Nanoparticle-Based Combination Therapy Enhances Fulvestrant Efficacy and Overcomes Tumor Resistance in ER-Positive Breast Cancer

Breast cancer remains the leading cause of death in women, globally. In particular, estrogen receptor (ER)–positive breast cancer has the highest incidence, accounting for over 70% of all cases (1, 2). Endocrine therapy is recommended as the first-line treatment regimen for ER-positive breast cancer, except in patients with extensive, life-threatening metastases (3–5). Among the endocrine therapies, fulvestrant, a selective estrogen receptor degrader, is commonly used for extended disease control in patients with advanced and metastatic breast cancer who have developed resistance to other antiestrogen therapies, such as tamoxifen. Fulvestrant competitively binds to the ER to downregulate the receptor with a much greater affinity than tamoxifen—approximately 89% that of estradiol, compared with 2.5% for tamoxifen (6, 7). Specially, unlike tamoxifen or other selective ER modulators, fulvestrant has no agonist effects and is not cross-resistant with other endocrine agents in clinical studies (8–10). With these unique modes of action of fulvestrant, the drug has clear and significant clinical value, offering an important therapeutic option, particularly for those patients with advanced disease requiring more intense treatment options (11–13). However, the poor solubility, invasive muscle administration, low drug bioavailability, and common side effects (e.g., pain, vomiting and hair loss) limit its broad application (9, 14, 15). In addition, the development of compensatory drug resistance associated with long-term administration makes the treatment prohibitively challenging to achieve complete management of tumor progression (16, 17).

Nanocarriers have enormous potential for the efficient and precise delivery of therapeutic drugs into tumor tissues, preventing rapid immune clearance and biodegradation, improving circulation time and bioavailability, ultimately improving efficacy without the need for increased doses (18–22). Several types of nanocarriers have been introduced for drug delivery in cancer treatment, with some achieving FDA approval, including Doxil (liposomal doxorubicin) in 1995 and Onivyde (liposomal irinotecan) in 2015, both of which show improved therapeutic effects relative to the respective free drugs (23, 24).

To this end, to stay the continual application of fulvestrant, in this study we designed a micelle nanocarrier to encapsulate fulvestrant to optimize its administration route and improve therapeutic outcomes. We also coencapsulated an inhibitor of cyclin-dependent kinases 4 and 6 (CDK4/6), abemaciclib (that can prevent DNA synthesis to block progression of the cell cycle from G₁- to S-phase) in the nanocarrier, to overcome drug resistance resulting from extended fulvestrant treatment. Numerous clinical trials have suggested that the patients on fulvestrant whose cancer progresses retain sensitive to subsequent abemaciclib therapy (25, 26). It also stands to reason that a combination of fulvestrant with abemaciclib in one nanoplatform may enhance response rates by circumventing the hurdles faced by sequential combination therapy wherein suboptimal scheduling and dosing may lead to antagonism between the two drugs (27–31). Our nanoformulation was fabricated from an amphiphilic chimeric poly(ethylene glycol) (PEG)-poly(β-amino ester) block copolymer macromolecule (PPAE) that could self-assemble into approximately 108-nm-sized micelle nanoparticles (NP) incorporating two drugs (illustrated in Fig. 1A). We surface modified the NPs with an active targeting cRGD peptide to direct them to tumor tissue and release drugs in a stable, controlled manner (32–34). We present data that show that the resultant biocompatible, hydrophilic nanoformulation, termed PPFA-cRGD, exhibited in vitro organoid targeting and in vivo tumor accumulation with good pharmacokinetics, after intravenous administration. Further it exhibited excellent antitumor efficacy with low toxicity in multiple ER-positive murine breast tumor animal models (Fig. 1B). Furthermore, the PPAE material is easily and cheaply obtained, thus enabling the scale-up of NP synthesis and reproducible characterization. Overall, the PPFA-cRGD demonstrates promising clinical application potential for patients with ER-positive breast cancer with improved therapeutic efficacy and good tolerability.

The primary objective of this study was to develop biocompatible NPs that respond to a weakly acidic environment for targeted delivery of fulvestrant (an ER antagonist) and abemaciclib (an inhibitor of CDK4/6) to tumor sites, and enhancement of the clinical efficacy of the drug in ER-positive breast cancer. For in vitro experiments, NP characterization, cell viability and targeting analysis, experiments examining tumor organoid uptake, as well as other effects, were performed at least three times, independently. To establish the breast cancer organoid and breast cancer patient-derived tumor xenograft (PDX) models, surgically resected tumor specimens were obtained from the Fourth Hospital of Hebei Medical University (Shijiazhuang, Hebei, P.R. China). Each sample was confirmed to be tumor tissue based on pathologic evaluation. Written informed consent for the research use of tissues from patients with breast cancer was obtained prior to the acquisition of human specimens and the protocol was approved by both the Hebei Medical University Fourth Affiliated Hospital Ethics Committee and the National Center for Nanoscience and Technology Institutional Review Board.

All in vivo experiments were performed using 4 to 6 weeks old BALB/c or BALB/c nude mice that were purchased from Vital River. The mice were randomized into the different treatment groups. The tumor-targeting ability of NPs was evaluated by IVIS imaging in MCF7 tumor-bearing BALB/c nude mice (n = 3 mice per group). The antitumor efficacy of NPs was evaluated in the MCF7, HCC1428, and ZR75-1 breast cancer xenograft models and a PDX model (n = 6 mice per group). The safety assessment of the NPs was evaluated in healthy mice (n = 6 mice per group) and Bama miniature pigs (n = 3 pigs per group). The mice and miniature pigs were monitored daily and euthanized at defined, humane endpoints. The tissue samples from these studies were collected and fixed for hematoxylin and eosin (H&E) and IHC staining or other experiments. All animal experiments were performed according to protocols approved by The National Center for Nanoscience and Technology Animal Care and Use Committee.

Fulvestrant and abemaciclib were obtained from Selleck. mPEG-PAE and cRGD peptides were synthesized by Ruixibio. RPMI1640, DMEM, and PBS were acquired from Wisent Corporation. The culture dishes and confocal microscopy dishes were purchased from Corning. The human breast cancer cell lines, MCF7 (RRID: CVCL_0031), ZR75-1 (RRID: CVCL_0588), and HCC1428 (RRID: CVCL_1252) were acquired from the ATCC. All cell lines are verified and periodically tested for Mycoplasma via the TransGen Biotech. DAPI and Cy5.5 were purchased from Solarbio Science & Technology Co. Ltd. The cell counting kit-8 (CCK-8) was brought from Dojindo and the Annexin V/propidium iodide apoptosis kit was purchased from BioLegend.

