Self-navigated NIR-II thermophoretic liposome nanomotors for directed tumor delivery and enhanced chemotherapy.

1
Introduction
Drug delivery to solid tumors is hindered by increased interstitial fluid pressure and biological barriers in tumor tissues, thus limiting their diffusion to perivascular regions [[1], [2], [3], [4]]. The resulting poor tumor penetration and insufficient accumulation of therapeutic agents in tumor tissues pose significant challenges for cancer therapy [5]. Nanomotors with asymmetric structures or properties are emerging as novel drug delivery vehicles with the ability to transduce diverse energy sources into their autonomous mechanical motion [[6], [7], [8], [9]], which is beneficial to overcome biological barriers for deep tissue penetration [10,11]. Control over the movement speed and direction of nanomotors is essential for improving the bioavailability of therapeutic agents with enhanced treatment.
Compared with chemically powered nanomotors whose motion behavior relies on surrounding fuel concentration or enzymatic activity [10,12,13], light is a stable external physical stimulus for driving nanomotors with the potential to be remotely manipulated with excellent spatial and temporal resolution [[14], [15], [16]]. Among light propelled nanomotors, thermophoretic ones driven by the second region near infrared light (NIR-II, 1000-1700 nm) have been widely investigated as drug delivery systems, on account of deep tissue penetration and good biocompatibility of NIR-II light [[17], [18], [19]]. Harnessing the local surface plasmon resonance (LSPR) of metal materials, incident NIR-II light can be converted into heat, thereby generating a temperature gradient field to drive the self-thermophoretic motion of nanomotors [[20], [21], [22], [23], [24]]. By modulating the power densities of incident light, it is handy to control the motion speed of photothermophoretic nanomotors.
Motion direction control of photothermophoretic nanomotors is still a challenge, while it's essential for improving their targeted accumulation at specific sites, especially in drug delivery to solid tumors. Precise directionality control via switching incident wavelength for photocatalytic reactions modulation is limited by the penetration depth of ultraviolet and visible light and tissue injury in drug delivery nanomotors [[25], [26], [27]]. And size-dependent shading effect utilized for directional propulsion of a photothermophoretic motor would not exist in nanometer-sized drug delivery systems [28].
Magnetic field is widely explored for navigation in nature. Rotating magnetic field with a high degree of programmability is usually used to offer precise direction control over nanomotors and corresponding swarms, which relies on complex and specialized devices as well as continuous exposure to external magnetic fields [[29], [30], [31], [32], [33]]. External magnetic field generated by permanent magnets is arising as a convenient means for direction guidance [[34], [35], [36]], making it a great candidate for motion direction steering of photothermophoretic nanomotors.
Herein, we report a 1064 nm NIR-II light-driven thermophoretic liposome nanomotor (LipNM) by integrating liposomal doxorubicin (DOX) with core-shell Fe3O4@Cu9S8 nanoparticles (NPs). LipNMs possess both magnetic steering capability and NIR-II-triggered self-thermophoresis, allowing precise control over its motion speed and direction for targeted tumor accumulation and improved therapeutic efficacy. LipNMs are fabricated by a modified thin film hydration method with Fe3O4@Cu9S8 NPs dispersed in the lipid precursor, and hydrated by citric acid buffer, followed by DOX encapsulation via a pH-gradient method. Phase separation between free lipid molecules and oleylamine stabilized Fe3O4@Cu9S8 NPs leads to asymmetric assembly of Fe3O4@Cu9S8 NPs in the lipid bilayer. The as-obtained LipNMs have an asymmetric distribution of Fe3O4@Cu9S8 NPs in one portion. The magnetic Fe3O4 core enables the orientation and positioning of LipNMs under magnetic guidance, facilitating their targeted accumulation in tumor tissue. Localized surface plasmon resonance (LSPR) of Cu9S8 shell in NIR-II region endows Fe3O4@Cu9S8 NPs with intensive photothermal property, and their asymmetric distribution results in temperature gradient inside liposome nanomotors for thermophoresis propulsion. LipNMs also contain DOX for chemotherapy and have a folic acid (FA) labeled lipid bilayer to facilitate cancer cells targeting and reduce side effects of DOX during delivery (Fig. 1A).
After in vivo administration of LipNMs, an external magnetic field (∼15 mT) generated by NdFeB magnets is applied for 5 min to precisely guide the motion direction of LipNMs. Magnetic stimulation enables liposomal nanomedicine to orient uniformly toward tumor tissue. Thereafter, 10 min of 1064 nm NIR-II light irradiation is performed, generating an asymmetric heat distribution within LipNMs. The resulting directional self-thermophoresis highly accelerates their motion and facilitates their targeted accumulation and deep penetration into tumor. After FA mediated endocytosis in cancer cells, the lipid membrane of LipNMs is decomposed by the acidic environment of lysosome, and DOX is released for chemotherapy (Fig. 1B). The as-presented magnetic-steering NIR-II light powered asymmetric liposomal nanomedicine demonstrates controllable motion speed and direction, therefore realizing their targeted accumulation and deep penetration at the tumor site for enhanced cancer chemotherapy.

2
Experimental section
2.1
Materials
Iron (III) acetylacetonate (95%) (Fe(acac)3), copper (II) acetylacetonate (97%) (Cu(acac)2), oleylamine (C18, 80-90%), ammonium chloride (BR), DSPE-PEG-folic acid (DSPE-PEG-FA, MW 2000), poly (ethylene glycol) diacrylate (PEG-DA-600, MW 600), 2-hydroxy-2-methyl-propiophenone (photo initiator, 97%), 1-Chloro-2,2,2-trifluoroethyl difluoromethyl ether (isoflurane, 98%) were purchased from Aladdin Ltd. (Shanghai, China). Sulfur (99%) and trisodium citrate (99%) were purchased from Alfa Aesar Co., Ltd. (Shanghai, China). Doxorubicin hydrochloride (DOX, 99%), DSPE-PEG (MW 2000, 98%) were purchased from Shanghai Bide Pharmatech Ltd. (Shanghai, China). Hydrogenated soya phosphatidylcholine (HSPC, 90%) was purchased from Psaitong Biotechnology Co. (Beijing, China). Cholesterol (99%), citric acid (99.5%), dimethyl sulfoxide (DMSO, 99.7%), Triton X-100 (99%) were purchased from Shanghai Macklin Biochemical Technology Co., Ltd (Shanghai, China). Sodium carbonate anhydrous (AR), L-histidine (BR), trichloromethane (AR) and ethanol (AR) were purchased from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). Fetal bovine serum (FBS) and poly (2-hydroxyethyl methacrylate) (pHEMA, BR) were purchased from Sigma-Aldrich Co., Ltd. (Shanghai, China). Lyso-Tracker Green was purchased from Beyotime Institude of Biotechnology Co., Ltd. (Shanghai, China). Dulbecco's modified Eagle's media (DMEM), phosphate-buffered saline (PBS), trypsin, paraformaldehyde and MTT cell proliferation and cytotoxicity assay kit were purchased from KeyGEN Biotech (Nanjing, China). Calcein-AM/PI Double Stain Kit was purchased from Abbkine Scientific Co.,Ltd (Wuhan, China). Aqueous solutions used in the experiments were prepared with deionized (DI) H2O (18.1 M Ω cm, Millipore).

