Near-infrared II fluorescence-guided glioblastoma surgery targeting monocarboxylate transporter 4 combined with photothermal therapy

Summary

Background

Surgery is crucial for glioma treatment, but achieving complete tumour removal remains challenging. We evaluated the effectiveness of a probe targeting monocarboxylate transporter 4 (MCT4) in recognising gliomas, and of near-infrared window II (NIR-II) fluorescent molecular imaging and photothermal therapy as treatment strategies.

Methods

We combined an MCT4-specific monoclonal antibody with indocyanine green to create the probe. An orthotopic mouse model and a transwell model were used to evaluate its ability to guide tumour resection using NIR-II fluorescence and to penetrate the blood–brain barrier (BBB), respectively. A subcutaneous tumour model was established to confirm photothermal therapy efficacy. Probe specificity was assessed in brain tissue from mice and humans. Finally, probe effectiveness in photothermal therapy was investigated.

Findings

MCT4 was differentially expressed in tumour and normal brain tissue. The designed probe exhibited precise tumour targeting. Tumour imaging was precise, with a signal-to-background (SBR) ratio of 2.8. Residual tumour cells were absent from brain tissue postoperatively (SBR: 6.3). The probe exhibited robust penetration of the BBB. Moreover, the probe increased the tumour temperature to 50 °C within 5 min of laser excitation. Photothermal therapy significantly reduced tumour volume and extended survival time in mice without damage to vital organs.

Interpretation

These findings highlight the potential efficacy of our probe for fluorescence-guided surgery and therapeutic interventions.

Funding

Jilin Province Department of Science and Technology (20200403079SF), Department of Finance (2021SCZ06) and Development and Reform Commission (20200601002JC); National Natural Science Foundation of China (92059207, 92359301, 62027901, 81930053, 81227901, U21A20386); and CAS Youth Interdisciplinary Team (JCTD-2021-08).

Research in context.

Evidence before this study

Gliomas often exhibit infiltrative growth, and current surgical techniques are inadequate for achieving complete tumour resection while preserving normal brain tissue. Near-infrared window II (NIR-II) imaging, with its strong penetration and low background signal, has emerged as a valuable tool for real-time imaging during surgery. However, the high heterogeneity of gliomas and varying degrees of disruption of the blood–brain barrier (BBB) can lead to uneven fluorescence intensity in tumoral tissue, potentially misleading surgeons.

Added value of this study

In this study, we developed an NIR-II probe (CCM-mAb-ICG) that can penetrate the BBB and selectively bind to monocarboxylate transporter 4 expressed in tumour tissue. We validated the effectiveness of this probe at the cellular level, in mouse and patient tumour samples, and in live mouse surgeries guided by NIR-II fluorescence. Furthermore, we demonstrated that photothermal therapy in mice injected with this probe suppressed tumour growth and prolonged survival.

Implications of all the available evidence

This study introduces a targeted probe designed specifically for low-heterogeneity targets capable of penetrating the BBB. The use of this probe during NIR-II fluorescence-guided glioma surgery provides enhanced assistance to surgeons, facilitating precise tumour resection and reducing the amount of residual tumour tissue, ultimately leading to improved patient survival.

Introduction

Glioblastoma (GBM), classified as a grade IV tumour of the central nervous system by the World Health Organization, ranks among the most malignant types of cancer.¹ Patients diagnosed with GBM have a low average 5-year survival rate of only 6–8%, which is because the diffuse nature of the tumour makes it difficult to distinguish the boundaries between cancerous and healthy tissue during surgical removal, leading to incomplete excision and treatment inefficacy.² Literature reports indicate that patients with new neurological impairments postoperatively, despite total resection of gliomas, have worse prognoses compared to those with partial tumour residue.³ In earlier studies, some classic research also indicated that the formidable challenge of defining brain tumour margins constitutes a significant barrier to improving outcomes for patients with brain tumours.⁴ Owing to the frequent infiltration of GBM into areas crucial for neurological function, a strong demand exists for real-time intraoperative differentiation between the tumour and normal tissues. Visible light microsurgery is a technique commonly used by neurosurgeons who are often faced with the challenge of accurately identifying the boundary of gliomas. An incomplete tumour resection due to inaccurate identification may contribute to early recurrence.⁵

Some conventional imaging techniques⁶ may aid surgeons in accurately visualising tumours and their surrounding structures. However, conventional imaging techniques such as computed tomography (CT) and magnetic resonance imaging (MRI) have limitations during surgery in accurately determining the gliomas.⁷ Developing a probe for real-time fluorescent surgical navigation is crucial for overcoming the limitations of conventional techniques. By utilising fluorescent dyes or probes that bind to tumour-specific biomarkers, this navigation system can provide immediate real-time visualisation of tumour tissue, enabling more precise and thorough tumour resection. Therefore, there is an urgent need to develop fluorescent surgical navigational probes.

