells at a final concentration of 50 μg mL⁻¹, and a co-incubation period of 4 hours was maintained. Subsequently, the nanoprobe was meticulously washed away, ensuring complete removal, and the cells were further incubated for an additional 4 hours to ensure the full activation of the nanoparticle signals. The intracellular Cy5 intensity was quantified using flow cytometry, and a comparative analysis was performed against the signal intensity exhibited by the non-inhibited control group.

The uptake of iP5B5 between tumour cells and normal cells at different pH values

The pH of the culture medium was adjusted to 6.8 using a 1 M monohydrate citric acid and 1 M HEPES solution. 4T1 cells and 3T3 cells were seeded at a density of 1 × 10⁴ cells/well in a 24-well plate and cultured until the confluence reached 60–70%. After removal of the culture medium, 200 μL of iP5B5-Cy5 solution (50 μg/mL) diluted with cell culture medium of different pH was added. After 4 hours incubation, the culture medium was replaced with fresh complete medium, and the cells were further incubated for 4 hours for full activation of fluorescent signal. After cell digestion and resuspension, the fluorescence intensity in the APC channel was detected using a flow cytometer. 10,000 cells were collected, and the data were analyzed using FlowJo or Cytexpert software.

In vivo near-infrared fluorescence cancer imaging

UPS-ICG or HUNPs-ICG nanoprobes at a concentration of 20 mg kg⁻¹ were injected into the 4T1 tumour-bearing mice (n = 3 for each nanoprobe). Time-dependent fluorescence images were captured by IVIS in vivo imaging system using an ICG filter. After imaging at 48 h post-injection, the mice were euthanized and a mock surgery was performed. The surgical process was captured by the IVIS system. The major organs were gathered for fluorescent images acquisition at 48 h post-injection. The fluorescence intensity was analyzed by the IVIS software. The following equation was used to calculate the contrast-to-noise ratios (CNR):

$${CNR}=\frac{{FI}\left(T{umour}\right)-{FI}({Normal})}{S.D.({Normal}t{issue})}$$

(2)

The fluorescence intensity of tumours and normal tissues is represented in this formula by the letters FI (Tumour) and FI (Normal), respectively. S.D. (Normal tissue), which is derived as the standard deviation of the normal tissue fluorescence, is used to present background noise. To calculate the tumour-to-normal tissue ratio (TNR), the mean fluorescence intensity of the tumour was divided by the mean fluorescence intensity of adjacent healthy tissues.

Image-guided tumour resection

To further evaluate the accuracy of image-guided surgery of iP5B5-ICG, MCF-7-GFP tumour model was established on the right flank of mice. A mock surgery was performed at 24 hours post-injection of the iP5B5-ICG. ICG and GFP signals were collected by IVIS Lumina Series III imaging system. After imaging, the co-localization of nanoprobes and GFP signals was analyzed. ICG signal intensity in tumour areas (with GFP signal) and non-tumour areas (without GFP signal) were collected in IVIS software for subsequent analysis. After image-guided tumour surgery, the resected tumours and adjacent normal tissues were fixed with 4% paraformaldehyde for histological examination.

In vivo imaging of peritoneal disseminated tumours

Seven days after intraperitoneal inoculation of 4T1-GFP tumour cells, iP5B5-ICG nanoprobe was intravenously injected at a dose of 20 mg kg⁻¹. Before the operation, fluorescent enzyme substrate D-Luciferin was intraperitoneally injected at a dose of 150 mg kg⁻¹. The surgery procedure began 5–10 minutes after injection. The mouse was dissected, and imaged for ICG and Luc signals using the IVIS Lumina Series III live imaging system. The intestine was then collected and the ICG and Luc signals of the disseminated tumour nodules and adjacent intestine were analyzed.

