Propofol Disrupts Aerobic Glycolysis in Colorectal Cancer Cells via Inactivation of the NMDAR-CAMKII-ERK Pathway

Colorectal cancer is the most common cancer of the digestive system and the third most common cause of cancer-related death globally, accounting for approximately 1.3 million new cases and 50, 000 deaths per year [1]. With the introduction of sigmoidoscopy and colonoscopy, both the incidence and death rate of colorectal cancer declined in the USA and high-income countries. However, due to the adoption of dietary patterns and risk factors associated with a western lifestyle, its incidence has increased in previously low-risk countries, such as countries in Eastern Europe and East Asia [1‒3].

Approximately 80% of patients are diagnosed with colorectal cancer at a local stage. Surgery is the standard protocol for the treatment of regional and liver metastasis tumors [3, 4]. However, local recurrence and distant metastases remain serious issues after surgery [5]. Despite advances in neoadjuvant therapy, the 5-year relative survival rate for colorectal cancer remains at less than 50% in low-income countries. Epidemiological studies have consistently shown an interaction between the risk of colorectal cancer and various factors, such as inflammatory bowel disease [6], obesity [7], excessive alcohol consumption [8], and diabetes [9]. Although genome-wide association studies have identified multiple factors related to the progression of colorectal cancer, the molecular mechanisms underlying its development are still understood poorly.

Cancer cells reprogram their metabolic type to meet the demands of proliferating cells by shifting from oxidative phosphorylation to aerobic glycolysis, which is known as the Warburg effect [10]. The increased uptake of glucose and glycolytic rate as a result of this shift favor cancer cells by providing energy and intermediates for biosynthesis. The activation of hypoxia-inducible factor 1α (HIF1α) is one of the main mechanisms of aerobic glycolytic metabolism [11]. HIF1α stimulates aerobic glycolysis by inducing the expression of glucose transporters (such as glucose transporter 1, GLUT1) and directly targeting enzymes that regulate glycolysis, such as hexokinase 2 (HK2), phosphoglycerate kinase 1 (PGK1), lactate dehydrogenase A (LDHA), monocarboxylate transporter 4, and pyruvate dehydrogenase lipoamide kinase isozyme 1 (PDK1) [12]. A number of studies have revealed the importance of this metabolic reprogramming in cancer cells [13‒17].

Surgery for malignant cancers aims at eliminating cancer cells; however, it may also promote the distant spread of tumor cells, leading to metastases [18]. Recent discoveries have suggested that anesthesia could induce metabolic, inflammatory, and immunological changes in the perioperative period [19]. Propofol (2, 6-diisopropylphenyl) is a widely used agent during the induction and maintenance of anesthesia. It has been revealed recently that propofol exerts antitumor effects against some malignancies via inhibition of cell viability and the dispersion and invasion capacity of cancer cells [20‒22]. However, its effect on glucose metabolism in cancer cells is not well understood. Therefore, in this study we investigated the effect of propofol on glucose metabolism using HT29 and SW480 colorectal cancer cells, and explored the potential mechanism by which propofol suppresses aerobic glycolysis.

Human HT29 (ATCC®HTB-38^TM) and SW480 (ATCC®CCL-228^TM) colorectal cancer cells, murine CT26 colon cancer cell lines (ATCC®CRL-2368^tm), human colon mucosal epithelial NCM460 cells, and human umbilical vascular endothelial cells (HUVECs) were purchased from the American Type Culture Collection. HT29, SW480, and CT26 cell lines were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Hyclone, Australia); the NCM460 cell line was cultured in RPMI 1640 medium (Hyclone, Australia); and HUVECs were cultured in DMEM (5 mM glucose). All culture media were supplemented with 10% fetal bovine serum (Gibco, Australia), 100 U/mL penicillin, and 100 mg/mL streptomycin, with the cells maintained in a humidified 5% CO₂ atmosphere at 37°C. Cell medium was changed every day, and the cells were passaged until they reached 80–90% confluence. Cells at passages 4–10 were used in this study.