To increase accumulation, fluorescent dyes have been used to bind membrane proteins that exhibit differential expression between tumour and normal tissues.8, 9, 10, 11 Antibodies, peptides, and inhibitors that can selectively bind to tumour cells have been extensively employed in the advancement of tumour-targeted fluorescent probes.¹² For instance, cetuximab-IRDye800, a conjugate of the epidermal growth factor receptor (EGFR) inhibitor cetuximab and the infrared fluorescent dye IRDye800, has been developed to selectively target tumours that overexpress EGFR, which are commonly observed in GBM.¹³ Similarly, to target gliomas specifically, the bombesin (BBN) peptide that binds to the gastrin-releasing peptide receptor (GRPR) was labelled with IRDye800 and ⁶⁸Ga, forming ⁶⁸Ga-IRDye800CWBBN.¹⁴ A clinical trial of this probe in patients with GBM proved the clinical effects of these probes in fluorescence-guided surgery.¹⁵

Unlike normal cells that use mitochondrial oxidative phosphorylation as their primary energy source, most cancer cells depend on aerobic glycolysis, a metabolic pathway known as the Warburg effect. Warburg observed that cancer cells tend to metabolize glucose to lactate through fermentation, even in the presence of oxygen, which potentially supports mitochondrial oxidative phosphorylation.¹³ Monocarboxylate transporter 4 (MCT4), which exports lactate from the cell, is upregulated under hypoxic conditions in many types of malignant tumours, including breast, colon, and prostate tumours, as well as gliomas.16, 17, 18, 19, 20 Additionally, elevated MCT4 expression is associated with shorter overall survival in patients with cancer. However, MCT4 is not expressed in normal brain tissue, making it a highly promising target for the fluorescence imaging of gliomas.²¹

Fluorescence-guided surgery (FGS) enables the real-time intraoperative visualisation of tumours using fluorescent probes.²² Clinical trials have extensively studied the application of fluorescent dyes, such as fluorescein sodium, 5-aminolevulinic acid, and indocyanine green (ICG).²³ While these techniques have shown promising results in enhancing tumour resection,24, 25, 26, 27, 28, 29 fluorescent probes generally lack high specificity for targeting tumours. Additionally, the light of these probes falls within the visible or near-infrared (NIR)-I range (700–900 nm), making the imaging quality and tumour delineation susceptible to autofluorescence of normal tissue. In addition, the effectiveness of tumour resection is further limited by the short emission wavelength, which restricts the penetration depth. Recent studies have applied NIR-II (wavelengths of 1000–1700 nm) FGS for biological imaging, which has surpassed conventional NIR-I FGS. NIR-II imaging has several advantages over NIR-I.³⁰ For instance, minimal autofluorescence of tissues in NIR-II facilitates accurate tumour detection. In addition, NIR-II light experiences minimal absorption and scattering when passing through tissues, which is advantageous for clinical applications. Owing to the unique properties of NIR-II light, NIR-II FGS has demonstrated remarkable performance in various preclinical applications.^31,32

The blood–brain barrier (BBB) plays a crucial role in preserving homeostasis in the internal environment of a healthy brain by safeguarding the central nervous system. The BBB prevents the entry of microorganisms and toxins into the brain. However, brain tumours often disrupt and impair this protective barrier, leading to increased permeability of blood vessels within and around the tumour mass.³³ Owing to this impaired integrity in patients with glioma, fluorescence during glioma surgery may be uneven, making it difficult to provide visual assistance during the operation. Conventional nanomaterials face challenges in effectively penetrating the tumour region by relying solely on the enhanced permeability and retention (EPR) effect owing to the heterogeneity and unique characteristics of the BBB,³⁴ which acts as the primary obstacle in brain-related applications.³⁵ To address these obstacles, membrane-functionalised probes leverage the properties of the transmembrane proteins present in cell membranes. This characteristic enables non-selective particles to traverse the BBB. Cancer cell membranes (CCMs) enhance the ability of particles coated with CCM to cross the BBB.³⁶

Methods

Ethics statement

All animal experiments in this study were approved by Ethics Committee of Institute of Automation, Chinese Academy of Sciences (CASIA Issue No. IA21-2302-420303).

Probe synthesis

Monoclonal antibody (mAb)-ICG

The mAb-ICG imaging probe was prepared by conjugating ICG-NHS with a monoclonal antibody specific for MCT4 (RRID: AB_10992036) via random site labelling. More specifically, ICG-N-hydroxysuccinimide (NHS) (Xi'an Qiyue Biology, Xi'an, China) dissolved in dimethyl sulfoxide (DMSO) at a concentration of 10 mg/mL was incubated with anti-MCT4 antibody (Santa Cruz Biotechnology, Dallas, TX) at a concentration of 1 mg/mL for 2 h in the dark at pH 8.5. Unbound ICG-NHS were separated using size-exclusion chromatography with phosphate-buffered saline (PBS) as the elution buffer at a flow rate of 0.5 mL/min.