A Zetasizer Nano ZS and HT7700 transmission electron microscope were applied for the estimation of size, distribution, and surface properties of NPs. An LC-20AT HPLC was used for quantification of the drug concentrations. A Leica TCS SP8 MP confocal microscope was used to capture fluorescence images. A BD Accuri C6 flow cytometer (Becton Dickinson) was used to determine the targeting ability of NPs and apoptosis status of MCF7 cells. In vivo and ex vivo fluorescence imaging was performed on a PerkinElmer IVIS Spectrum imaging system. Absorbance was read using a Synergy HTX Multi-Mode Reader (BioTek) at 405 nm in CCK-8 assay.

Briefly, 60 mg PPAE-cRGD was dissolved in 2 mL methylene chloride. Next, 3 mg fulvestrant and 6 mg abemaciclib were added to the solution, which was combined with 7.5 mL deionized water. Emulsification was generated by ultrasonication for 1 minute at 100 W using a probe ultrasonicator (NingBo Scientz). Organic solvent was then removed by rotary evaporation for thirty minutes to acquire drug-loaded NPs (PPFA-cRGD). Afterward, the NPs were collected by centrifugation at 15,000 × g for 5 minutes and washed three times with water. PPFA-cRGD was sterilized through 0.22 μm filters before use. The other nanoformulations were also prepared following the above method.

NP samples (PPAE-cRGD, PPFA-mRGD, or PPFA-cRGD) were diluted with pure water and examined on Zetasizer Nano ZS for size distribution and surface charge analysis at 25°C. For morphologic characterization, NPs were deposited on a carbon-coated copper grid after dilution. All samples were negatively stained with 2% uranyl acetate solution for 10 minutes and examined under a transmission electron microscope.

Freshly prepared PPFA-cRGD was freeze-dried overnight under vacuum, then 5 mg of the solid powder was dissolved in methanol. The drug concentration was calculated via measurement of the absorbance at 280 nm for fulvestrant and 254 nm for abemaciclib using high performance liquid chromatography (HPLC). The encapsulation efficiency (EE) and drug loading rate (DLR) were respectively calculated according to the following equations: EE (%) = mass of drug encapsulated in NPs/initial mass of drug × 100%; DLR (%) = mass of drug encapsulated in NPs/total mass of NPs × 100%.

The in vitro release of fulvestrant and abemaciclib were conducted in PBS at pH 5.0, 6.5, and 7.4, respectively. PPFA-cRGD was dialyzed against 10 mL PBS in a dialysis bag with a molecular weight cutoff of 10 kDa at 37°C and shaken at 100 rpm. A dialysate sample (100 μL) was collected at chosen time points and the removed volume from the dialysis was replaced with fresh PBS. The concentrations of released drugs were determined using HPLC and quantified against standard curves.

MCF7 cells were cultured in RPMI1640 medium supplemented with 10% FBS and 1% penicillin/streptomycin at 37°C in a humidified, 5% CO₂ incubator. Cells (3,000 cells/well) in the logarithmic growth phase were seeded into 96-well plates and incubated with drug-loaded NPs at 37°C for 24 hours. The cell viability was evaluated via CCK-8 kits and the optical density of formazan in each well was measured using a microplate reader at 450 nm. For apoptosis assays, the cells were harvested after treatment with the different NP formulations in 6-well plates and stained with an Annexin V APC/PI kit. The analysis of apoptotic cells was performed on a BD C6 flow cytometer.

To analyze the cell targeting ability of the NPs, different NP formulations were labeled with Cy5.5 fluorescent dye. Cy5.5-labeled NPs were prepared using the single emulsion method. Briefly, the copolymer was prepared and was then mixed with Cy5.5 dye in the presence of free fulvestrant and abemaciclib. The subsequent emulsification generated by ultrasonication made the amphiphilic polymer to self-assemble into a micelle structure with the Cy5.5 dye and free drugs encapsulated within. MCF7 cells were seeded onto confocal microscopy dishes or 6-well plates at a density of 5 × 10⁴ cells/mL and incubated with the different Cy5.5-loaded NPs on ice for 30 minutes, after which, the NPs in the supernatant were removed and the cells were washed three times in ice-cold PBS. For confocal microscopy imaging, the cells were fixed in 4% paraformaldehyde for 15 minutes, then stained with DAPI to label the nuclei and PKH67 to label the cell membranes. For flow cytometry analysis, the cells treated with Cy5.5-loaded NPs were collected and centrifuged at 1,000 rpm for 5 minutes to sediment the cells. Afterward, the cells were washed three times in PBS and analyzed by flow cytometry.

Tumor tissues of patients with breast cancer were immediately collected postsurgical resection in advanced DMEM/F12 medium (containing 1% penicillin-streptomycin and 50 μg/mL Primocin). The tumor tissues were photographed and minced into 1–4 mm³ pieces. The pieces were then digested with collagenase (1.5 mg/mL, Sigma) in organoid culture medium on an orbital shaker at 37°C for 2 hours. Afterward, the digested fragments were diluted with 10 mL AdDFGH+/+/+ medium (containing advanced DMEM/F12, 10 mmol/L HEPES, 1 × Glutamax, and 1% penicillin-streptomycin) and strained with a 100 μm Cell Strainer (Falcon) followed by centrifugation at 350 × g. The resulting pellet was resuspended in 10 mL AdDFGH+/+/+ and centrifuged at 350 × g. Next, the pellets were resuspended in ice-cold Matrigel (Growth Factor Reduced, Corning) and 10 μL droplets of Matrigel-cell suspension were seeded onto preheated 96-well, round bottom, ultra-low attachment culture plates (Corning). The plates were maintained at 37°C in a cell incubator to allow the Matrigel to solidify. After 30 minutes, prewarmed organoid culture medium was added to each well of the plate. Advanced DMEM/F12 containing 250 ng/mL human R-Spondin 1 (Peprotech), 5 nmol/L heuregulin 1 (Peprotech), 5 ng/mL KGF (Peprotech), 20 ng/mL human FGF10 (Peprotech), 5 ng/mL human EGF (Peprotech), 100 ng/mL human Noggin (Peprotech), 500 nmol/L A83-01 (Tocris), 5 μmol/L rho-associated kinase (ROCK) inhibitor (Y-27632; Selleck), 500 nmol/L SB202190 (Selleck), 1 × B27 supplement (Gibco), 1.25 mmol/L N-acetylcysteine (Sigma), 5 mmol/L nicotinamide (Sigma), 1 × GlutaMax (Gibco), 10 mmol/L HEPES (Gibco), 1 × penicillin/streptomycin (Gibco), and 50 μg/mL Primocin (Invivogen) were used to culture the breast cancer patient-derived organoids. The medium was changed every 3 days and the breast cancer organoids were passaged between days 7 and 14. Briefly, breast cancer organoids embedded in Matrigel were physically released by repeated pipetting using a 1 mL pipette. After centrifugation at 350 × g, the pellet was further dissociated by resuspension in prewarmed TrypLE Express (Gibco) and incubated at room temperature for 3 minutes, followed by the addition of 10 mL AdDFGH+/+/+ and centrifugation at 350 × g. The pellets were resuspended in ice-cold Matrigel and layered onto preheated culture plates at an appropriate ratio (1:2 to 1:5). After initial splitting, 5 μmol/L ROCK inhibitor was added to aid the outgrowth of organoids from single cells.