Propofol (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich). KN93, an inhibitor of calcium/calmodulin-dependent protein kinase II (CAMKII), and dizocilpine, also known as MK801, an uncompetitive antagonist of the N-methyl-D-aspartate receptor (NMDAR), were purchased from Selleck (Shanghai, China). Rapastinel, a selective, partial agonist of NMDARs, was purchased from Sigma-Aldrich. KN93, MK801, and rapastinel were dissolved in DMSO, respectively.

Colorectal cancer RNASeqV2 data and clinical data were downloaded using TCGA Assembler [23]. Patients were classified into low or high gene expression groups using median expression calculated by Cox regression analysis. Survival differences between each group were assessed by Kaplan-Meier methods and compared using the log-rank test. All statistical tests were two-sided and performed with SPSS software.

During the induction and maintenance of anesthesia, the plasma concentration of propofol ranges from 5 to 50 µM [24]. In this study, propofol was administered to the cells in a concentration gradient (5, 25, 50, and 100 µM) for 4 h. DMSO (<0.1%) was administered to the cells as a control group in extracellular acidification rate (ECAR) analysis. The concentration of propofol that had a significant inhibitory effect on glucose metabolism was determined in the ECAR assay, and then used in the following experiments. Since DMSO had no effect on ECAR compared with the blank control group, this group was not included in the following experiments. KN93 (10 µM) and MK801 (500 µM) were administered to the indicated groups, respectively. Propofol (50 µM) and rapastinel (50 µM) were administered to the cells simultaneously.

Briefly, the cells were collected and solubilized with RIPA lysis buffer (Beyotime Biotechnology, Suzhou, China) supplemented with protease and phosphatase inhibitors (Roche, Rotkreuz, Switzerland). The lysates were mixed with 5× loading buffer and boiled for 5 min at 100°C. The proteins were resolved in a 10% SDS-polyacrylamide gel and transferred to a polyvinylidene fluoride membrane (Millipore, Billerica, MA) using a semidry electroblotting system. Then, the membranes were blocked with 5% skim milk in phosphate-buffered saline-Tween 20 (PBST) for 1 h at room temperature, followed by incubation with a 1: 1, 000 dilution of the primary antibody at 4°C overnight. The membranes were then washed with PBST and incubated with a 1: 5, 000 dilution of goat anti-rabbit IgG (H+L), HRP conjugate (Proteintech, Wuhan, China) for 1 h at room temperature. After washing, the signals were detected using a LAS-4000 mini CCD camera (GE Healthcare).

Total RNA was extracted using TRNzol (Tiangen, Beijing, China). Chloroform was added and the lysates were centrifuged at 12, 000 × g for 15 min at 4°C. The supernatant was transferred to a new centrifuge tube. RNA was precipitated with 100% ethanol. The samples were rinsed with 75% DEPC-alcohol. Each RNA sample was diluted in 30 µL RNase-free water. The quantity and quality of RNA were assessed using a NanoDrop 2000c Spectrophotometer (Thermo Fisher, USA).

cDNA was synthesized by reverse transcription using a TaKaRa PrimeScript® RT Reagent Kit with gDNA eraser. Briefly, a mixture containing 1 µg RNA, 2.0 µL 5× gDNA eraser buffer, and RNase-free water up to 10 µL was incubated at 42°C for 2 min. Then, 4.0 µL of 5× PrimeScript® Buffer2, 1.0 µL PrimeScript®RT Enzyme Mix I, 1.0 µL RT Primer Mix, and RNase-free water up to 20 µL were added to the mixture. The RT reaction mixture had a final volume of 20 µL and was incubated at 42°C for 15 min and 85°C for 5 s. The expression of candidate genes and β-actin was determined by qRT-PCR. The PCR was run using a QuantStudio 7 Flex RealTime PCR System (Applied Biosystems) starting with denaturation at 95°C for 30 s followed by 40 cycles of 95°C for 5 s, 60°C for 30 s (annealing), a terminal extension step at 95°C for 10 s, and a final holding stage at 4°C. Melt curve controls were run to ensure primer specificity. All of the reactions were run in triplicate. The primer sequences used are listed in Table 1.