CCM vesicle extraction

The Beyotime Membrane and Cytosol Protein Extraction Kit was used to extract CCM vesicles (Beyotime, Shanghai, China). The first step involved washing adherent cells with PBS and scraping them off using a cell scraper. The collected cells were then centrifuged at 1000 rpm for 5 min. The collected cells underwent a second round of washing with cold PBS and were subsequently suspended in extraction reagent A containing 1% phenylmethylsulfonyl fluoride (PMSF). Afterward, the mixture was centrifuged at 700×g for 10 min at 4 °C to separate the supernatant. The supernatant was then centrifuged at 14,000×g for 30 min at 4 °C to precipitate the cell membrane fragments. The collected CCM was then suspended in PBS and sonicated in a glass bottle in a bath sonicator for 30 min at a frequency of 42 kHz and power of 100 W at 4 °C. The obtained CCM vesicles were sequentially extruded through porous polycarbonate membranes with pore diameters of 1000, 600, 400, and 200 nm using an Avanti mini-extruder (Avanti, Birmingham, AL).

CCM vesicle coating

The CCM vesicles and mAb-ICG solutions were combined and sonicated for 30 min at 4 °C. The mixture was then consecutively passed through polycarbonate porous membranes with pore diameters of 200 nm and 100 nm. To ensure a uniform and effective dispersion of the synthesized probe, the extrusion process was repeated three times.

Cell culture

U87MG-Luc cells (RRID: CVCL_5J15) genetically modified to express luciferase, ln229 cells (RRID: RRID: CVCL_0393), HS683 cells (RRID: CVCL_0844), and U251 cells (RRID: CVCL_0021) were cultured in Dulbecco's modified Eagle's medium (DMEM; Gibco, Thermo Fisher Scientific, Waltham, MA) supplemented with 10% foetal bovine serum (FBS, Gibco) and 1% penicillin-streptomycin solution (Gibco) in a CO₂ incubator at 37 °C and 5% CO₂. bEnd.3 cells were cultured separately in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin solution under a 5% CO₂ atmosphere and at 37 °C. After each cell thawing, mycoplasma testing was conducted and no mycoplasma infection was detected.

Western blot analysis

After euthanising the subcutaneous tumour-bearing mice, the tumour tissue was excised. Upon sectioning the tumour tissue, signs of necrosis were observed in the interior. The necrotic fluid was collected and the remaining tumour tissue was washed with PBS. Non-necrotic tumour tissue was dissected to obtain internal tumour tissue (diameter 0.5 cm), while the surrounding tissue was considered as peripheral tumour tissue. Tumour tissue was homogenised at 4 °C. The resulting cells were sonicated and then centrifuged at 4 °C and 12,000 rpm for 10 min. The collected supernatant was then combined with the loading buffer and boiled for 10 min. Proteins were resolved using sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene fluoride (PVDF) membranes. After blocking with 5% skim milk for 1.5 h, the membranes were incubated overnight at 4 °C with specific antibodies as indicated. The membranes were then incubated with secondary antibodies for 1.5 h at room temperature. Subsequently, electrochemiluminescence was detected.

Immunohistochemistry and haematoxylin and eosin staining

Fresh tissue was placed in a cryotome cryostat to prepare the sections. The tissue sections were fixed with cold acetone at 4 °C. The slides were then immersed in a 3% H₂O₂ solution in PBS at room temperature for 10 min to block endogenous peroxidase activity. Subsequently, the samples were incubated for 30 min at 37 °C in a 5% bovine serum albumin blocking solution before incubation with primary antibodies (Anti-SC-376140; RRID: AB_10992036) overnight at 4 °C. Next, after washing with PBS, the samples were covered with appropriately diluted secondary antibodies (HRP Conjugated goat anti-mouse IgG; RRID: AB_10808067) at room temperature. Following another round of washing, the sections were placed in a solution for reaction with 3,3′-diaminobenzidine (DAB) to visualise the target antigen. Brain tissue samples collected from mice were preserved in 4% paraformaldehyde for in vitro imaging. After sectioning, tissue sections were embedded in an embedding agent, frozen, and fixed. Finally, the tissue sections were subjected to haematoxylin and eosin (HE) staining for pathological analysis.

Immunohistochemical staining of samples from patients

Initially, specimens excised after surgery were collected from 109 patients (purchased from Bioaitech Co., Xi'an, China). Race or ethnicity data were not collected, as there is currently no research to definitively indicate a correlation between the grading of gliomas and race or ethnicity. Than, specimens were placed in a cryotome cryostat to prepare the sections. The tissue sections were fixed with cold acetone at 4 °C. The slides were then immersed in a 3% H₂O₂ solution in PBS at room temperature for 10 min to block endogenous peroxidase activity. Subsequently, the samples were incubated for 30 min at 37 °C in a 5% bovine serum albumin blocking solution before incubation with primary antibodies (Anti-SC-376140; RRID: AB_10992036) overnight at 4 °C. Next, after washing with PBS, the samples were covered with appropriately diluted secondary antibodies (HRP Conjugated goat anti-mouse IgG; RRID: AB_10808067) at room temperature. Following another round of washing, the sections were placed in a solution for reaction with 3,3′-diaminobenzidine (DAB) to visualise the target antigen.