2.2
Cells lines
The 4T1 (RRID: CVCL_0125) cell line and human umbilical vein endothelial (HUVEC) (RRID: CVCL_9Q53) cell line were purchased from KeyGEN Biotech. Co., Ltd (Nanjing, China) at September 23, 2023. All cell lines were subjected to short tandem repeat analysis and mycoplasma testing to confirm the absence of contamination. The cells were cultured at 37 °C in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS in a humidified incubator containing 5% CO2 and 95% air.

2.3
Animals
All experimental procedures in this study were performed in accordance with the NIH guidelines for the care and use of laboratory animals (NIH Publication no. 85-23 Rev. 1985) by qualified operators (Certificate Number of 220212754 and 220195151), and the ethical approval for related animal experiments were obtained from the Institutional Animal Care and Use Committee (IACUC) of Southeast University with approval number of 20250103031. Female BALB/c mice with 5-6 weeks old were purchased from Jiangsu Qinglongshan Biotechnology Co., Ltd. (Nanjing, China) and used for preparing the subcutaneous breast cancer model. The mice were housed in a specific pathogen-free barrier facility under controlled environmental conditions, with a temperature of 20-26 °C, humidity of 40%-70%, a 12-h light/dark cycle, and ad libitum access to food and water.

2.4
Synthesis of magnetic Fe3O4 core
Magnetic Fe3O4 nanoparticles (NPs) were synthesized by a modified thermal decomposition method [[37], [38], [39]]. Briefly, 15 mL of oleylamine in three-necked flask was stirred at 600 rpm under nitrogen atmosphere and heated to 300 °C at a heating rate of 2 °C min−1. Then, 0.5 mmol Fe(acac)3 dissolved in 3 mL of oleylamine and 2 mL of N-methyl-2-pyrrolidone was added to the flask. After 10 min reaction at 300 °C, the resultant mixture was naturally cooled down to 60 °C at room temperature. The product obtained was precipitated by adding 30 mL of ethanol and collected by centrifugation at 10000 rpm for 10 min. Magnetic Fe3O4 NPs were finally redispersed in 20 mL of cyclohexane for further use.

2.5
Synthesis of core-shell structured Fe3O4@Cu9S8 NPs
A thin layer of Cu9S8 was grown to magnetic Fe3O4 core to prepare core-shell structured Fe3O4@Cu9S8 NPs according to a previously reported approach [37,40,41]. Typically, 1 mmol sulfur dissolved in 3 mL of oleylamine was rapidly injected into 10 mL cyclohexane dispersion of the as-prepared Fe3O4 nanoparticles at 70 °C and stirred at 800 rpm under nitrogen atmosphere for 10 min. Then, 0.5 mmol Cu(acac)2 dissolved in 4 mL of chloroform and 1 mL of oleylamine was injected into the solution. The resulting mixture was further reacted at 70 °C for another 30 min, leading to a gradual change in color from brown to dark green. Subsequently, the product was collected by centrifugation (10000 rpm, 10 min) and washed with ethanol and chloroform three times. The as-obtained Fe3O4@Cu9S8 NPs were dispersed in chloroform at a final concentration of 2 mg mL−1 and kept at room temperature for further use.
To facilitate their photothermal performance characterization, amphiphilic DSPE-PEG was then coated to the above prepared Fe3O4@Cu9S8 NPs with a modified literature procedure [[42], [43], [44]]. 20 mg of DSPE-PEG (MW 2000) powder was dissolved in 10 mL of chloroform dispersion containing 10 mg of Fe3O4@Cu9S8 NPs, and stirred overnight at 800 rpm at room temperature. After removing chloroform under a nitrogen flow, the resulting hydrophilic Fe3O4@Cu9S8 NPs were collected, ultrasonicated in deionized water for dispersion, and stored at room temperature for further use.

2.6
Synthesis of liposome nanomotors (LipNMs) and liposomal DOX
Liposome nanomotors encapsulating DOX and Fe3O4@Cu9S8 NPs (LipNMs) were fabricated by a thin film hydration method according to the synthesis procedure of commercially available liposomal Doxorubicin (DOX). Hydrogenated soya phosphatidylcholine (HSPC), cholesterol, DSPE-PEG (MW 2000), and DSPE-PEG-folic acid (DSPE-PEG-FA, MW 2000) were dissolved in chloroform in a molar ratio of 25: 17: 1: 1 to prepare the lipid precursor. Chloroform dispersed Fe3O4@Cu9S8 NPs at a concentration of 2 mg mL−1 were added to the lipid precursor at a volume ratio of 1:50. The solvent was removed by rotary evaporation under vacuum at 45 °C to form a thin lipid film. The as-obtained lipid film was hydrated with an equal volume of citric acid buffer to prepare the non-curative liposome nanomotors (denoted as nc-LipNMs). Successive extrusion through 400 nm, 200 nm, and 100 nm polycarbonate membranes was used to homogenize their size distribution.
DOX was loaded into the as-obtained liposomes via a pH-gradient method to generate LipNMs. Typically, 2 mL of nc-LipNMs was mixed with 3 mL of PBS, and the pH was adjusted to 7.0 with 0.1 M Na2CO3. Then, 2 mL of PBS containing 1 mg mL−1 DOX (pH 7.4) was added to the solution and incubated at 50 °C for 50 min. The final product LipNMs were then collected by ultrafiltration and stored at 4 °C for further use.
Liposomal DOX was prepared using nearly identical lipid composition and procedures as LipNMs without Fe3O4@Cu9S8 nanoparticles, and DSPE-PEG was used instead of DSPE-PEG-FA.

2.7
Photothermal performance characterization
0.1 mg mL−1 hydrophilic Fe3O4@Cu9S8 NPs dispersed aqueous solution and LipNMs dispersed aqueous solution containing equivalent concentration of Fe3O4@Cu9S8 NPs were irradiated respectively under 1064 nm NIR-II laser (1.5 W cm−2) for 10 min, and the solution temperatures were recorded with an IR thermal camera (Fortic 225-1, Fotric, China). Deionized water was employed as a negative control.
Aqueous dispersions of 0.1 mg mL−1 hydrophilic Fe3O4@Cu9S8 NPs dispersed were exposed to 1064 nm laser irradiation at various power densities (0.5, 1.0, 1.5, and 2.0 W cm−2) for 10 min, respectively, and their corresponding temperature curves were recorded. Besides, 0.1 mg mL−1 hydrophilic Fe3O4@Cu9S8 NPs dispersed aqueous solution was subjected to NIR-II laser for 80 min (10 min break for each 10 min exposure), and corresponding time-dependent temperatures were recorded to evaluate their photothermal stability.

2.8
Stability evaluation
LipNMs were stored in PBS and DMEM containing 10% fetal bovine serum at 4 °C for 14 days, respectively. Their size and corresponding polydispersity index (PDI) were recorded at Day 1, 3, 5, 7, and 14, respectively. Besides, LipNMs were exposed to 10 min 1064 nm laser irradiation under power density of 1.5 W cm−2. DLS analysis was carried out to characterize their sizes, and transmission electron microscope (TEM) was employed to characterize their morphology.
Besides, the fluorescence stability of DOX under NIR-II laser irradiation was evaluated, as its intensity was used to monitor the in vitro and in vivo behavior of lipNMs. Free DOX and LipNMs solution were both exposed to 1064 nm NIR-II laser (1.5 W cm−2) for 10 min, and the corresponding fluorescence emission spectra (λex = 470 nm, λem = 520-700 nm) were recorded before and after NIR-II laser treatment. The fluorescence intensity at 595 nm wavelength was used for quantitative comparison.