Breast cancer organoids were passaged and digested into single cells by TrypLE treatment. Droplets containing suspended cells were seeded into each well of preheated 96-well round bottom ultra-low attachment culture plates (Corning), followed by solidification of the Matrigel at 37°C in a tissue culture incubator. The prewarmed organoid culture medium of 100 μL was added to each well, the time of which was denoted day 0. A concentration series of nanodrug carrying abemaciclib and/or fulvestrant was added on day 3, free abemaciclib and/or fulvestrant were also added into 96-well plates in the same concentration series. Each concentration condition was performed in triplicate. Three days later, ATP levels were measured using the CellTiter-Glo three-dimensional (3D) Cell Viability Assay (Promega) according to the manufacturer's instructions. The results were normalized to a blank control, which contained only Matrigel.

Breast cancer organoids embedded in Matrigel were fixed with 4% paraformaldehyde in PBS at room temperature for 30 minutes, which typically led to the complete degradation of the Matrigel and the release of the embedded organoids. Suspended tissues were collected and centrifuged at 300 × g to remove the fixation buffer, followed by washing with ultra-pure water and centrifuged again to obtaining pellets. Following resuspension in water, the organoids were spread on glass slides and allowed to attach by natural drying. The attached organoids were rehydrated with PBS, followed by permeabilization with 0.25% Triton X-100 in PBS for 1 hour at room temperature and blocked with 3% BSA in PBS containing 0.1% Triton X-100 for at least 3 hours at room temperature. Subsequently, breast cancer organoids were incubated overnight at 4°C with blocking buffer-diluted primary antibodies against ER (Abcam, catalog no. ab108398, RRID: AB_10863604), cytokeratin 8 (Abcam, catalog no. ab53280, RRID: AB_869901), and Ki67 (Abcam, catalog no. ab92742, RRID: AB_10562976). The organoids were washed three times with PBS, each time for 1 hour, then incubated with a goat anti-rabbit secondary antibody (Bioss, diluted in blocking buffer) overnight at 4°C. After extensive washing, stained organoids were incubated with DAPI (Solarbio) and imaged by confocal microscopy (Zeiss LSM 710).

Breast cancer organoids were treated with TrypLE and dissociated into single cells. Matrigel droplets containing the cell suspension were seeded into preheated, ultra-low attachment, 96-well culture plates (Corning), followed by incubation at 37°C for the solidification of the Matrigel. Afterward, prewarmed organoid culture medium was added into each well. Three days later, Cy5.5-labeled nanodrugs were added to each well of the plate, free Cy5.5 was added as the control. Two hours later, the organoids were treated with TrypLE at 37°C for 5 minutes, followed by the addition of 10 mL AdDFGH+/+/+ medium and centrifugation at 350 × g for another 5 minutes. The precipitates were resuspended in PBS and analyzed with a flow cytometer (Agilent). The data were processed and analyzed using NovoExpress Software (version 1.4.1).

For plasma pharmacokinetics evaluation, free Cy5.5, Cy5.5-loaded PPFA, Cy5.5-loaded PPFA-mRGD, or Cy5.5-loaded PPFA-cRGD was intravenously administered to MCF7 tumor-bearing female nude mice (n = 3). At the indicated time points postinjection, the blood samples were collected via the tail vein and evaluated by the PerkinElmer IVIS Spectrum imaging system to determine Cy5.5 fluorescence intensity. In addition, we conducted plasma pharmacokinetic study of free fulvestrant, free abemaciclib, and PPFA-cRGD nanodrug in a rat model using a Thermo Fisher Scientific TSQ Quantis Plus LC/MS. In brief, the Sprague-Dawley rats (RRID: RGD_70508) were randomly divided into four groups (n = 4) and administered a single intravenous dose of 0.5 mL PPFA-cRGD nanodrug (each milliliter of PPFA-cRGD nanodrug contained 0.53 mg abemaciclib and 0.27 mg fulvestrant), 0.5 mL abemaciclib (0.53 mg/mL), 0.5 mL fulvestrant (0.27 mg/mL), or saline. Blood samples (200 μL) were then collected into heparinized tubes at 0, 0.083, 0.167, 0.5, 1, 2, 4, 8, 24, 48, and 72 hours after injection, and were centrifuged immediately at 3,000 rpm for 15 minutes to obtain the plasma. Each plasma sample of 50 μL was transferred to a 1.5 mL centrifuge tube, and 10 μL of carbazepine internal standard working solution (1 μg/mL) and 200 μL acetonitrile solution were then added. The samples were vortexed at 2,500 rpm for 3 minutes and then centrifuged at 10,000 rpm for 10 minutes. A total of 100 μL of the supernatant was injected into a sample bottle for subsequent analysis of fulvestrant plasma levels using the LC/MS system.

For in vivo tumor-targeting and biodistribution analysis of PPAE-cRGD, the nude mice bearing MCF7 tumor were anesthetized after tail-vein injections of Cy5.5-labeled NPs and scanned with an IVIS imaging system for whole-body fluorescence imaging at selected time points. At the endpoint of the experiment, the mice were euthanized and the major organs (heart, lung, spleen, liver, and kidney) and tumor tissues were excised for ex vivo fluorescence imaging and quantitative analysis.