HUNPs distribution in tumour tissues by immunofluorescence analysis

4T1-bearing mice were administered with Cy5-labeled HUNPs (20 mg kg⁻¹) via tail vein injection. The mice were euthanized at 48 h post-injection, and the tumours were dissected and snap frozen for tissue sectioning. The sections were fixed with 4% paraformaldehyde solution for 15 min and then were permeabilized with 0.1% Triton X-100 solution for 5 min. After washing with PBS, the samples were blocked with BSA solution for 1 hour and incubated with anti-LAMP1 (1:100) antibodies. After incubation and washing 3 times with PBS, an AF-488-labeled secondary antibody (Goat anti-rabbit 1:400) was added. The fluorescence images of tumour slides were collected by confocal microscope (Nikon, AXR). Hoechst 33342, AF-488, and Cy5 channels were captured. The images were exported using NIS Viewer and ImageJ software.

In vivo pharmacokinetics (PK) studies of HUNPs

Fifteen BALB/c mice (18–22 g) were randomly divided into five groups (n = 3 biologically independent mice for each group) to assess the in vivo pharmacokinetics of HUNPs. The mice were intravenously administered with iP10-ICG, iP7HE3-ICG, iP9B1-ICG, iP7B3-ICG, and iP5B5-ICG nanoprobes. After intravenous injection, blood sample (25 μL) was withdrawn at the following intervals: 2 min, 15 min, 0.5, 1, 3, 6, 12, and 24 h. Collected plasma (10 μL) was diluted with 200 μL of acidic methanol, and then centrifuged at 10,000 g for 10 min. Fluorescence signals of supernatant were determined by Odyssey DLx imaging system (Licor, USA) at 800 nm channel.

Evaluation of tumour imaging and nanoparticle internalization with a binary ratiometric system

ICG-labeled nanoparticle and Cy5-labeled nanoparticle were encoded as ‘On-Off’ and ‘Always-ON’ modules for in vivo fluorescence cancer imaging, respectively. The cocktail of the two modules (On-Off: 20 mg kg⁻¹; Always-ON: 40 mg kg⁻¹) was intravenously injected into the 4T1 tumour-bearing mice (n = 3 biologically independent mice per group) to monitor the tumour accumulation and internalization of nanoparticles. Time-dependent fluorescence images were captured on the IVIS in vivo imaging system using ICG and Cy5 filters, respectively. Fluorescence imaging was also performed on CT26 and MC38 tumour-bearing mice to further probe the internalization capacity of HUNPs. The Cy5 and ICG channel images were unmixed using Living Image Software, and the ratiometric images were created by ImageJ software (NIH).

To evaluate the nanoparticle internalization, BDP-labeled nanoparticle was encoded as the ‘On-Off’ module, and Cy5-labeled nanoparticle was encoded as the ‘Always-ON’ module. 4T1 tumour-bearing mice were administered with a cocktail of the two modules (O–Off: 20 mg kg⁻¹; Always-ON: 40 mg kg⁻¹) via tail vein injection. The mice were euthanized at 24 h post-injection of the nanoprobes. After fixation with 4% paraformaldehyde for 15 min and staining with Hoechst 33342, fluorescence images of whole tumour slides were collected in Hoechst 33342, BDP, and Cy5 channels by quantitative slide scanner (Phenochart 1.0.8, PerkinElmer Vectra Polaris). The images were exported using the Phenochart software and the ratiometric images were created by ImageJ software (NIH).

Evaluation of accumulation and internalization of iP5B5 with a binary ratiometric system

4T1-bearing mice were administered with the binary ratiometric nanoreporter of iP5B5 (Cy5, On-Off; Cy7.5, Always-On, molar ratio = 9.4: 0.6) (20 mg kg⁻¹) via tail vein injection. The mice were euthanized at 48 h post-injection. The heart was perfused until the red color faded from the liver. The heart, liver, spleen, lungs, kidneys, tumour, and normal muscle tissues were dissected. Imaging was performed with the IVIS. The acquired data were imported into the Living Image software, where regions of interest (ROIs) were drawn around each organ/tissue, and the intensity was measured and the ratio values were calculated.

Co-localization of iP5B5 with macrophages at tumour margin

MCF-7-GFP tumour-bearing mice model was established through intramuscular inoculation. Cy5-labeled iP5B5 was injected into mice at a dosage of 20 mg kg⁻¹. At 24 h post-administration, the tumour and adjacent muscle were dissected, and sliced into three adjacent sections. Slide one was used for histological examination, slide two was used to analyze the distribution of nanoprobes in GFP-labeled tumour cells, and slide three was used to analyze the distribution of nanoprobes in F4/80-labeled macrophages, which were stained with AF-594 secondary antibodies. The fluorescence images of whole tumour slides were collected by quantitative slide scanner (Phenochart 1.0.8, PerkinElmer Vectra Polaris). Hoechst 33342, AF-488, and Cy5 images were captured in slide two, and AF-594 channel for slide three was captured to avoid fluorescence cross-talk.