2.9
Physical simulation of asymmetric thermal variation in LipNMs
Theoretical simulations were conducted using ANSYS Fluent 2024 R1 to illustrate the asymmetric temperature distribution within LipNMs. The liposome nanomotor was constructed as a spherical vesicle composed of an aqueous core and a lipid bilayer shell with asymmetrically anchored photothermal domains. The external computational domain was filled with water to mimic physiological fluid conditions. Based on the photothermal conversion efficiency, the confined photothermal region was simplistically defined as a constant heat source to simulate heat generation under continuous NIR-II laser irradiation. A pressure-based coupled solver was adopted for all simulations and heat transfer was calculated according to the following equation:where is the specific heat capacity, T is temperature, k is the thermal conductivity of the fluid, and represents the viscous dissipation term.
Given the low Reynolds number characteristic of the nanoscale system, convective effects are neglected and temperature transport is dominated solely by thermal conduction. A local fluid subdomain was constructed around the nanoparticle with all six outer boundaries specified as pressure outlets and backflow enabled. Steady-state three-dimensional energy equations were solved to obtain the temperature distribution inside the nanomotor and in the surrounding aqueous medium.

2.10
Movie capture of LipNMs motion
10 μL of LipNMs dispersed solution was dropped onto a glass slide and irradiated with a 1064 nm laser. The motion behavior of LipNMs were observed with an inverted optical microscope (DMi8, Leica, Germany) equipped with a 40x objective, and movies of 30 frames at a rate of 6 frames per second (FPS) were recorded. Movement trajectories were traced and analyzed with Image J and Origin software. Based on the extracted trajectories, the motion speed of LipNMs was calculated according to the equation:Where χ and t represent the sailing distance and duration time of LipNMs, respectively.
Corresponding mean square displacements (MSD) was measured according to the equation below [45]:Where (x0, y0) and (xΔt, yΔt) referred to the positions of LipNMs at time point of t0 and after time interval of Δt, respectively.

2.11
DOX loading capacity and release assay
DOX encapsulation and release were analyzed using fluorescence spectroscopy. A fluorescence calibration curve was established by measuring fluorescence intensities for various concentrations of DOX at 595 nm. The DOX loading capacity was determined as the mass ratio of encapsulated DOX in LipNMs to the nanoparticles with the same volume. For DOX release, 2 mL of LipNMs (equivalent DOX concentration, 0.2 mg mL−1) was suspended in PBS and irradiated with 1064 nm NIR-II laser (1.5 W cm−2 for 10 min). The accumulated amount of DOX released from LipNMs at 37 °C was calculated by measuring fluorescence intensities of released DOX in the dialysate at series time points.

2.12
Intracellular localization of LipNMs
500 μL of 4T1 cells suspension (∼105) was seeded in a single well confocal dish and cultured at 37 °C for 24 h. Then, 4T1 cells were treated with LipNMs (500 μL, equivalent DOX concentration, 12 μg mL−1), and the confocal dish was placed in a home-built external magnetic field generated by two oppositely arranged NdFeB magnets for 5 min. The confocal dish was then irradiated with 10 min NIR-II laser (1.5 W cm−2) and further incubated for subsequent treatments. Each dish was washed with PBS thoroughly three times to remove free LipNMs. 4T1 cells were subsequently stained with Lyso-Tracker Green (500 μL, 150 nM) for 15 min to indicate cell lysosomes, and the co-localization of DOX fluorescence and Lyso-Tracker Green fluorescence were monitored with a confocal laser scanning microscope (CLSM) (TCS SP8, Leica, Germany). CLSM images were analyzed with LAS X Office and corresponding Pearson's R values were calculated via Image J software.

2.13
In vitro models for penetration capability characterization of LipNMs

(i)Hydrogel Model. Hydrogel precursor containing 5.0% (v/v) PEG-DA-600 and 1.0% (v/v) 2-hydroxy-2-methyl-propiophenone (photo initiator) was poured into a plastic cuvette and polymerized into hydrogel upon UV irradiation to simulate the tumor extracellular matrix. The resulting hydrogel was equilibrated in PBS (pH 7.4) at room temperature overnight and subsequently incubated with 100 μL of LipNMs in the presence (NIR-II(+) B(+)) and absence (NIR-II(+) B(−)) of 5-min external magnetic field stimulation before 30 min NIR-II irradiation (1.5 W cm−2, with a 10 min break for each 10 min exposure). Passive LipNMs treated hydrogel without NIR-II irradiation (NIR-II(−) B(−)) was set as control. The penetration ability of nanomotors was characterized by measuring the depth of LipNMs (red) infiltrated into the hydrogel.

(ii)Transwell assay. To evaluate the penetration capability of LipNMs at the tumor boundary, 1 × 105 of HUVEC cells were seeded in the upper chamber of a 12-well Transwell per insert (polycarbonate filter, 3 μm pore, Corning) to mimic the vascular barrier of tumor tissue. HUVECs monolayer was treated with 500 μL of LipNMs (equivalent DOX concentration, 12 μg mL−1) in the presence (NIR-II(+) B(+)) and absence (NIR-II(+) B(−)) of 5-min external magnetic field stimulation before 30 min NIR-II irradiation (1.5 W cm−2, with a 10 min break for each 10 min exposure), and co-cultured with 4T1 cells in the lower chamber for another 12 h. Passive LipNMs treated HUVEC cells without NIR-II irradiation (NIR-II(−) B(−)) were set as control. The lower 4T1 cells were stained with Lyso-Tracker Green (500 μL, 150 nM), and colocalized with LipNMs via CLSM characterization. To evaluate the penetration capability of LipNMs inside tumor tissues, a similar experiment was conducted with the upper chamber HUVECs monolayer replaced by tumor cells (4T1) to mimic the interior of tumor tissue.

(iii)3D multicellular tumor spheroids (MTSs) model. 3D MTSs model of 4T1 cells was constructed according to previously reported literature [[46], [47], [48], [49]]. A thin film of poly (2-hydroxyethyl methacrylate) (pHEMA) was coated on the bottom of a T-25 cell culture flask, with subsequent exposure to ultraviolet light for 2 h to sterilize. 5 × 105 4T1 cells suspended in 5 mL DMEM containing 10% FBS were seeded in the above prepared pHEMA coated T-25 cell culture flask, and incubated at 37 °C in a humidified incubator containing 5% CO2 and 95% air. The culture medium was replaced every other day. 4T1 3D MTSs (∼200 μm in diameter) were formed spontaneously in 7 days. To investigate the penetration capability of LipNMs in 3D MTSs, 1000 μL of LipNMs (equivalent DOX concentration, 12 μg mL−1) were incubated with 3D MTSs in the presence (NIR-II(+) B(+)) and absence (NIR-II(+) B(−)) of 5-min external magnetic field stimulation before 30 min NIR-II irradiation (1.5 W cm−2, 10 min break for each 10 min exposure). Free LipNMs treated 3D MTSs without NIR-II irradiation (NIR-II(−) B(−)) was set as control. After continuous incubation for 12 h, 3D MTSs were washed and re-suspended by PBS, and imaged with a CLSM imaging system. Besides, the distribution of fluorescence intensity within the 3D MTSs was further analyzed with LAS X office software.