The orthotopic MCF7, HCC1428, or ZR75-1 breast cancer xenograft mouse model and a PDX orthotopic breast cancer mouse model were constructed to evaluate the antitumor effect of PPFA-cRGD. Before construction of tumor models, 17β-estradiol pellets (0.72 mg, 60-day release, Innovative Research of America) were transplanted subcutaneously into the mice for external estrogen supplementation. To establish MCF7, HCC1428, or ZR75-1 orthotopic breast cancer mouse model, female nude mice received injection of 5 × 10⁶ breast cancer cells/mouse into the mammary fat pad. For generating the breast cancer PDX model, ER-positive tumor tissues were collected from patients of the Fourth Hospital of Hebei Medical University (Shijiazhuang, Hebei, P.R. China) and transplanted into the mammary fat pads of NOD-SCID mice (4–6 weeks old) after the resection of necrotic tissue. Once the longest diameter reached up to 1 cm, tumors were divided into small pieces (∼4 mm³) and subsequently passaged by implantation into the fat pads of different nude mice. The mice bearing orthotopic MCF7, HCC1428, or ZR75-1 tumor xenografts or PDX tumors were randomly divided into seven groups (n = 6). When the tumor volume reached about 150 mm³, the mice were treated with saline, PPAE-cRGD (empty NPs), free fluvestrant plus abemaciclib (Ful + Abe), PPAE-F-cRGD (NPs loaded with fulvestrant), PPAE-A-cRGD (NPs loaded with abemaciclib), PPFA-mRGD (NPs loaded with fulvestrant and abemaciclib and surface modified with mutant peptide mRGD), or PPFA-cRGD (NPs loaded with fulvestrant and abemaciclib and surface modified with cRGD peptide) every 3 days for five total treatments. The free fulvestrant and abemaciclib were predissolved in a small amount of ethanol as a stock solution prior to use. Free drugs and different formulations (contained equivalent 0.27 mg fulvestrant and/or 0.53 mg abemaciclib) were administrated via intravenous injection per treatment. The tumor volume was monitored using a vernier caliper every 2 days and calculated by the following formula: tumor volume = length × width²/2. The body weight of the mice was also measured over the course of the experiment with a digital balance. At the endpoint of the experiment, the major organs and tumor tissues were collected for IHC or immunofluorescence analysis. In addition, blood samples were obtained for biosafety assessments. All animal experiments were approved by the ethics committees of the National Center for Nanoscience and Technology.

To acquire a comprehensive safety assessment, healthy Balb/c mice were treated with saline (n = 6), free drugs (Ful + Abe, n = 12), or PPFA-RGD (n = 12). In the free drug combination group, the mice were treated with fulvestrant (5 mg/kg per mouse, solubilization in castor oil) by intramuscular (IM) injection, and abemaciclib (10 mg/kg per mouse) via oral gavage for a total of seven doses. Saline and PPFA-cRGD were administrated by intravenous injection, for which the administration dosage of PPFA-cRGD was 0.1 mL per mouse, with the concentrations of fulvestrant at 0.27 mg/mL and abemaciclib at 0.53 mg/mL. The main observable indicators of toxicity and side effects during the treatment process included inflammation at the injection site, diarrhea, and body weight loss. At the endpoint of the experiment, the mice were sacrificed and the major organs (heart, liver, spleen, lung, kidney, and colon) and leg were collected for H&E staining analysis. Blood samples were also acquired for blood biochemical analysis. We further explored the safety of PPFA-cRGD in Bama miniature pigs due to the similarities of this model animal to human physiology, anatomy, and genome. Bama miniature pigs (n = 3) were administered 5 mL of PPFA-cRGD/pig (the dosage was converted from the therapeutic dose of each mouse) or saline via the marginal ear vein every 3 days for a total of five treatments. The major organs were removed after the last treatment for histopathologic analysis and the blood samples were collected for routine blood and biochemistry analysis.

The data were analyzed using GraphPad prism 8.0 (RRID:SCR_002798) and SPSS 17.0 (RRID:SCR_002865). All error bars indicate SD of the mean. One-way ANOVA was applied for multiple comparisons and Student t test was used for comparison between two groups, *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

All data associated with this article are present in the article or the Supplementary Materials and Methods. All other raw data can be obtained by contacting the corresponding author upon request.

To efficiently package fulvestrant and abemaciclib, we developed a biodegradable micelle NP based on the self-assembly of PPAE copolymer macromolecules (Supplementary Fig. S1), with the ability to form a nanoformulation with the two drugs owing to its excellent loading capacity for hydrophobic drugs. First, we prepared a series of fulvestrant/abemaciclib mixtures in which we initially used a fixed amount of fulvestrant, then added different amounts of abemacliclib to yield fulvestrant/abemaciclib ratios in the range of 1:4 to 4:1. Cell viability experiments in the MCF7 cell line showed that the strongest synergistic killing effect was observed at a fulvestrant/abemaciclib ratio of 1:2 (Supplementary Fig. S2). Therefore, we used this ratio for subsequent experiments. Under these conditions, the maximal encapsulation efficiency of fulvestrant and abemaciclib in the nanoformulation separately reached 92.2% and 99.2% (Supplementary Table S1). To confer the NPs with tumor-targeting ability, we modified cRGD [cyclic (Arg-Gly-Asp-d-Phe-Cys)] peptides (that can bind to αvβ3 integrin, which is highly overexpressed on the surface of tumor cells and endothelial cells) on the particle surfaces through the chemical reaction of sulfhydryl group on the peptides and maleimide groups on the PPAE (35–37). ¹H NMR results demonstrate the successful conjugation of the peptides, wherein the disappearance of peak of Mal indicated that the Mal groups of PPAE-Mal were completely consumed by the sulfhydryl groups of cRGD to form the conjugates (Supplementary Fig. S3). The resultant drug-loaded nanoformulation, PPFA-cRGD, had an average diameter of approximately 108 nm and a moderate zeta-potential of approximately 23 mV, displaying typical spherical structures with smooth surfaces (Fig. 1C–E). Compared with the empty NPs (PPAE-cRGD), addition of the two drugs significantly increased the particle diameters from 73 to 108 nm, but did not have significant effect on surface charges of the particles (Fig. 1D and E). PPFA-cRGD showed a maximal drug release rate during the first 6 hours at pH 5.0 compared with pH 6.5 or 7.4 (Supplementary Fig. S4), suggesting a low pH-sensitive capability of the nanomaterials. This would ensure the efficient release of the loaded drugs in the weekly acidic tumor microenvironment and the acidic lysosomes. In addition, the PPFA-cRGD exhibited high stability in size distribution in human serum for up to 72 hours at pH 7.4 (Supplementary Fig. S5), which is a critical requirement for NPs to achieve a long circulation time in the body.