Mitochondrial function was measured using a Seahorse Bioscience XF96 Extracellular Flux Analyzer according to the manufacturer’s instructions. Briefly, the cells were seeded at a density of 2.0 × 10⁴ cells per well in a 96-well cell culture XF microplate for 24 h prior to assay. The cells were switched to Seahorse buffer (DMEM with phenol red containing 25 mmol/L glucose, 2 mmol/L sodium pyruvate, and 2 mmol/L glutamine) at 1 h prior to assay. Then, 25 µL each of 10 mmol/L glucose, 1 µmol/L oligomycin, and 100 mmol/L 2-deoxy-glucose were added to measure the ECAR. Each measurement cycle consisted of a mixing time of 3 min and a data acquisition period of 3 min for the XF96. Glucose stimulation indicates the cell glycolysis rate and oligomycin inhibition indicates the cell glycolysis capacity. ECAR levels were calculated after normalization to cell number and are plotted as mean ± standard deviation (SD).

Intracellular Ca²⁺ measurements were performed using a Fluo-3 AM Fluorescence Kit (Beyotime, China). The cells were seeded at a density of 2.0 × 10⁶ cells per well in a 6-well plate. After treatment, the cells were washed with PBS prior to the addition of the Fluo-3 AM fluorescence probes (0.5 µM) to each well. The cells were then incubated with the probes for 60 min at 37°C protected from light. After incubation, the cells were washed with serum-free medium 3 times to remove extracellular Fluo-3 AM that might have entered the cells. The intensity of Fluo-3 AM was assessed using flow cytometry at 488 nm according to the manufacturer’s instructions. Intracellular Ca²⁺ concentration was proportional to fluorescence intensity. Each assay was performed in three replicates, and the levels of intracellular Ca²⁺ are presented as mean ± SD.

BALB/c mice (female, 4–6 weeks of age, 18–20 g; Shanghai SLAC Laboratory Animal Co., Ltd.) were housed under pathogen-free conditions and given food and acidified water ad libitum. Twenty-five mice were divided randomly into 5 groups. CT26 cells were treated with propofol (50 µM), KN93 (10 µM), MK801 (500 µM), propofol (50 µM) with rapastinel (50 µM), or DMEM, as a control group, for 4 h. The mice of each group were injected subcutaneously with 2.0 × 10⁵ CT26 cells in 100 µL PBS. At 2 weeks after tumor implantation, the mice were subjected to micro-positron emission tomography/computed tomography (microPET/CT) before they were euthanized, and the tumors were then dissected, photographed, and subjected to qRT-PCR and western blot analyses. Data are presented as the mean ± standard error (SE) or SD. All animal experiments were performed according to the Guidelines for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of Fudan University.

MicroPET/CT imaging was performed according to the manufacturer’s instructions. Briefly, ¹⁸F-FDG was produced in a cyclotron (Siemens CTI RDS Eclipse ST) using an Explora FDG4 module. Each mouse was injected with 11.1 MBq (300 µCi) of ¹⁸F-FDG through the tail vein. Scanning started at 1 h after injection. Images were reconstructed using a three-dimensional ordered-subset expectation maximization/maximum algorithm. An Inveon^TM Research Workplace was used to obtain the percentage injected dose per gram (%ID/g) and standardized uptake values (SUVs). SUVmax was calculated.