2.14
MTT assay
MTT assay was performed to evaluate the therapeutic effect of LipNMs in vitro. 5 × 103 4T1 cells were seeded in a 96-well microplate per well and cultured at 37 °C overnight. Subsequently, 100 μL of samples containing a series of equivalent DOX concentrations (including free DOX, liposomal DOX, LipNMs) and 100 μL of corresponding noncurative nc-LipNMs at lipid particle concentrations identical to LipNMs but without DOX loading were added into each well and co-cultured for 24 h, respectively. Then 50 μL of MTT reagent (1 × ) was added to each well and co-cultured with 4T1 cells for 4 h. After removing the remained MTT solution, the formazan precipitates produced by live cells were dissolved in 150 μL DMSO and quantified by measuring the absorbance at 490 nm with a microplate reader. The same experiment was conducted with 50% inhibitory concentration (IC50) of LipNMs and equivalent concentration of nc-LipNMs in the presence and absence of 10 min 1064 nm laser irradiation (1.5 W cm−2) to investigate the effect of laser exposure on cell viability.

2.15
Live/dead cells staining
The therapeutic effect of LipNMs was further evaluated by Calcein AM/PI staining. 1 × 105 4T1 cells were inoculated in a 6-well plate per well and incubated at 37 °C for 24 h. Then, 4T1 cells were treated with PBS, LipNMs (12 μg mL−1, equivalent DOX concentration) (free LipNMs in the presence and absence of 10 min NIR-II irradiation, and magnetic-steered LipNMs with 10 min NIR-II irradiation) and equivalent nc-LipNMs, respectively. After incubation for an additional 12 h, 4T1 cells were stained with Calcein AM (green for living cells) and PI (red for dead cells) solution at 37 °C for 30 min and washed with PBS three times. Fluorescence images were recorded with an inverted fluorescence microscope.

2.16
In vitro hemolysis assay
Whole blood samples containing heparin were collected from healthy BALB/c female mice and centrifuged at 1000 rpm for 10 min to discard the supernatant. The sedimented red blood cells (RBCs) were washed and diluted with saline to prepare 2% RBCs suspension. Series concentrations (0.25-4.0 mg mL−1) of nc-LipNMs dispersed in saline were mixed with 2% RBCs suspension and co-incubated at 37 °C for 3 h. The mixture solutions were then centrifuged and the absorbance of supernatants at 570 nm wavelength (Asample) was measured. 2% RBCs suspension mixed with equal-volume saline and DI water were set as negative control (Asaline) and positive control (Awater), respectively. The hemolysis percentage was calculated by the following equation:

2.17
In vivo antitumor effect of LipNMs
Female BALB/c mice with 5-6 weeks old were purchased from Jiangsu Qinglongshan Biotechnology Co., Ltd. (Nanjing, China) and used for preparing the subcutaneous breast cancer model by implanting 4T1 cells (2 × 106) into the lateral side of the right hind limb subcutaneously. The tumor volumes were calculated using formula V=(L×W2)/2, where W and L referred to the width and length of the tumor, respectively. After the tumor volumes reached 80 mm3, these mice were weighed and stochastically divided into five groups (n = 5). Then randomly grouped mice were intratumorally injected with 100 μL of (1) saline, (2) DOX solution, (3-5) LipNMs at an equivalent DOX dose of 2 mg kg−1. Once injected, the mice in group 4 (G4) were irradiated with 1064 nm laser (1.5 W cm−2) for 30 min (with a 10 min break for each 10 min exposure) while the mice in group 5 (G5) were subjected to a home-built magnetic field generated by NdFeB magnets for 5 min before 30 min NIR-II irradiation (with a 10 min break for each 10 min exposure). Mice in group 3 (G3) were set as a passive LipNMs control, receiving neither magnetic field stimulation nor NIR-II irradiation. The treatments were operated at Day 1, 4 and 7, repeatedly, meanwhile the tumor volumes and body weights of mice were recorded. All the mice were sacrificed at Day 10, and the tumors were dissected and photographed. For the quantitative assessment of therapeutic efficacy, the relative tumor volume (RTV) and the tumor growth inhibition rate (TGI) can be calculated using the following formulas:Where V0 is the tumor volume measured at the time of initial treatment (day 0), and Vt is the tumor volume measured at each subsequent time point (day t).Where RTVT and RTVC represent the relative tumor volume of the mice in the treatment group and the control group, respectively.

2.18
In vivo biodistribution of LipNMs
The in vivo biodistribution of LipNMs was assessed by bioimaging of DOX fluorescence (excitation/emission 465/600 nm) via a small animal imaging system (IVIS Lumina Series III, PerkinElmer, USA). The subcutaneous tumor-bearing mice were intratumorally injected with 100 μL of LipNMs (equivalent DOX concentration, 4 mg kg−1), and magnetically stimulated for 5 min to orient motion before 30 min NIR-II laser (with a 10 min break for each 10 min exposure). In order to avoid the interference from the signal of non-interested biological tissues, tumor tissues and the main organs including heart, liver, spleen, lung, and kidney of mice were harvested at different time points (1, 3, 6, 12, 24, 48 h post administration) for ex vivo imaging. Besides, LipNMs injected mice without magnetic field stimulation were further treated in the presence and absence of 30 min NIR-II irradiation (10 min break for each 10 min exposure). The mice were sacrificed and their heart, liver, spleen, lung, kidney, and tumor tissues were collected for fluorescence bioimaging after 24 h. The normalized DOX fluorescence intensity of specific tissues (tumor, liver, etc.) was further calculated as the fluorescence intensity of the target tissue divided by the total fluorescence intensity of the tumor and major organs in the mouse.
Furthermore, the tissue localization of LipNMs in vivo was also investigated in an orthotopic breast cancer model. 4 Tl cells (3 × 105) suspended in PBS were orthotopically injected into the mammary fat pads of female BALB/c mice. When the orthotopic 4T1 tumors reached approximately 100 mm3, the mice were randomly divided into five groups and intravenously injected with the following formulations at an equivalent DOX dose of 4 mg kg−1: DOX solution, Liposomal DOX, or LipNMs. The mice receiving LipNMs were further divided into three subgroups that received distinct treatments: one served as an untreated control; a second group received NIR-II laser irradiation at the tumor site at 8 h post-injection; and a third group was exposed to a magnetic field for 5 min prior to identical NIR-II irradiation at the same 8-h time point. The DOX fluorescence intensity of tumor at the predetermined time points (2, 4, 8, 12, 24, and 48h) was monitored via bioimaging after intravenous administration.

2.19
Histopathology analysis and hematological examination
After the whole treatment, tumor tissues and the main organs including heart, liver, spleen, lung, and kidney of mice were harvested from the five groups and fixed in 4% paraformaldehyde, and embedded in paraffin. The paraffin-embedded tissues were then sectioned and stained with hematoxylin and eosin (H&E) for observation with an optical microscope. For TUNEL assay, the paraffin slices of tumor tissues were treated sequentially with xylene, gradient ethanol, and washed with PBS twice. The slices were then incubated with proteinase K at 37 °C for 20 min, rinsed with PBS, and stained with TUNEL at 37 °C for 1 h in the dark. Subsequently, DAPI was used to stain the nuclei of tumor cells for observation of cell apoptosis on CLSM.
Besides, whole blood and serum samples of mice were also collected for hematological examination. White blood cell (WBC), red blood cell (RBC), blood platelet (PLT), hemoglobin (HGB), mean corpuscular hemoglobin concentration (MCHC), mean corpuscular volume (MCV), red blood cell distribution width (RDW) and hematocrit value (HCT) were detected for blood routine analysis, while alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBiL), blood urea nitrogen (BUN), creatinine (CREA), and uric acid (UA) were measured for liver and kidney function tests.