To study the cellular targeting ability of PPFA-cRGD, we incubated MCF7 cells, a well-established ER-positive human breast cancer cell line, with Cy5.5-labeled nontargeted PPFA (without cRGD modification), PPFA-mRGD (with mutant cRGD peptide modification) or PPFA-cRGD. Thirty minutes later, the confocal laser scanning microscopy (CLSM) imaging revealed that significant fluorescence intensity was observed at the cell membrane of the cells treated with PPFA-cRGD, while little fluorescent signal was present in the other two controls (Supplementary Fig. S6). Flow cytometry results also indicated the strongest cell-associated fluorescent signal in the PPFA-cRGD–treated group after incubation for 30 minutes (Fig. 2A). These data suggested that PPFA-cRGD harbors a high targeting ability for ER-positive breast cancer cells. We proceeded to assess the cellular uptake of Cy5.5-labeled PPFA-cRGD in MCF7 cells with a prolonged incubation time for up to 4 hours at 37°C. In contrast to nontargeted PPFA and PPFA-mRGD, PPFA-cRGD–treated cells displayed a much stronger fluorescent signal inside the cells over time, as demonstrated by CLSM (Supplementary Fig. S7A) and flow cytometry data (Supplementary Fig. S7B), indicative of efficient uptake of PPFA-cRGD by MCF7 cells. Next, we evaluated cell killing efficacy of PPFA-cRGD in three different ER-positive breast cancer cell lines, MCF7, HCC1428, and ZR75-1 cells. Our results demonstrate that PPFA-cRGD treatment effectively inhibited tumor cell proliferation, with a significantly higher proportion of apoptotic cells (Supplementary Figs. S8A–S8C and S9A–S9C) and a much lower ER expression level (Supplementary Fig. S10A–S10C) than those of other control treatments.

Tumor organoid is a highly promising preclinical model that recapitulates gene expression, drug response, and cancer histology (38, 39). We next evaluated the effects of PPFA-cRGD on an in vitro organoids established using primary tumor tissues. ER-positive breast cancer organoids were successfully established by culturing the mixtures of collagenase-treated tumor tissues collected from patients with clinical cancer (Fig. 2B). Detailed patient information for obtaining the tumor tissue samples is shown in Supplementary Table S2 (Case1–Case4). The organoids were comprised of spherical clusters of cells (Fig. 2C). CLSM imaging and Western blot analysis revealed significant expression of the cellular proliferation marker, Ki67 protein and breast cancer–specific makers, CK8 and ER proteins, in the organoids (Fig. 2C; Supplementary Fig. S11). We then assessed the uptake of PPFA-cRGD by the organoids by incubating the organoids with different Cy5.5-labeled nanoformulations. One hour later, a significantly greater fluorescent signal was observed in the PPFA-cRGD–treated group compared with the nontargeted PPFA and PPFA-mRGD groups (Fig. 2D and E; Supplementary Fig. S12). CellTiter-Glo 3D cell viability assay further demonstrated that PPFA-cRGD treatment exhibited the most powerful breast cancer cell killing effect in the organoids from different patients with breast cancer (Fig. 2F). Importantly, compared with free fulvestrant plus abemaciclib, PPFA-cRGD exerted a much stronger killing effect in the organoids, verifying the higher efficacy of NP-mediated delivery of therapeutic agents to the tumor cells. These findings indicate that PPFA-cRGD can effectively target the tumor organoids and penetrate into the tissues, leading to rapid death of tumor cells.

We then tested the in vivo tumor-targeting ability of PPFA-cRGD via intravenously administering the Cy5.5-labeled NPs into female nude mice bearing orthotopic breast tumors (∼300 mm³) which were established by inoculating 5 × 10⁶ MCF7 tumor cells to the mammary fat pad of each mouse. The fluorescence labeling stability in the PPFA-cRGD was first verified by incubating the Cy5.5-labeled PPFA-cRGD samples in PBS supplemented with 10% FBS in the dark at room temperature for 48 hours. The result of fluorescence spectrophotometer showed that after 48 hours, the fluorescence intensity of the Cy5.5-labeled NPs merely decreased by approximately 20% compared with original fluorescence intensity, indicating a favorable fluorescence stability of the nanodrug (Supplementary Fig. S13). Then, the Cy5.5-labeled PPFA-cRGD was intravenously administering into the tumor-bearing mice and the real-time fluorescent signal was monitored using a noninvasive near-infrared optical imaging system. Although there is a possibility that a certain degree of dye dissociating from NPs, the encapsulation or surface modification of these particles with fluorescent dye is still considered a feasible method for analyzing biodistribution of nanodrugs within the body (18, 19). A strong fluorescence was present exclusively in the tumors of PPFA-cRGD–treated mice 1 hour postadministration, with a maximal fluorescence accumulation observed 2 hours later (Fig. 3A and B). In contrast, no specific fluorescence signal was observed in the tumors in the free Cy5.5, nontargeted PPFA or PPFA-mRGD–treated groups over time. We next excised the tumor tissues and major organs from the mice to measure the biodistribution of PPFA-cRGD more accurately at 24 hours postadministration. Ex vivo imaging and quantitative analysis revealed that PPFA-cRGD accumulated in the tumor about 7-fold more efficiently than free Cy5.5 and 3-fold higher than the nontargeted PPFA. A weak tumor fluorescence was also detected in the PPFA-mRGD–treated group, largely attributing to the general passive accumulation of NPs to tumors. In addition, the kidneys, as the major organ, were found to be responsible for PPFA-cRGD metabolism and excretion over time (Fig. 3C and D). To further analyze the long-term in vivo biodistribution of PPFA-cRGD, we extended the real-time monitoring time to 72 hours and found that 72 hours postadministration, PPFA-cRGD accumulated at tumor tissues in an even higher fluorescence intensity than that observed at 24 hours (Supplementary Fig. S14A and S14B). However, the fluorescence signals in the other major organs, such as the liver and heart nearly disappeared compared with those at 24 hours, indicating that PPFA-cRGD possesses excellent tumor-targeting ability and can be cleared from the body over time, which is an essential characteristic for a nanodrug intended for clinical application.