In vitro targeting evaluation

The U87-Luc glioma cell line was cultured by evenly spreading approximately 5 × 10⁵ cells/mL. Once the cells reached the desired density (60–70% confluence), they were incubated with tetramethylindocarbocyanine perchlorate (DiI) for 30 min and washed three times with PBS. Subsequently, 200 μL of CCM-mAb-ICG, mAb-ICG, or ICG (0.01 mg/mL) were added to separate dishes. The cells were incubated in the dark at room temperature for 2 h. Next, the cells were washed three times with PBS, fixed with 4% paraformaldehyde for 10 min, and stained for 10 min with 4′,6-diamidino-2-phenylindole (DAPI). The cells were then observed under a laser confocal microscope.

In vitro BBB model assay

The bEnd.3 cells (1.0 × 10⁵ cells/well) were seeded in the 12-well transwell plate with a membrane with a mean pore size of 0.4 μm to simulate the BBB environment. The transendothelial electrical resistance (TEER) values were recorded by the Millicell ERS-2 Epithelial Volt-Ohm Meter voltohmmeter (Millipore, Burlington, MA) to assess the cell monolayer integrity during cell culture. When the TEER value reaches 200 Ω cm² or above, it can be considered as an in vitro BBB model.

Nude mouse glioma xenograft subcutaneous model

All the mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. Approximately 5 × 10⁶ cancer cells were mixed with 125 μL PBS and then inoculated into the right lower back of BALB/c nude mice to construct a subcutaneous tumour model. Imaging was performed when the tumour grew to 0.5 cm–1.5 cm.

Orthotopic transplantation brain tumour model

Orthotopic GBM models were established in 5-week-old BALB/c nude mice (Supplementary Material). A mixture containing 1 × 10⁵ U87MG-Luc cells and 5 μL of PBS was surgically injected into the right cerebral area of each mouse at 1 μL/min. For all procedures, the animals were anaesthetised using isoflurane.

The signal-to-background ratio (SBR) was used to assess imaging performance. First, the tumour location was determined based on the bioluminescence imaging (BLI) results from each mouse, and five positions were selected at the tumour site. The total fluorescent signal at each pixel point divided by the number of pixels was calculated as the average tumour signal intensity. The same method was used to determine the average background signal intensity surrounding the tumour. Finally, the average tumour fluorescence signal was divided by the average background fluorescent signal to calculate the SBR.

BLI

Tumour location was detected using BLI on an IVIS Spectrum in vivo imaging system (Revvity, Waltham, MA). Mice harbouring tumours were anaesthetised using isoflurane and subsequently administered injections of 150 mg/kg D-luciferin dissolved in 100 μL of normal saline. BLI was performed 15 min after D-luciferin administration.

NIR-II fluorescence imaging

The CCM-mAb-ICG fluorescent probe was administered at a dose of 1 mg/kg via tail vein injection to mice with tumours. NIR-II fluorescence was detected using an indium gallium arsenide (InGaAs) camera equipped with a lens capable of transmitting NIR-II fluorescence. A filter wheel (Thorlabs, Newton, NJ) for specific wavelengths (1000–1500 nm) was attached to the front end of the lens. Imaging was performed using an excitation laser with a wavelength of 808 nm and a power setting of 50 mW/cm². The exposure times ranged from 0.3 to 2.0 s.

NIR-II fluorescence imaging in the in vivo mouse model

CCM-mAb-ICG (1 mg/kg), mAb-ICG (1 mg/kg), and ICG (1 mg/kg) were injected into the tail vein of the mice (n = 5 in each group), and fluorescent images were acquired using an NIR-II imaging system. After injecting the probes, the fluorescent signals at the tumour site were measured at 2, 4, 6, 8, 10, 12, 24, 48, 72, 96, 120, and 144 h using different exposure times (300, 500, 1000, and 2000 ms) and filters of different wavelengths (1000, 1100, and 1200 nm) with the infrared camera.

NIR-II fluorescence FGS in the in vivo mouse model

NIR-II FGS was performed to resect the orthotopic brain gliomas in mice (n = 5). Following a probe injection period of 10 h, the mice were euthanised, and tumours were removed under fluorescence guidance. First, the skin of each mouse was dissected using scissors to fully expose the cranial bone. Forceps were used to separate the cranial bone and expose the mouse brain tissue. Imaging was performed before surgery, after skin removal, and after skull removal.

Photothermal conversion behaviours of the probe

The temperature changes in solutions containing probes at concentrations ranging from 0 to 40 μg/mL in water were investigated using a laser at 808 nm (0.8 W/cm²) applied for 0–5 min. The temperature variations were captured using an infrared thermal camera (FOTRIC 225s, Dallas, TX).

Statistical analysis

Quantitative analysis of the NIR fluorescent images was performed using ImageJ software (National Institutes of Health, Bethesda, MD), where the images were transformed into pseudocolour representations. Data are presented as mean ± standard deviation (SD). After confirming the normality of the data distributions and homogeneous variances, the differences between groups were assessed using Student's t-test, and the levels of significance were indicated as follows: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. The statistical analysis was conducted using GraphPad Prism version 7.0 (GraphPad Software, San Diego, CA).

Role of the funding source

The sponsors of the study had no role in the study design, data collection, data analyses, data interpretation, manuscript writing, or the decision to submit the paper for publication.