2.20
Statistical analysis
Data were presented as the mean ± standard deviation unless indicated otherwise. The unpaired two-tailed t-test and one-way analysis of variance (ANOVA) were used to evaluate statistically significant differences using GraphPad Prism. The p values of less than 0.05 were considered significant. *p < 0.05, **p < 0.01, ***p < 0.001. Correlations were computed using Pearson's r analysis with ImageJ software.

3
Results and discussion
3.1
Synthesis of LipNMs
As the major component in LipNMs to control motion direction and speed for targeted delivery and deep tissue penetration, core-shell structured Fe3O4@Cu9S8 NPs were prepared according to a modified literature procedure [37,39,50]. Fe3O4 core contributed to the magnetic-steering property of LipNMs and demonstrated a diameter of 6.47 ± 0.836 nm (Fig. S1A and B). High-resolution TEM (HRTEM) image showed characteristic Fe3O4 lattice spacing of 0.32 nm (Fig. S1C). Continuous reaction with Cu(acac)2 precursor resulted in the coating of Cu9S8 shell on Fe3O4 core. Core-shell structured Fe3O4@Cu9S8 with oleylamine as stabilizing ligand demonstrated a diameter of 8.81 ± 1.52 nm (Fig. 2A) and corresponding HRTEM image demonstrated characteristic Cu9S8 lattice spacing of 0.27 nm (Fig. S2A), indicating the successful coating of Cu9S8 on Fe3O4 core. The electron diffraction patterns demonstrated (4,0,0), (2,2,2), (3,1,1) crystal faces for Fe3O4 and (1,1,13), (0,0,34), (1,0,21), (1,0,13) crystal faces for Fe3O4@Cu9S8, which corresponded with Fe3O4 and Cu9S8 phase respectively, further validating the successful coating of Cu9S8 on Fe3O4 core (Fig. S1D and Fig. S2B). The magnetism of Fe3O4 and Fe3O4@Cu9S8 NPs was measured by sweeping the external magnetic field between −10 and 10 kOe at 298.15 K with a saturated mass magnetization value of ∼42.4 emu g−1 and ∼21.9 emu g−1, respectively (Fig. 2B). Corresponding magnetic hysteresis curves demonstrated no remanence magnetization or coercivity, suggesting that the superparamagnetic nature of Fe3O4 core for steering motion direction. The weight contribution of the non-magnetic Cu9S8 shell might contribute to the decrease in saturated mass magnetization value of Fe3O4@Cu9S8 NPs. Localized surface plasmon resonances (SPR) for vacancy-doped Cu9S8 shell enhanced its absorption and resonances [41,51], therefore Fe3O4@Cu9S8 NPs demonstrated intense absorption band in the region around 1100 nm (Fig. 2C). The strong absorbance endowed Fe3O4@Cu9S8 NPs great photothermal capability in NIR-II region. To measure its photothermal performance, oleylamine stabilized Fe3O4@Cu9S8 NPs were coated with amphiphilic DSPE-PEG to make them hydrophilic [43,44]. Aqueous dispersion of 100 μg mL−1 Fe3O4@Cu9S8 NPs demonstrated a power density dependent rapid temperature increase upon 1064 nm light irradiation (Fig. S3A), and reached 66.1 °C within 10 min under 1.5 W cm−2 NIR-II laser irradiation (Fig. 2D, Fe3O4@Cu9S8). Besides, the temperature variations in response to NIR-II laser irradiation for four on/off cycles revealed the favorable photothermal stability of Fe3O4@Cu9S8 NPs (Fig. S3B).
Liposome nanomotors (LipNMs) encapsulating DOX and Fe3O4@Cu9S8 NPs were fabricated by a thin film hydration method according to the synthesis procedure of commercially available liposomal doxorubicin (DOX) [52,53]. The lipid precursor containing chloroform dispersed hydrogenated soya phosphatidylcholine (HSPC), cholesterol, DSPE-PEG, DSPE-PEG-FA, and Fe3O4@Cu9S8 NPs was rotary evaporated to form a uniform thin lipid film, and hydrated by citric acid buffer. Phase separation between free lipid molecules and oleylamine stabilized Fe3O4@Cu9S8 NPs led to asymmetric assembly of Fe3O4@Cu9S8 NPs in lipid bilayer. Afterwards, DOX was loaded into the as-obtained liposomes via a pH-gradient method to generate LipNMs (Fig. 1A). TEM image, energy dispersive X-ray (EDX) elemental mapping image, and cryo-TEM image of LipNMs confirmed their morphology and asymmetric distribution of Fe3O4@Cu9S8 nanoparticles (Fig. 2E and Fig. S4). Additionally, the as-obtained LipNMs demonstrated a hydrodynamic diameter of 219 ± 6.83 nm with polydispersity index (PDI) of 0.0593 ± 0.0102, and zeta potential of −1.34 ± 0.214 mV, and kept stable in PBS and cell culture medium over 14 days (Figs. S5 and S6). The stability of LipNMs under 1064 nm laser irradiation was also investigated. TEM images and DLS analysis demonstrated that LipNMs retained their intact morphology and stable size under NIR-II laser irradiation (Fig. S7). LipNMs also demonstrated characteristic absorption of FA at 284 and 356 nm in UV-vis absorption spectrum compared with FA-mPEG-DSPE, and DOX characteristic absorption at 479 nm verified successful encapsulation of DOX in LipNMs (Fig. S8). The loading capacity of DOX was calculated as 11.7 μg per mg LipNMs. According to the in vitro release experiment, efficient DOX release at pH 6.5 reached 65.9% within 96 h (Fig. S9), which ensured the efficient drug release in tumor microenvironment.
Fe3O4@Cu9S8 NPs cluster distributed asymmetrically in LipNMs acted as a heat source under NIR-II irradiation and generated temperature gradient inside LipNMs. The corresponding self-thermophoresis resulted in the powerful motion of LipNMs. Numerical simulations were performed to profile the temperature gradient inside LipNMs according to previously reported researches [20,54,55], which exhibited a high temperature of 55 °C at Fe3O4@Cu9S8 NPs located position, and gradually decreased to 25 °C at the other side of LipNMs where Fe3O4@Cu9S8 NPs did not locate (Fig. S10). Fe, Cu, and S content (originating from Fe3O4@Cu9S8 NPs) in LipNMs was measured via ICP-MS to be 0.168 ± 0.00157, 0.281 ± 0.00376, and 0.239 ± 0.00596 μg mg−1, respectively, confirming the small portion of Fe3O4@Cu9S8 NPs in LipNMs. Due to the spatially confined heat generation of asymmetrically anchored Fe3O4@Cu9S8 NPs and inherently low thermal conductivity of phospholipids (0.20 ± 0.02 W m−1 K−1 at 300 K) [56], LipNMs had little impact on the ambient temperature under NIR-II laser irradiation. Aqueous dispersion of 14.0 mg mL−1 LipNMs containing 0.4 μg mL−1 Fe3O4@Cu9S8 showed similar extent of temperature change as water in response to 1064 nm NIR-II irradiation (Fig. 2D, LipNMs), which validated the simulation predictions and ensured negligible thermal injury to tissues when administrated in vivo. The LipNMs dispersion with an equivalent Fe3O4@Cu9S8 concentration of 100 μg mL−1 exhibited a comparable temperature rise to that of the aqueous dispersion of pure Fe3O4@Cu9S8 nanoparticle at the same concentration (Fig. S11), confirming the heating ability of Fe3O4@Cu9S8 in LipNMs. The temperature gradient inside LipNMs generated propulsion force for its powerful motion. The motion videos (Movies S1-S4) and corresponding movement trajectories of LipNMs (Fig. 2F) demonstrated increased motion speeds in response to enhanced power densities from 0 to 1.5 W cm−2 (Fig. 2G), because higher power density produced larger temperature gradient inside LipNMs for more powerful propelling force [57]. Mean square displacement (MSD) for LipNMs under different 1064 nm laser irradiation power densities demonstrated quadratic functions of time (Fig. 2H), indicating that LipNMs movement originated from the self-driving effect rather than a heat enhanced Brown motion. Although LipNMs demonstrated outstanding motion capability in response to NIR-II laser irradiation, their motion directions were random (indicated by the arrow in Fig. 2F), which composed a barrier for further improving their accumulation at specific site.