We next investigated the blood pharmacokinetic behavior of PPFA-cRGD after intravenous administration into healthy mice, through measuring the fluorescence intensity in the blood as a function of time. The results show that PPFA-cRGD had the longest blood circulation time with a half-life of 10.3 hours, relative to the half-life of 4.8, 4.2, and 1.5 hours for PPFA-mRGD, nontargeted PPFA and free Cy5.5, respectively, according to a one-compartment open model (Fig. 3E and F). In addition, we further generated drug–time curves of free fulvestrant, free abemaciclib and PPFA-cRGD by LC/MS. PPFA-cRGD significantly prolonged the in vivo half-life of the free drugs: from 0.06 to 1.28 hours for fulvestrant and from 0.05 to 3.79 hours for abemaciclib (Supplementary Fig. S15). These findings indicate that PPFA-cRGD is expected to provide a longer window of time for more efficient drug delivery to solid tumors in the body.

To evaluate the in vivo therapeutic effects of PPFA-cRGD, the nude mice bearing approximately 150 mm³ orthotopic breast tumors established by MCF7, HCC1428, or ZR75-1 cell inoculation were randomized into seven groups (n = 6 per group), and administered different formulations, including saline, PPAE-cRGD (empty NPs), free fulvestrant plus abemaciclib (Ful + Abe), PPAE-F-cRGD (NPs loaded with fulvestrant), PPAE-A-cRGD (NPs loaded with abemaciclib), PPFA-mRGD (NPs loaded with fulvestrant and abemaciclib and surface modified with mutant peptide mRGD) and PPFA-cRGD (NPs loaded with fulvestrant and abemaciclib and surface modified with cRGD peptide), via the tail vein at intervals of every 3 days for a total of five treatments (Supplementary Fig. S16A and S16B). In the PPFA-cRGD treated group, the tumors of the mice shrank substantially over time and were almost completely regressed at the end of experiments. Furthermore, the PPFA-cRGD treatment group exhibited the lowest average tumor weight and the longest animal survival in three breast tumor mouse models (Fig. 4A–F). In contrast, PPAE-F-cRGD and PPAE-A-cRGD treatments displayed a significant, but very limited inhibition to tumor growth compared with saline group (Fig. 4A, C, and E). Unexpectedly, PPFA-mRGD treatment elicited a much stronger inhibition on tumor growth than the two free drugs, although this inhibitory effect was far lower than PPFA-cRGD treatment. This confirms that the effective drug delivery to tumor sites is achievable only by employing nanocarrier. No inhibitory effect was found in the empty PPAE-cRGD group in contrast to saline group. The tumor weight and survival time analysis of tumor-bearing mice correlated well with the tumor growth curves (Fig. 4A–F). dT-mediated dUTP nick-end labeling (TUNEL) staining of MCF7 tumor tissues also showed the strongest apoptosis-positive cell signals in the PPFA-cRGD treated group (Fig. 4G; Supplementary Fig. S17), with widespread tumor tissue necrosis (Supplementary Fig. S18). IHC and Western blot analyses further reveal that PPFA-cRGD treatment substantially decreased ER expression within MCF7 tumor tissues (Fig. 4H and I; Supplementary Fig. S19).

PDX mouse model has been shown to have a high level of similarities to the patient's original disease and preserve both intertumoral and intratumoral heterogeneity (40, 41). To evaluate the therapeutic potential of PPFA-cRGD in PDX models, we established an ER-positive orthotopic breast cancer PDX model through subcutaneous transplantation of freshly collected clinical primary breast cancer tissue into immune-compromised mice (Fig. 5A). H&E and IHC staining verified that the PDX tumors retained the histopathologic characteristics of the primary breast tumor and retained a high level of ER expression (Fig. 5B). The mice bearing approximately 150 mm³ tumors were then randomized into seven groups (n = 6 per group). They were intravenously administered with different formulations at intervals of every 3 days for a total of five treatments. After five injections, the mice treated with PPFA-cRGD had a mean tumor volume of 342.6 ± 225.19 mm³, which is far smaller than those of saline (1973.4 ± 209.2 mm³), empty PPAE-cRGD (1799.3 ± 343.6 mm³), free two drugs (1082.1 ± 325.2 mm³), PPAE-F-cRGD (1171.6 ± 218.5 mm³), PPAE-A-cRGD (1233.6 ± 345.0 mm³), and PPFA-mRGD (889.9 ± 216.2 mm³) treated groups (Fig. 5C and D). Clearly, the PPFA-cRGD outperforms other controls in suppressing tumor growth, which correlated with a substantial lower tumor weight (Fig. 5E; Supplementary Fig. S20). At the end of experiments, the average tumor weight in the PPFA-cRGD group was approximately 6-fold lower than that in the saline group, and was nearly 4-fold lower than those of the free drugs and PPFA-mRGD groups. Histologic analysis further revealed that, compared with other controls, PPFA-cRGD treatment more effectively reduced intratumoral ER expression and markedly induced tumor necrosis (Fig. 5F–H) and cell apoptosis (Supplementary Fig. S21). Thus, with several doses of injection, PPFA-cRGD provides a substantial therapeutic effect on PDX tumors, suggesting its potential for clinical utility.

Preclinical studies and clinical practice have established that long-term IM injection of fulvestrant aggravates pain at the injection site and leads to severe side effects, including hepatic and renal toxicity, headache, muscle pain, and gastrointestinal disorders (42). Clinically, the direct combination of fulvestrant with abemaciclib can dramatically improve antitumor efficacy in patients with ER-positive breast cancer. Unfortunately, this also increases the incidence of side effects with more serious symptoms, limiting the broad application of the combined strategy in the clinic (43). In our study, to investigate the biosafety of PPFA-cRGD, we first noted the body weight changes of PDX tumor-bearing mice following five treatments with different formulations. Similar to the saline group, a slight but not significant increase in the body weight of the mice was observed in the PPFA-cRGD and other nanodrug-treated groups within the entire therapeutic regimen, while a significant decrease occurred in the free drug (Ful + Abe) group (Supplementary Fig. S22). This indicates that NP-based delivery effectively reduces the systematic toxicity of the free drugs. We then performed serum biochemical analysis of the mice, and similarly found that the free drug treatment caused significant hepatic toxicity, while no any organ toxicity was detectable in the PPFA-cRGD or other nanoformulation treated groups (Supplementary Figs. S23A–S23D and S24).