Results

CCM-mAb-ICG characterisation

The mAb-ICG imaging probe prepared by conjugating ICG-NHS with a monoclonal antibody specific for MCT4 demonstrated a ratio of fluorescence to protein of 1.75, consistent with previous reports,³⁷ corresponding to an average of 1.75 fluorescent molecules per antibody. Subsequently, the cancer cell membranes were covered with mAb-ICG. Transmission electron microscopy (TEM) observations (Fig. 1a and b) revealed a hydrodynamic CCM-mAb-ICG size of approximately 138 nm (Fig. 1c). Particles of this size facilitate BBB penetration in GBM.³⁸

Fig. 1.

CCM-mAb-ICG characterisation. (a, b) TEM images. (c) The hydrodynamic size of CCM-mAb-ICG. (d) Probe absorbance and emission spectra. (e) Normalised emission spectra in the NIR-II range. (f) Linear correlation of CCM-mAb-ICG fluorescence intensity with a concentration within a certain range. CCM-mAb-ICG, cancer cell membrane-monoclonal antibody-indocyanine green; TEM, transmission electron microscopy; NIR-II, near-infrared window II.

The absorption spectrum of the probe exhibited a peak at 794 nm, whereas the NIR-I emission spectrum showed a peak at 828 nm. Because of the NIR-II emission-trailing characteristic of ICG, the emission spectrum of the probe was observed in the NIR-II range, with an NIR-II signal at 1500 nm (Fig. 1d and e). Investigation of the correlation between CCM-mAb-ICG concentration and fluorescence intensity revealed a linear correlation within a specific concentration range, with an equation of Y = 2.531 × X + 5.141 and an R² value of 0.9956 (Fig. 1f).

MCT4 expression in human glioma tissue

MCT4 expression was measured in four glioma tissues established in the U87, ln229, HS683, and U251 cell lines and various tumour regions by Western blot analysis (Fig. 2a). Quantitative analysis revealed that the grey-value ratios of the MCT4 bands to the internal reference GAPDH bands in U87-Luc, ln229, HS683, and U251 cells were 1.21, 1.15, 1.15, and 1.14, respectively (Fig. 2b), which differed significantly from the values in normal brain tissue. After euthanising 10 mice with subcutaneous tumours, the tumour tissues were extracted. Among them, five tumour tissues showed signs of necrosis. Necrotic fluid was collected, and the remaining were considered tumour tissues. The diameters of the remaining five tumours ranged from 1.1 to 1.6 cm. These tumours were dissected and the interior tumour tissue (diameter, 0.5 cm) was extracted, while the surrounding tissues were considered lateral tumour tissues. Regardless of the location within the tumour (central or peripheral) or necrotic status, MCT4 expression was stable (Fig. 2c).

Fig. 2.

MCT4 expression in gliomas. (a) WB results. (b) MCT4 expression in normal mouse brain tissue and different human glioma cell lines. (c) MCT4 expression in tumour tissue, necrotic tissue, and central and peripheral regions of subcutaneous tumours in mice. (d–h) MCT4 IHC staining in normal brain tissue and gliomas of different WHO grades in postoperative specimens from human patients with glioma. Normal tissue (d) and WHO grades I (e), II (f), III (g), and IV (h); scale bar: 50 μm. (i) Differences in MCT4 expression between normal human brain tissues and glioma tissues of different WHO grades (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001, Student's t-test). MCT4, monocarboxylate transporter 4; WB, Western blot; IHC, immunohistochemistry; WHO, World Health Organization.

Immunohistochemical staining of postoperative tissue samples from 109 patients with gliomas (specimens were purchased from Bioaitech Co., Xi'an, China) was also performed. The samples included normal (11 cases, Fig. 2d) and WHO grades I (seven cases, Fig. 2e), II (32 cases, Fig. 2f), III (23 cases, Fig. 2g), and IV (36 cases, Fig. 2h) tissues. MCT4 expression differed significantly between tumour and normal brain tissues, with no significant differences in expression among tumours of different grades (Fig. 2i).

In vivo tumour fluorescence imaging of the probe

CCM-mAb-ICG (1 mg/kg), mAb-ICG (1 mg/kg), and ICG (1 mg/kg) were injected into the tail vein of the mice employed as an in vivo tumour model, and fluorescent images were acquired using an NIR-II imaging system. After injecting the CCM-mAb-ICG probe into the U87-Luc tumour model (n = 5), a fluorescent signal was detected in the tumour area after 4 h. The fluorescent signal gradually increased over time, with the tumour-to-background ratio (TBR) peak at 10 h. After 72 h, the tumour fluorescent signal was similar to that of the surrounding tissue (Fig. 3a, probe). When mAb-ICG was injected into the U87-Luc tumour model (n = 5), the brain fluorescent signal was lower than that of the surrounding tissues from 2 to 12 h, indicating that mAb-ICG had an insufficient ability to penetrate the BBB compared to CCM-mAb-ICG. A fluorescent signal was observed in the tumour at 24 h, which peaked at 48 h and gradually diminished after 72 h (Fig. 3a, mAb-ICG). When ICG alone was injected into the U87-Luc tumour model (n = 5), a weak signal was observed in the mouse head; however, no specific signal was detected at the tumour site. After 120 h, no fluorescence was detected (Fig. 3a, ICG).