3.2
Controllable manipulation of LipNMs
External magnetic field exerted on asymmetric magnetic nanoparticles made them steer till the magnetic ends acquired a unified orientation towards the south pole along the direction of the magnetic field lines due to the magnetic dipole interaction, the principle of energy minimization, and the directional effect of the external magnetic field. Core-shell structured Fe3O4@Cu9S8 NPs not only acted as heat source for self-thermophoresis of LipNMs under NIR-II light irradiation, but also, their magnetic core Fe3O4 steered LipNMs self-thermophoresis in the opposite direction of magnetic field lines (Fig. 3A). This enabled controllable manipulation of LipNMs for tumor-targeted delivery.
LipNMs demonstrated magnetic steering behavior with uniform orientation in an external magnetic field (∼15 mT) generated by NdFeB magnets without NIR-II laser irradiation and obtained an initial velocity (denoted as ν(−)) along magnetic field lines (Movie S5). The corresponding representative trajectories of LipNMs showed universal swarming direction (Fig. 3B). The magnetic core-shell structured Fe3O4@Cu9S8 NPs ends in LipNMs were steered towards the south pole of external magnetic field. NdFeB magnets were removed 5 min later, and LipNMs with magnetic-steered orientation were irradiated with NIR-II laser afterwards. Fe3O4@Cu9S8 NPs ends in LipNMs absorbed NIR-II light irradiation with temperature increase, and generated a temperature gradient field opposite to former exerted magnetic field, thus LipNMs moved in a reversed direction (denoted as ν(+)) (Fig. 3C–Movie S6). Once the NIR-II laser was switched off and NdFeB magnets were set, the nanomotors rapidly decelerated to ν(−). The cyclic “off/on” activation of LipNMs motion were repeated, which indicated great promise for on-demand manipulation of LipNMs by a switch between external magnetic field guidance and NIR-II exposure (Fig. 3D–Movie S7).
Altering the incidence direction of the NIR-II laser by 90° and 180° respectively while keeping the magnetic field unchanged, negligible impact on the motion trajectories of the LipNMs was observed, indicating that motion direction of LipNMs was irrelevant to that of incident NIR-II laser (Fig. 3E, F and Movies S8, S9). Besides, altering the external magnetic field reversed by 180° with NIR-II laser irradiation from the same direction, the motion trajectories of LipNMs reversed compared with that of Fig. 3C–F, confirming the motion direction control of LipNMs by external magnetic field (Fig. 3G–Movie S10). The motion videos (Movies S11-S14) and corresponding movement trajectories of magnetic-steered LipNMs (Fig. S12A) demonstrated increased motion speeds in response to enhanced power densities from 0.5 to 1.5 W cm−2 (Fig. S12B). Mean square displacement (MSD) for LipNMs under different 1064 nm laser irradiation power densities demonstrated quadratic functions of time (Fig. S12C), indicating their movement originated from self-driving effect rather than a heat enhanced Brown motion.

3.3
Penetration capability of LipNMs
LipNMs with controllable motion speed and direction were capable of traversing biological barriers and enhancing their accumulation at tumor site, and facilitated active transportation from exterior tumor tissue to interior as well as endocytosis process, thus improving the therapeutic efficiency and minimizing the off-target toxicity of chemotherapeutic agents. A PEG-DA hydrogel was used to simulate the tumor extracellular matrix in vitro, and incubated with LipNMs under various conditions. Although NIR-II light irradiated LipNMs (NIR-II(+) B(−)) demonstrated deeper penetration compared with free diffused ones (NIR-II(−) B(−)), magnetic guidance before NIR-II irradiation endowed LipNMs initial motion direction and ensured their deepest penetration into PEG-DA hydrogel (Fig. S13).
The intracellular DOX fluorescence was imaged to monitor endocytosis of LipNMs as it was insensitive to 1064 nm laser irradiation (Fig. S14). Under NIR-II laser irradiation, magnetic-steered LipNMs rapidly accumulated in the vicinity of 4T1 cells, and demonstrated intense intracellular DOX fluorescence after 30 min of incubation (Fig. 4A and B, NIR(+) B(+)). Much weaker DOX fluorescence was observed from free LipNMs incubated 4T1 cells in the presence (Fig. 4A and B, NIR(+) B(−)) and absence (Fig. 4A and B, NIR(−) B(−)) of NIR-II irradiation during the same incubation time. Corresponding fluorescence spectra showed a pronounced overlap between the red fluorescent signals (LipNMs) and green fluorescence (lysosomes) in magnetic-steered thermophoretic LipNMs (NIR(−) B(−)), indicating that the autonomous locomotion of LipNMs may effectively facilitate the cellular uptake process (Fig. 4C). Otherwise, long incubation time of 180 min was required to observe DOX fluorescence from 4T1 cells incubated with magnetic-steered LipNMs in the absence of NIR-II irradiation (Fig. S15A). These results indicated that external magnetic field stimulation only contributed to the motion direction control while NIR-II light irradiation induced self-thermophoresis of LipNMs promoted endocytosis process. To evaluate the delivery specificity of LipNMs, magnetic-steered nonFA-LipNMs was set as a control and incubated with 4T1 cells in the presence of NIR-II light irradiation, which demonstrated negligible intracellular DOX fluorescence (Fig. S15B).
The capability of LipNMs penetrating among different cell layers in tissue was evaluated with Transwell experiments. To mimic the process of nanomotors traversing from blood vessel to tumor site, two-dimensional cellular models were established by seeding HUVEC cells in the upper chamber and 4T1 cells in the lower chamber (Fig. 4D) [58]. The upper HUVECs were treated with LipNMs under various conditions, and co-incubated with the lower chamber subsequently. DOX fluorescence was negligible from 4T1 cells that seeded in the lower chamber for free LipNMs treated group in the absence of NIR-II irradiation (Fig. 4E, NIR-II(−) B(−)), while brighter DOX fluorescence was observed from 4T1 cells for the group treated with free LipNMs in the presence of NIR-II irradiation (Fig. 4E, NIR-II(+) B(−)). This result indicated the contribution of NIR-II light as a driving force to the penetration of liposome nanomotors from blood vessel to the tumor site. Furthermore, magnetic stimulation was exerted on LipNMs before NIR-II irradiation, which endowed them with initial motion orientation for targeted accumulation and demonstrated the brightest intracellular DOX fluorescence for 4T1 cells in the lower chamber (Fig. 4E, NIR-II(+) B(+)). Co-localization analysis of the green lysosome and the red LipNMs in Fig. 4F also confirmed this trend, in which much higher fluorescence intensity was found inside 4T1 cells for the magnetic-steered thermophoretic group. Similar Transwell experiment was performed with 4T1 cells seeded both in the upper chamber and lower chamber to verify the intratumoral penetration capability of LipNMs (Fig. S16A). CLSM images also demonstrated much stronger DOX fluorescence from the lower chamber seeded 4T1 cells for the NIR-II light irradiated group with magnetic stimulation (Fig. S16B), indicating NIR-II light irradiation and magnetic-steered motion orientation also improved intratumoral accumulation and penetration capability of liposome nanomotors.
To further investigate the penetration capability of LipNMs in vitro, three-dimensional (3D) multicellular tumor spheroids (MTSs) model was set up to simulate the spatial complexity and heterogeneity of tumor tissues (Fig. 4G) [[46], [47], [48], [49],59]. Z-stack imaging of different cross-sections of 3D 4T1 cell spheres from the bottom to the top was performed with CLSM to evaluate the penetration of LipNMs in MTSs (Fig. 4H). Magnetic stimulation-endowed initial motion orientation and NIR-II irradiation-propelled thermophoresis effectively enhanced LipNMs penetration depth in 3D MTSs, and DOX fluorescence was almost observed in the whole section area with 60 μm away from the bottom (Fig. 4H, NIR-II(+) B(+)). On the contrary, negligible DOX fluorescence appeared in the free diffusion group (Fig. 4H, NIR-II(−) B(−)), or only concentrated at the edge of 3D MTSs for free LipNMs incubated group with NIR-II irradiation (Fig. 4H, NIR-II(+) B(−)), indicating low penetration efficiency and inefficient accumulation. Quantitative analysis of the fluorescence spectra across the 60th μm section of the MTSs demonstrated the strongest DOX fluorescence signal for the group incubated with magnetic-steered LipNMs in the presence of NIR-II irradiation (Fig. 4I). These results confirmed that NIR-II irradiation as well as motion direction control could effectively improve LipNMs penetration in tissues.