We next performed tolerance and safety evaluation experiments using healthy mice given multiple PPFA-cRGD administrations over a period of time. Healthy BALB/c mice were randomly divided into three groups treated with saline (intravenous injection), PPFA-cRGD (intravenous injection), or free fulvestrant (IM injection) plus abemaciclib (intragastric administration) to simulate a real-world clinical dosing regimen of combination therapy at intervals of every 3 days for a total of seven treatments. During the treatment process, toxicity responses of the mice, such as pain, vocalizing, rough fur, diarrhea, and weight loss were observed in the free drug group but not in the PPFA-cRGD or saline group (Supplementary Figs. S25 and S26A–S26D). More importantly, the muscle injury caused by multiple injections of fulvestrant at injection sites disappeared in the PPFA-cRGD group owing to the change of drug administration route, while serious injury was obvious in the free drug treatment group (Fig. 6A). We also explored the gastrointestinal toxicity of PPFA-cRGD because fatal diarrhea was reported when fulvestrant was combined with abemaciclib in the clinic (44). The results show that the severe gastrointestinal symptoms, such as vomiting, diarrhea, abdominal pain, and hematochezia occurred in the free drug group, while no any significant adverse reaction was observed in the PPFA-cRGD group (Supplementary Fig. S26A–S26D), suggesting that nanocarrier dramatically improves the tolerance of the drugs. At the experimental endpoint, we collected the stomach and intestinal tissues of the mice and carried out histopathology analysis. The length of small intestines of the mice was measured to assess the severity of colitis, in which much shorter intestines were found in the free drug group compared with PPFA-cRGD or saline group (Fig. 6B). H&E staining revealed severe injuries of the gastrointestinal tract resulting from free drug treatment, presenting as enteritis, loss of gastrointestinal tract structure, the blunting of villi and a nearly complete obliteration of stem cell crypts, while PPFA-cRGD treatment showed a favorable safety profile similar to saline (Fig. 6C). In addition, PPFA-cRGD treatment did not cause any morphologic changes or functional damage to major organs (Fig. 6D–I; Supplementary Fig. S27); however, free drug treatment caused obvious heart and liver toxicity as per change in serum biochemistry indicators (Fig. 6D–G).

Finally, we assessed the biosafety of PPFA-cRGD in healthy Bama miniature pigs, which are more similar to humans in anatomy and physiology (45). We randomized healthy Bama pigs into two groups (n = 3 per group) and administered PPFA-cRGD or saline via the ear vein five times at intervals of every 3 days (Fig. 6J). At the end of the experiments, we collected the major organs of the pigs and performed H&E staining. The results showed that compared with saline, multiple PPFA-cRGD treatments did not cause any morphologic damage to the major organs (Fig. 6K). The routine blood parameters and biochemistry parameters, such as alanine aminotransferase (ALT), aspartate aminotransferase (AST), creatine kinase (CK), white blood cell count, red blood cell count, and platelet count were also not significantly affected by PPFA-cRGD treatments (Supplementary Fig. S28A and S28B). Taken together, these results provide evidence that PPFA-cRGD exhibits a superior safety profile in both mouse and Bama pig models and is particularly attractive for clinical use on account of its benign tolerability and safety profiles.

According to the most recent cancer statistics (2020), female breast cancer incidence has already surpassed lung cancer as the most frequently diagnosed cancer worldwide (46). Metastasis in distal organs eventually occurs in about 20%–40% of patients with breast cancer; this phenomenon is especially frequent in patients with ER-positive breast cancer (47). ER expression is a definitive driver of tumor progression in the majority of breast malignancies (∼80%); thus, the ER is the most suitable target for endocrine therapy (ET) to improve the overall cure rate of breast cancer (48). Fulvestrant is the most effective drug in ET-naïve patients with advanced tumors and is also recommended for patients experiencing disease progression and metastasis following other ET treatments (49). Although prolonged overall survival and progression-free survival have been achieved, how to improve its efficacy, overcome drug resistance, and minimize side effects have remained intractable challenges to the continual clinical application of fulvestrant.

Similar to steroidal compounds, rapid excretion by the hepatobiliary route and poor solubility lead to low oral bioavailability of fulvestrant (50). Consequently, “Faslodex” (commercial fulvestrant for clinical application at 200 mg/5 mL) must be administered by IM injection to acquire a steady plasma concentration of 24–28 ng/mL (51). Unfortunately, the clinical data suggest that the current maximum tolerated dose of Faslodex (500 mg for each treatment) is insufficient to achieve a blockade of the ERα receptor, possibly due to poor bioavailability and pharmacokinetic dynamics (52). In addition, pharmacologic studies have demonstrated that a higher dose of fulvestrant is expected to increase plasma levels of the drug and improve its efficacy (12, 53). However, the only way to achieve this is to develop a more concentrated formulation of fulvestrant, because 5 mL has been reported for adults as the maximum tolerated volume for a single IM (54). Of note, previous research has indicated that this strategy is unlikely to generate a fulvestrant injection with higher concentrations than Faslodex, due to poor solubility (12). Herein, we describe a biocompatible polymer micelle for the delivery of fulvestrant at a much higher plasma concentration than that of Faslodex. To further enhance its tumor-targeting ability, cRGD peptides were decorated on the surface of NPs. With intravenous administration, the resultant nanodrug, termed PPFA-cRGD, exhibits high tumor-specific accumulation and locally releases fulvestrant in response to the weakly acidic tumoral microenvironment (Figs. 2 and 3; Supplementary Fig. S4). The nanoformulation exhibited strong antitumor effects in three orthotopic ER-positive breast cancer cell inoculated tumor-bearing mouse models and a patient-derived ER-positive breast tumor xenograft mouse model (Figs. 2, 4, and 5). Moreover, the large-scale stable production of PPFA-cRGD can be easily achieved owing to the simple but efficacious structure of material, indicating translational potential of the drug as a new therapeutic for patients with ER-positive breast cancer.