Fig. 3.

In vivo tumour fluorescence imaging. (a) Imaging effects of three types of dye: CCM-mAb-ICG (probe), mAb-ICG, and ICG, at different times in an in vivo tumour-bearing mouse model. (b) Changes in MFI in CCM-mAb-ICG over time at different wavelengths. (c) Changes in CCM-mAb-ICG SBR over time at different wavelengths. CCM-mAb-ICG, cancer cell membrane-monoclonal antibody-indocyanine green; MFI, mean fluorescence intensity; SBR, signal background ratio.

With a 300-ms exposure time, the fluorescence intensity was strongest at 10 h, gradually decreased from 12 to 24 h, slightly increased from 24 to 48 h, and then decreased again. Owing to the fluorescence properties of ICG, the maximum fluorescence intensity was observed at 1000 nm, whereas the lowest intensity was observed at 1200 nm. These trends were roughly consistent with exposure times of 500 ms, 1000 ms, and 2000 ms (Fig. 3b).

The highest SBR (2.8) was observed 10 h after probe injection, with a 300-ms exposure time and a 1000-nm filter. As the exposure time increased, the tumour fluorescent signal increased, along with an increase in the background signal. At 300-ms exposure, the SBRs for the 1000-nm and 1100-nm filters were similar, but the average fluorescence intensity at 1000 nm (Fig. 3b, 300 ms-MFI) was approximately three times higher than that at 1100 nm, indicating a more significant reduction in the background signal at 1100 nm (Fig. 3c).

NIR-II FGS of glioma in vivo

Imaging was performed at three stages: before surgery, after skin removal, and after skull removal (Fig. 4a). The influence of different tissue barriers on tumour imaging was also analysed. As the barrier tissues were reduced, the level of fluorescence accumulation in the tumour region gradually increased, whereas the average fluorescent signal in the background remained almost unchanged (Fig. 4b). The SBR values before skin removal, after skin removal, and after skull removal were 2.8, 4.8, and 6.3, respectively (Fig. 4c). Next, forceps were used to separate the mouse brain tissue along the fluorescence boundaries, and the tissue with the fluorescent signal was completely excised (Supplementary Video 1). Finally, the mouse brain and tumour tissues were separated for in vitro fluorescence imaging (Fig. 4d–f). Slicing was performed following the red line shown in Fig. 4g. HE staining (Fig. 4h and i) demonstrated complete excision of the tumour tissue, with no residual tumour detected at the margin of the brain tissue.

Fig. 4.

NIR-II FGS of glioma. (a) Comparison of the effects of preoperative, skin removal, and skull removal on fluorescence imaging. (b) MFI and SBR (c) of the tumour and background at different stages of surgery (preoperative, skin removal, and skull removal). (d–f) In vitro imaging of mouse brain and tumour tissues following resection. (g) Incision location (red line). (h–i) HE staining of brain and tumour tissues. NIR-II FGS, near-infrared window II fluorescent image-guided surgery; MFI, mean fluorescence intensity; SBR, signal-to-background ratio; WL, white light; FL, fluorescence; merge, fluorescent signal combined with a grayscale WL image using a pseudocolour overlay technique; HE, haematoxylin and eosin.

In vitro evaluation of probe performance

Ten hours after probe injection, the mice were anaesthetised, followed by white light (WL), NIR-II, and BLI imaging. Subsequently, the mice were euthanised, and the brain tissue was dissected for additional BLI imaging (Fig. 5a). The results demonstrated a strong correlation between NIR-II and BLI localisation, indicating specific binding of the probe to the tumour. Mouse brain tissue was then sliced and subjected to immunohistochemistry (Fig. 5b) and HE staining (Fig. 5c). The tumour regions in HE staining corresponded to strong positive signals in immunohistochemistry staining. Examination of immunohistochemistry staining under high-power microscopy allowed clear differentiation between normal brain and tumour tissues (Fig. 5d and e). Image analysis using the IHC Profiler ImageJ plugin³⁹ revealed a significant difference in the average optical density of MCT4 between tumour and normal tissue (p < 0.001; Fig. 5f, Student's t-test). Confocal microscopy to examine the HE-stained sections at 5× magnification (Fig. 5g, ×5) and a 740-nm filter revealed fluorescence aggregates within the tumour (Fig. 5g, FL) and no fluorescent signal in the surrounding normal tissue. To enhance visualisation, HE staining and fluorescent signals were merged (Fig. 5g, merge). Higher (20×) magnification was used to investigate the binding location of the probe to tumour cells (Fig. 5h, ×20). The fluorescent signal was significantly weaker (Fig. 5h, FL), and after merging, the fluorescent signal was observed primarily on the surface of the tumour cells (Fig. 5h, merge), indicating good specificity of the probe.

Fig. 5.