3.4
In vitro therapeutic evaluation of LipNMs
LipNMs were internalized by cancer cells via FA-mediated endocytosis, and released DOX intracellular for chemotherapy. Endocytosis pathway of LipNMs were characterized by monitoring the intracellular co-localization of DOX fluorescence (derived from LipNMs) and LysoTracker green fluorescence (derived from lysosome) at different time points. Intracellular DOX fluorescence and LysoTracker Green fluorescence demonstrated obvious overlap in 30 min incubation with Pearson correlation coefficient increased to 0.932 ± 0.0149, and clear separation was observed in 180 min incubation with Pearson correlation coefficient decreased to 0.513 ± 0.0177, indicating successful endosome escape of LipNMs (Fig. 5A and B). Additionally, we labeled the lipid membranes of LipNMs with Dil and monitored their colocalization with lysosomes and DOX (Fig. S17). The results showed that after being internalized into lysosomes, the lipid fraction of LipNMs remained within lysosomes for subsequent metabolism, while DOX was released into the cytoplasm to exert its antitumor effects.
MTT assays revealed DOX-loaded samples, including the free DOX, liposomal DOX and passive LipNMs, exerted significant concentration-dependent antitumor effects in vitro without NIR-II irradiation after 24 h of incubation. Notably, the free DOX demonstrated the strongest cytotoxic activity, as it is immediately bioavailable upon diffusion into cells. Furthermore, liposomal DOX and LipNMs showed comparable antitumor efficacy in the absence of external stimuli, confirming that Fe3O4@Cu9S8 NPs engineered LipNMs maintained chemotherapeutic effect as liposomal DOX. LipNMs treated 4T1 cells at an equivalent DOX concentration of 48 μg mL−1 survived only 25% (Fig. 5C, LipNMs), while noncurative LipNMs (nc-LipNMs) without DOX encapsulation exhibited good biocompatibility with over 90.5% of cell viability across all concentrations tested (Fig. 5C, nc-LipNMs). Only NIR-II light irradiation barely demonstrated cytotoxicity (Fig. 5D–PBS). 4T1 cells treated with nc-LipNMs showed negligible decrease of cell viability in response to NIR-II irradiation, while LipNMs incubated group demonstrated a decrease from 52.5 ± 1.26% to 39.6 ± 2.65% of cell viability corresponding to NIR-II exposure (Fig. 5D, nc-LipNMs, LipNMs). These results indicated the good biocompatibility of liposome nanomotors and their active motion driven by NIR-II irradiation enhanced cellular endocytosis without cytotoxicity.
Live/dead cell staining assay also confirmed the contribution of autonomous motion of liposome nanomotors to 4T1 cell death. The brightest red fluorescence was observed for the group treated with magnetic-steered LipNMs in the presence of NIR-II irradiation, and the corresponding ratio of live to dead cells was the lowest, indicating nanomotors with directional autonomous movement result in the most cell death, thus validating the importance of deep tissue permeation as well as targeted accumulation of drugs for improving their therapeutic effect (Fig. 5E and F, NIR-II(+) B(+)). In contrast, when LipNMs were directed by a reversed magnetic field (opposite to B) followed by NIR-II irradiation, a much weaker red fluorescence was acquired along with a little bit increase of the ratio of live to dead cells (Fig. 5E and F, NIR-II(+) B’(+)). These findings demonstrate that appropriate magnetic orientation is essential for maximizing the therapeutic effect of thermophoretic LipNMs.
In addition, nc-LipNMs demonstrated low hemolysis with less than 5% of RBCs hemolyzed even at the concentration of 4.0 mg mL−1, indicating good hemocompatibility of the liposome nanomotors (Fig. S18).