Drug resistance is the second major clinical obstacle to the success of many experimental chemotherapeutics. Indeed, the excellent therapeutic effects of fulvestrant are impeded by the development of drug resistance; the majority of patients with ER-positive breast cancer will develop de novo (primary) or acquired (secondary) resistance to fulvestrant (55). Previous work has shown that the molecular mechanisms underlying fulvestrant resistance and tumor progression converge at the cell cycle (56). Both preclinical data and clinical trials suggest that fulvestrant plus abemaciclib (a selective, small-molecule inhibitor of CDK 4 and 6) is able to induce cell-cycle arrest and inhibit the proliferation of cancer cells to significantly overcome fulvestrant resistance; however, this aggressive approach results in dangerous adverse effects (44, 57). In consideration of these properties of abemaciclib, we coencapsulated the inhibitor at an optimal ratio with fulvestrant in our nanoplatform to precisely deliver the two therapeutics simultaneously into tumors with high biosafety. Compared with the slow and inadequate drug bioavailability of Faslodex delivered by conventional IM injection, the burst release of fulvestrant and abemaciclib that can be achieved with PPFA-cRGD in the tumor microenvironment is expected to eliminate cancer cells prior to development of drug resistance (Supplementary Fig. S4). Thus, the combination of abemaciclib with fulvestrant not only improves the antitumor response, but also avoids drug-resistant occurrence, possibly providing a rationale for the powerful therapeutic efficacy of PPFA-cRGD against ER-positive breast tumors (Figs. 4 and 5; Supplementary Fig. S9). The current consensus in cancer treatment is to reduce side effects while increasing or maintaining therapeutic effects, thus providing the ability to live more comfortably with breast cancer is as important as curing cancer (58, 59). Faslodex needs to be intramuscularly injected into each buttock (gluteal areas) as two 5 mL injections on days 1, 15, and 29 of each treatment cycle, then once monthly thereafter at 500 mg per dose. Few evidence-based guidelines for IM injections are available (60). As a result, few health care workers have the training to manage such large IM injections, lacking the appropriate injection skills, knowledge of the possible side effects, and the potential efficacy. It has been shown that the success rates of gluteal IM injections vary between only 32% and 52%, with many or most injections inadvertently leading to subcutaneous drug deposition and/or a reduced dose, both of which decrease the efficacy of Faslodex therapy (61–63). Serious side reactions, including injection site pain, especially neuropathic pain, sciatica, neuralgia, and peripheral neuropathy, due to inaccurate landmarking and inexperienced long-term injection, have been reported to occur in a significant proportion of patients (64–66). These adverse effects have led to significantly decreased quality of life, because routine activities, including walking and sleeping, are critically impacted by injection site pain. In addition, a total of six cases involving necrotic ulceration of the injection site have been reported in patients receiving Faslodex via IM by the Pharmaceuticals and Medical Devices Agency of Japan in 2020 (67). Needless to say, patient adherence to Faslodex treatment is impacted by these adverse effects. In the current study, we designed PPFA-cRGD to enable an increased availability of intravenously administered fulvestrant to avoid injection site reactions caused by invasive IM; the results of our safety evaluation experiments in healthy BALB/c mice proved this advantage (Fig. 6A–C). Hepatic impairment, presenting as elevated ALT and AST, is a common adverse reaction from Faslodex treatment; we were able to recapitulate this effect in mice by free fulvestrant plus abemaciclib treatment. Importantly, no liver function abnormality was detectable in PPFA-cRGD treated mice (Fig. 6D and E). We also utilized a Bama miniature pig model to demonstrate the lack of systematic toxicity or side effects of PPFA-cRGD in a model closer to humans than mice. Neutropenia was reported as one of the most frequent adverse events in the MONARCH 2 trial, which aimed to evaluate the efficiency of fulvestrant plus abemaciclib (44). In the miniature pigs, we did not detect any reduction in the concentration of neutrophils after PPFA-cRGD treatment (Supplementary Fig. S28B), suggesting excellent in vivo safety and tolerability of PPFA-cRGD.

To facilitate clinical translation, we plan to both standardize the process of PPFA-cRGD preparation and comprehensively explore the nanoformulation's mechanism of action as well as its biosafety and immunogenicity in non-human primates. Interestingly, previous work has shown that abemaciclib can increase tumor immunogenicity by enhancing tumor antigen presentation and suppressing the proliferation of regulatory T cells (68). However, we here used immunodeficient mice for the in vivo biodistribution and therapeutic experiments. Future work should explore PPFA-cRGD–induced antitumor immunity responses and the combination of PPFA-cRGD with immunotherapeutics, such as checkpoint inhibitors, in immunocompetent mice. It may also be possible to develop predictive biomarkers that can identify patients who are more likely to benefit from PPFA-cRGD treatment. For example, measuring integrin receptor expression or tumor vascularization levels in breast tumor tissues may provide valuable information about the potential for RGD targeting and the enhanced permeability and retention effect in the given patient.

In summary, we developed a biocompatible micelle nanodrug, PPFA-cRGD, for the tumor-specific codelivery of fulvestrant and abemaciclib. The nanomaterials used here can be easily modified with targeting motifs, and are degraded into water and carbon dioxide in vivo without apparent side effects. PPFA-cRGD not only successfully addresses the clinical limitations of fulvestrant (e.g., unmet efficacy, drug resistance, severe side effects) to endow it with the prospect of continual application, but also circumvents drug resistance by the integration of abemaciclib. As a paradigm for the delivery of multiple hydrophobic drugs, the PPFA-cRGD nanoplatform may enable the extension of the clinical application of fulvestrant to accomplish the ultimate goal of eradicating ER-positive breast tumors.

S. Sukumar reports grants from The Fetting Fund during the conduct of the study. No disclosures were reported by the other authors.

B. Li: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. F. Qi: Data curation, investigation, methodology, writing–original draft. F. Zhu: Data curation, investigation. Z. Lu: Data curation, investigation, methodology. M. Wang: Resources, data curation. T. Chu: Resources, data curation, software, formal analysis, methodology. S. Wu: Data curation, software, formal analysis. J. Wei: Data curation, supervision. Z. Song: Resources, supervision. S. Sukumar: Conceptualization, supervision, writing–review and editing. C. Zhang: Conceptualization, data curation. J. Xu: Conceptualization, resources. S. Li: Conceptualization, supervision, writing–original draft. G. Nie: Conceptualization, resources, supervision, funding acquisition, writing–original draft.

This work was supported by grants from the Beijing Distinguished Young Scientist program (JQ20037 to S. Li), CAS Interdisciplinary Innovation Team (JCTD-2020-04 to S. Li), CAS Project for Young Scientists in Basic Research (no. YSBR-036 to S. Li), the Key Area R&D Program of Guangdong Province (2020B0101020004 to S. Li), National Basic Research Plan of China (2018YFA0208900 to G. Nie), the Strategic Priority Research Program of Chinese Academy of Sciences (XDB36000000 to G. Nie).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.