In vitro evaluation of probe performance. (a) WL, NIR-II, and BLI imaging of mouse heads and BLI imaging of brain tissue. (b, d, e) IHC staining of mouse brain tissue (magnification: 1×; 5×; 20×). (c) HE staining of mouse brain tissue in mice. (f) IHC demonstrating different MCT4 between tumour and brain tissues (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001, Student's t-test). (g) HE staining and fluorescence imaging, as well as superimposed images at 5 × magnification. (h) HE staining and fluorescence imaging, as well as superimposed images at 20 × magnification. WL, white light; NIR-II, near-infrared window II; BLI, bioluminescence imaging; IHC, immunohistochemistry; MCT4, monocarboxylate transporter 4; HE, haematoxylin and eosin.

Probe in vitro targeting and BBB infiltration

The confocal microscopy results showed that the CCM-mAb-ICG probe and DiI colocalized on the tumour cell membrane and that mAb-ICG also colocalised with DiI. However, ICG did not bind specifically to cells (Fig. 6a). Furthermore, an in vitro BBB model established using bEnd.3 cells (Fig. 6b) showed that the CCM coating enabled the probes to penetrate tightly interconnected vascular endothelial cells. As shown in Fig. 6c, the control group without bEnd.3 cells show the free passage of CCM-mAb-ICG through the pores of the transwell culture dish. In contrast, in the remaining three groups, which included bEnd.3 cells, the probes had a limited ability to freely traverse through the pores. The ratio between the MFI values of the basolateral and apical chambers calculated to infer the ability of the probe to penetrate the BBB showed ratios of approximately 0.47, 0.3, and 0.07 for CCM-mAb-ICG, mAb-ICG, and ICG, respectively (Fig. 6d).

Fig. 6.

In vitro targeting and blood–brain barrier penetration by the probe. (a) Confocal results of different probes. (b) Schematic illustration of the in vitro transwell model. (c) Fluorescence imaging of the liquid in the apical and basolateral chambers of transwell culture plates. (d) Mean fluorescence intensity ratios in the basolateral and apical chambers. CCM, cancer cell membrane; ICG, indocyanine green; mAb, monoclonal antibody; DAPI, 4′,6-diamidino-2-phenylindole; DiI, tetramethylindocarbocyanine perchlorate.

In vitro and in vivo photothermal effects

Considering the tumour-targeting ability of the probe at the cellular and in vivo levels, as well as the photothermal effect of ICG, the probe may exhibit both tumour-targeting capability and photothermal effect at the cellular and in vivo levels. We compared the photothermal ablation effects of PBS, probe, and ICG in vitro and in vivo. The photothermal treatment was carried out according to the scheme shown in Fig. 7a, and the change in tumour diameter was recorded during the treatment. Firstly, to assess the photothermal capability of the probe, we evaluated the temperature change of the material with laser irradiation time. Specifically, an 808-nm laser with a power density of 0.8 W/cm² was applied to the probe solution for 300 s, and the temperature change of the probe over time was observed. The results showed that the probe could reach a temperature of 53.9 °C after 300 s of in vitro irradiation (Fig. 7c). An infrared thermal camera (Fluke Ti32; Fluke, Everett, WA) was used to capture the temperature changes. Then, the temperature change with irradiation time was investigated for the probe at different concentrations (Fig. 7d). In addition, in vivo experiments were conducted on subcutaneous tumour-bearing mice that had been injected with CCM-mAb-ICG (1 mg/kg). The mice were irradiated with the same power of an 808-nm laser, and the temperature changes at the tumour site were recorded after 60, 120, 180, 240, and 300 s of irradiation (Fig. 7b). The results showed that the probe could elevate the temperature of the tumour tissue to 50.3 °C after 300 s of laser irradiation in vivo. The photothermal effect of the probe was verified in both in vitro and in vivo environments, and the next step was to validate the effect of the probe on tumour tissue in vivo. Firstly, the mice were randomly divided into three groups (n = 5) and subcutaneous tumour models were established on the same day. The experiment was conducted according to the procedure shown in Fig. 7a, where the mice in the PBS group were intravenously injected with PBS solution, the probe group received intravenous injection of the probe (1 mg/kg), and the ICG group received intravenous injection of ICG (1 mg/kg). The tumour size was recorded from the 10th day after tumour cell implantation when the tumour diameter was approximately 1 mm. The tumour size was observed until the 18th day (Fig. 7g), and then the tumours were dissected (Fig. 7f). Based on these results, photothermal treatment with the probe can inhibit tumour growth, and the ICG group showed better results than the PBS group due to the EPR effect of ICG. Furthermore, important organs were sliced after injection of the probe, and histopathological results confirmed that the probe had no tissue toxicity (Fig. 7e). To evaluate the impact of the probe's photothermal effect on the survival of subcutaneous tumour-bearing mice, a subcutaneous tumour model was established and randomly divided into PBS, probe, and ICG groups (n = 6). Photothermal treatment was performed following the procedure shown in Fig. 7a, and the survival time of the mice was observed and the survival curve was plotted (Fig. 7h).

Fig. 7.