3.5
In vivo chemotherapy effect verification of LipNMs
The as-designed LipNMs address the biocompatibility and thermophoresis directionality issues of previously reported NIR-II light driven nanomotors, and their central specialty lies in controlled motion direction for drug delivery. Direction control of thermophoretic nanomotors aims not to solve the tumor targeting problem in systemic circulation [60], but the accumulation problem after liposomal drugs reach the tumor site via the bloodstream following tail vein injection. Besides, intratumoral injection can eliminate the differences in systemic circulation accumulation caused by individual heterogeneity among mice, which would affect the final experimental results. Therefore, intratumoral injection was chosen to quickly verify the tissue penetration and accumulation effects.
Subcutaneous 4T1 breast cancer model was established on BALB/c mice to evaluate in vivo antitumor effect of LipNMs. Mice were injected with saline (G1), DOX (G2), passive LipNMs without external stimuli (G3), NIR-II thermophoretic LipNMs without magnetic orientation (G4) and magnetically oriented NIR-II thermophoretic LipNMs (G5) respectively on Days 1, 4, and 7. The in vivo tumor inhibition effects were determined on Day 10 (Fig. 6A).
DOX fluorescence was used to represent LipNMs, and measured via bioimaging to demonstrate the biodistribution of magnetically oriented NIR-II thermophoretic LipNMs at different time points after administration (Fig. S19). Upon the injection, the DOX fluorescence was concentrated in the tumor position, and the proportion of fluorescence in the liver gradually increased over time, indicating that most LipNMs were phagocytosed and metabolized by reticuloendothelial cells in liver [61]. DOX fluorescence observed in kidney might be ascribed to the filtration of a minor quantity of free drug through the glomerulus. Negligible DOX fluorescence was observed in other major organs, which indicated no drug accumulation occurred and confirmed the biosafety of LipNMs. Furthermore, DOX fluorescence of isolated tumor tissue and major organs including the heart, liver, spleen, lungs, and kidneys at 24 h post administration was evaluated. Tumor tissues harvested from mice injected with magnetic-steered LipNMs in the presence of NIR-II irradiation (Fig. 6B, C, G5) demonstrated fluorescence intensity increased by 186.2% compared to the group injected with non-magnetic stimulated LipNMs in the absence of NIR-II light irradiation (Fig. 6B, C, G3), while LipNMs injected group with only NIR-II light irradiation (Fig. 6B, C, G4) showed merely 23.10% increase of fluorescence intensity. This result indicated that magnetic guidance before NIR-II irradiation effectively manipulated LipNMs for directed tumor accumulation. Besides, fluorescence intensities of liver tissues showed a gradual decrease from G3 to G5, confirming that motion-controllable active transportation effectively enhanced drugs retention at tumor site (Fig. S20).
With magnetic stimulation-endowed motion direction orientation, active liposome nanomotors (LipNMs) (Fig. 6D–F and Fig. S21, G5) demonstrated best anti-tumor effect with 94.9% inhibition of tumor size during therapeutic period. In comparison, free LipNMs in the presence and absence of NIR-II irradiation (Fig. 6D–F and Fig. S21, G3 and G4), and simply DOX treatment (Fig. 6D–F and Fig. S21, G2) all showed tumor size increase to some extents. The body weights of mice remained stable for all groups during the whole therapeutic process (Fig. 6G). Magnetic-steered LipNMs with NIR-II light irradiation (Fig. 6H–H&E, G5) also demonstrated severe cell damages in H&E staining image, while only part of cancer cells was killed in other groups. TUNEL experiment also showed highest level of green fluorescence from magnetic-steered active LipNMs injected group (Fig. 6H, TUNEL, G5), indicating enhanced chemotherapy of LipNMs with magnetic-steered motion orientation and active transportation.
H&E staining histopathology analysis of major organs (heart, liver, spleen, lung, and kidney) harvested from mice on Day 10 from magnetic-steered active LipNMs injected group (G5) exhibited little lesion compared with the untreated control group (G1), and avoided DOX-associated cardiotoxicity compared with simply DOX treatment (G2) (Fig. 7A). The hematological results demonstrated that magnetic-steered active LipNMs treated group restored both serum biochemical and blood indices to physiological reference ranges (Fig. 7B and C). The red dashed lines in the inset figures of Fig. 7B and C denote the normal reference range for each individual parameter. These results confirmed the superior biocompatibility of the as-proposed liposome nanomotors (LipNMs), making it potential for clinical application.
Given the critical challenge of inadequate penetration and inefficient accumulation of therapeutic agents within tumor, nanomotors with autonomous mechanical motion capabilities have emerged as appealing drug delivery vehicles for overcoming biological barriers to achieve deep tumor tissue penetration [62]. However, previous attempts to develop active nanomotors for drug delivery encountered two critical hurdles: 1. Difficulty in balancing motion efficiency with biocompatibility. Chen's group proposed Cu2-xSe&PMO-DOX Janus nanomotors relying on NIR light driven autonomous motion for deep tumor penetration, and applied it in synergistic chemo-photothermal-immune therapy against triple-negative breast cancer [63]. Meng and his coworkers reported a thermophoresis-driven nanomotor through partially coating of polydopamine caps on pyroelectric tetragonal BaTiO3 (tBT) NPs, achieving pyroelectric potential-enhanced cell internalization and augmented antitumor treatment [64]. Although these drug delivery vehicles demonstrated outstanding chemotherapy efficiency, its further application in clinic was restricted by biocompatibility issues. 2. Inability to synchronize motion direction with predefined targets. Migle et al. successfully constructed chemically powered Janus platinum-mesoporous silica nanomotors, which asymmetrically catalyze the local decomposition of H2O2, to achieve deep penetration and disruption of the extracellular polymeric matrix of biofilm [65]. Ma et al. fabricated deformable polymer-based nanomotors that respond to NIR light by stretching and contracting to generate a mechanical force propelled motion [66]. In those cases, precise spatial control of artificial colloidal motors, critical for diverse applications, remains highly challenging, as their motion is continuously perturbed by ubiquitous Brownian rotation under low Reynolds number conditions and most macroscale control methods are ineffective [67].
The LipNMs reported in this work address both limitations. By repurposing the synthesis procedure of commercially approved liposomal doxorubicin, we retained the inherent biocompatibility and drug-loading capacity of liposomes, which exhibited key advantage over nanomotors constructed using novel and unvalidated scaffolds [68]. Furthermore, the magnetic pre-orientation step ensured directed-tumors thermophoresis rather than dissipation in non-target regions. The directional modulation accounted for the improvement in drug delivery efficiency of LipNMs compared with NIR-II thermophoretic liposome nanomotors. The underlying mechanism lied in their targeted movement, which minimized off-target drug loss and maximized accumulation within tumor tissues. The therapeutic performance of LipNMs achieved tumor chemotherapy efficiency of 94.9%, highlighting their potential to bridge the “efficacy gap” in current chemotherapy, which limits by inadequate drug accumulation and shallow tissue penetration.

4
Conclusion
In summary, we developed a clinically relevant NIR-II thermophoretic nanomotor based on liposomal doxorubicin with precisely controlled motion direction and speed for the first time, and successfully applied it as an intelligent nanocarrier for enhanced chemotherapy. Liposome nanomotors (LipNMs) containing DOX and Fe3O4@Cu9S8 NPs were synthesized via a modified thin film hydration method using the synthesis procedure of commercially available liposomal DOX. Phase separation between free lipid molecules and oleylamine stabilized Fe3O4@Cu9S8 NPs led to asymmetric distribution of Fe3O4@Cu9S8 NPs in the as-obtained LipNMs. Fe3O4 core enabled the motion direction control of the LipNMs under 5 min exposure to a weak magnetic field (∼15 mT), afterwards the strong absorption of Cu9S8 shell in NIR-II region resulted in asymmetric heat distribution inside liposome nanomotors and propelled them along magnetic-steered orientation to traverse robust biological barriers for targeted tumor tissue delivery. The as-reported liposome nanomotors effectively increased the delivery efficiency of DOX by 186% compared to NIR-II thermophoretic ones without motion direction modulation. Their targeted tumor accumulation combined with deep tissue penetration enhanced tumor chemotherapy efficiency to 94.9%. The as-reported LipNMs represent a step forward in precision chemotherapy that bridges nanotechnology innovation with clinical needs, underscoring the promise of “intelligent” nanomotors in revolutionizing cancer treatment.

CRediT authorship contribution statement
Qing Hu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft. Yingfei Wang: Conceptualization, Funding acquisition, Investigation, Methodology, Supervision, Writing – review & editing. Wei Chen: Investigation, Methodology. Nan Feng: Resources, Visualization. Keju Tao: Investigation. Jie Wu: Conceptualization, Methodology, Resources, Supervision. Ying Liu: Conceptualization, Methodology, Resources, Supervision. Fan Zhou: Visualization. Qing Hao: Supervision. Daoping Xiang: Formal analysis, Methodology, Software. Junjie Chi: Resources, Supervision. Hong Liu: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.