The Phenochart was used to export images of the slices. The images were then split by ImageJ software. Three samples were obtained, and ten regions were randomly collected from the tumour and muscle areas of each sample to analyze the differences in fluorescence signals between the two regions. Additionally, the co-localization between the nanoprobe and macrophages at the tumour margin was analyzed.

In vitro cytotoxicity of HUNPs

4T1 cells were cultivated in 96-well plates at a density of 3000 cells/well and incubated for 12 hours. The cells were exposed to increasing concentrations of polymeric nanoparticles for 48 hours. MTT solution (0.5 mg/mL) was added to the cells and incubated for 2 hours. The produced formazan was then dissolved in 200 L of DMSO after media removal. To assess polymer dose-response relationships, the OD values were calculated at 540 nm using a microplate reader (Thermo Fisher Scientific, USA).

In vivo toxicity evaluation of HUNPs

Thirty BALB/c mice (18–22 g) were randomly divided into six groups (n = 3 biologically independent mice per group) to assess the toxicity of HUNPs at an imaging dose (20 mg kg⁻¹). The body weight of the mice was monitored every day. At 24 h post-injection, the major organs were collected and immediately fixed in 4% paraformaldehyde for histological examination, including the heart, liver, spleen, lung, and kidney. In addition, iP5B5-ICG nanoprobes were intravenously administered to healthy BALB/c mice at a dose of 300 mg kg⁻¹. The body weight of the mice was monitored every day. At 24 h, 7 days, 11 days, and 18 days post-injection, the major organs were collected and immediately fixed in 4% paraformaldehyde for histological examination, including the heart, liver, spleen, lung, and kidney.

Statistics and reproducibility

Data were expressed as mean ± s.d. Statistical analyses were performed using GraphPad Prism. Significant differences of quantitative data were analyzed by t test or analysis of variance (ANOVA) post Brown-Forsythe test depending on the number of treatment groups and distribution. For time-dependent data, the statistical significance of the differences in the data for each group at the last time-point was assessed using specific tests for different figures. Specifically, the data for each group at the last time-point in Fig. 2f, g, Fig. 3b, e, Supplementary Figs. 5, 8b, c, 11, 20d, 21b, 25d, 26b, 32b, c, 33b, c were analyzed using the one-way ANOVA post Brown-Forsythe test, followed by Tukey’s multiple comparisons tests. Supplementary Figs. 13b–d, 14b–d, 23d, 24d, 27b, 28a, and 28b were analyzed using the unpaired t-test to evaluate the significance of differences among groups at the last time-point. Figure 6c, i, and Supplementary Fig. 35b were analyzed using the paired t-test to assess the differences between groups.

For the whole-mount imaging analysis of HUNPs distribution and internalization in tumour slides (Fig. 3g and Supplementary Figs. 7 and 12), the experiment was repeated thrice with similar results, and a series of representative images were shown. Confocal imaging of the uptake of HUNPs by 4T1 (Fig. 4f) or A549 cells (Supplementary Fig. 16) was repeated at least three times with similar results, and a series of representative images from each group were shown. The experiment of iP5B5-ICG guiding the resection of residual tumour in Fig. 6a, b was repeated three times using biologically independent mice with similar results, and a series of representative images were shown. The experiment of iP5B5-ICG delineating tumour margin in MCF-7-GFP tumour model was repeated three times using biologically independent mice with similar results (Fig. 6d), and a series of representative images were shown. For the whole-mount imaging analysis of the co-localization of iP5B5 with macrophages (Fig. 6f), the experiment was repeated thrice with similar results, and a series of representative images were shown. For histological analysis of major organs from mice after injection of HUNPs (Supplementary Fig. 8d and Supplementary Fig. 27c), the experiment was repeated thrice with similar results, and a series of representative images were shown.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Source data

Source Data