In vitro and in vivo photothermal effects. (a) Visual representation of the photothermal therapy process. (b) In vivo validation of the photothermal therapy performance of the probe. (c) In vitro validation of photothermal therapy performance of the probe. (d) Photothermal therapy performance of the probe at different concentrations. (e) Histopathological results of vital organs after probe injection. scale bar: 100 μm. (f) Ex vivo tumour size. (g) Tumour growth curves after photothermal therapy with different probes. (h) Survival curves of mice after photothermal therapy with different probes.

Discussion

FGS is becoming increasingly prevalent in clinical practice; however, most studies have concentrated on the NIR-I window (700–900 nm) with restricted imaging resolution and depth.⁸ The advantages of NIR-II imaging include reduced scattering, minimal absorption, and negligible autofluorescence.^40,41

GBM is the most common central nervous system tumour in clinical practice, and patients have a poor prognosis. Previously, experienced clinicians relied on preoperative imaging to determine the approximate tumour location and boundary, with refinement during surgery based on visual and tactile feedback. While various auxiliary methods are currently used in clinical practice to determine tumour boundaries, these methods are not ideal owing to limitations such as the inability to provide real-time assistance to surgeons during surgery or the requirement for specialised equipment that is not widely available. Hence, there is a pressing need to develop a novel probe that can specifically bind to tumours, with binding sites that do not exist in normal brain tissue, and that can cross the BBB. To address these needs, we designed a probe that targeted MCT4 expressed on the surface of glioblastoma cells.

Confocal microscopy demonstrated that our designed probe exhibited excellent tumour cell targeting at the cellular level. In vivo imaging using NIR-II technology in mice demonstrated that the probe allowed specific imaging of the tumour, with an SBR of 2.8. FGS using the probe led to postoperative pathological results showing no residual tumour cells in the brain tissue and an SBR of 6.3 without scalp or skull obstruction. HE staining of brain tissue revealed fluorescent signals present in the tumour, but absent in normal brain tissue. Under high magnification, the fluorescent signal was distributed on the tumour cell membrane. The results from the transwell assay indicated that CCM-mAb-ICG had a strong ability to penetrate the BBB, whereas mAb-ICG had a lower ability, and ICG hardly penetrated the BBB. The photothermal therapy results showed that our designed probe elevated the temperature up to 50 °C after 5 min of laser excitation in vitro. Lastly, photothermal therapy significantly reduced the tumour volume and prolonged the survival time of the mice. The pathological results confirmed that the probe did not damage vital organs in the mice.

This study has some limitations. The probe designed in this study was encapsulated in the tumour cell membrane, allowing it to penetrate the BBB. However, this raises ethical concerns regarding its clinical application. Currently, only a few substances specifically bind to MCT4. However, whether these substances could be used to modify dyes remains to be determined. Due to experimental constraints, we temporarily utilised the U87 model. We recognise that the U87 tumour model may not be as compelling as the PDX model. Our future research will involve the use of the PDX model to assess the imaging capabilities of the probe.

FGS can enable tumour illumination in robot-assisted surgeries. Surgical robots have numerous advantages, and an increasing number of researchers have reported remarkable progress in this field.42, 43, 44, 45 In the future, robotic surgeries will replace human surgeries,⁴⁶ as robots possess higher degrees of freedom in their “fingers,” allowing them to perform more complex operations. Robotic “hands” provide greater stability, resulting in more precise surgeries.⁴⁷ However, autonomous surgery requires methods that enable robots to identify tumours independently. NIR-II FGS navigation may be the optimal surgical approach to enable robots to autonomously remove tumours.

In recent research, numerous research teams have identified the potential of MCT4 as an imaging target,⁴⁸ while others propose its utility as a therapeutic target.⁴⁹ Furthermore, various MCT4 ligands have been discovered in recent studies,^50,51 collectively indicating the emerging significance of MCT4 as a pivotal target in cancer therapy.

In conclusion, our findings highlight the potential efficacy of the CCM-mAb-ICG probe for FGS and therapeutic interventions in glioma treatment. Our results demonstrated the superior ability of this combination probe to cross the BBB to target the tumour. Future studies are needed to refine this technology and confirm its effects for additional targets and cancer types, as well as to find solutions to address ethical concerns and ensure the safety of this technology.

Contributors

Study design: Tian Jie, Zhenhua Hu, Yufei Gao, Xiaohua Jia.

Literature search: Hongyang Zhao, Jinnan Zhang, Chunzhao Li, Xiaojing Shi.

Data collection: Hongyang Zhao, Jinnan Zhang, Xiaojing Shi.

Data analysis: all authors.

Data interpretation: all authors.

Figures: Hongyang Zhao, Xiaojing Shi.

Writing: Hongyang Zhao, Chunzhao Li.

Revision: all authors.

Access and verification of the data: all authors.

Responsibility for the decision to submit the manuscript: all authors.

All authors read and approved the final version of the manuscript.

Verified the underlying data: all authors.

Data sharing statement

The main data supporting the results of this study are available in the paper and Supplementary Material. Raw data can be obtained by asking the corresponding author to clarify the purpose of their use.

Declaration of interests

The authors declare that they have no conflicts of interest.