Data were obtained from at least 5 independent experiments and expressed as the mean ± SD. Groups were compared using one-way analysis of variance followed by pairwise comparisons between groups with Dunnett’s t test. On the basis of our previous experiments, we set α at 0.05, power at 0.8, to achieve a moderate effect size (0.5 = moderate [1/2 of an SD]), and a minimum sample size of 5 mice in each group. Overall survival and disease-free survival curves were calculated according to the Kaplan-Meier method, and comparisons were performed using the log-rank test. Univariate and multivariate Cox regression analyses were conducted to explore the independent predictors of prognosis. SPSS and GraphPad Prism software version 6.0 (GraphPad Software, Inc., La Jolla, CA, USA) were used for statistical analyses, and differences were considered significant at P < 0.05.

Cancer cells reprogram their metabolic pathway to glycolysis (Fig. 1A). First, we examined glucose uptake and lactate production by performing ECAR analysis. HT29 and SW480 cells were exposed to different concentrations of propofol (5, 25, 50, and 100 µM) for 4 h. As shown in Fig. 1B, propofol decreased glycolytic capacity in a dose-dependent manner. This decrease was significant at concentrations of 50 and 100 µM propofol. Therefore, we chose 50 µM for the following experiments. Moreover, no significant effect was observed in DMSO-treated cells. To assess further the role of propofol in glucose metabolism, key enzymes of the glycolysis cascade were examined. Enzymes related to glycolysis, such as GLUT1, HK2, PGK1, and LDHA, were decreased both at the mRNA and protein levels (Fig. 1C and 1D). To validate the specific inhibitory role of propofol on cancer cells, we examined its effect on NCM460 and HUVECs. Fig. 1E, 1F and 1G showed that propofol had little inhibitory effect on the glucose metabolism of normal cells.

Activation of the HIF1α transcription factor drives the metabolic changes necessary to increase glycolysis in tumor cells [12]. By recognizing hypoxia-responsive elements in the promoters of target genes, HIF1α directly regulates the enzymes necessary for glycolysis. Therefore, we investigated the expression of HIF1α in HT29 and SW480 cells following exposure to different concentrations of propofol (5, 25, 50, and 100 µM) for 4 h. Propofol reduced HIF1α expression in a dose-dependent manner (Fig. 2A).

In the central nervous system, CAMKII is best known for its role in learning and memory through Ca²⁺ signaling. However, recent studies have established critical roles for CAMKII in cancer progression. In cancer cells, remodeling of Ca²⁺ signaling helps to sustain most of the hallmarks of cancer. We next examined the effect of propofol on intracellular Ca²⁺ concentration and CAMKII phosphorylation. Treatment with a concentration gradient of propofol resulted in a dose-dependent decrease in intracellular Ca²⁺ concentration (Fig. 2B and 2C) and CAMKII phosphorylation (Fig. 2A). However, the total expression of CAMKII was not altered.

CAMKII activation reportedly leads to extracellular signal-regulated kinase (ERK) activation and cell proliferation, and ERK activation is essential for HIF1α activation [25]. Thus, we determined the level of phosphorylated (p)-ERK after exposure to propofol. As shown in Fig. 2A, ERK1/2 phosphorylation was decreased in propofol-treated cells in a dose-dependent manner.

Previous studies have shown that propofol inhibits NMDAR, which maintains Ca²⁺ flux and was recently found to be involved in cancer progression. Therefore, we hypothesized that NMDAR signaling is a key mechanism in propofol-induced glycolysis inhibition. To elucidate the NMDAR-CAMKII-ERK-HIF1α signaling pathway, MK801 (500 µM), an NMDAR antagonist, and KN93 (10 µM), a CAMKII antagonist, were administered to cells separately to mimic the effect of propofol. Compared with the control group, MK801 and KN93 induced a significant decrease in glycolytic capacity (Fig. 3A), and the mRNA and protein levels of glycolytic enzymes were also reduced, which was consistent with the effect of propofol (Fig. 3B and 3C).

To assess further the role of NMDAR in propofol-induced glucose metabolism inhibition, we used an agonist of NMDAR, rapastinel, and evaluated the co-effect of rapastinel and propofol on glucose metabolism. Rapastinel (50 µM) was administered to the cells together with propofol. We observed that NMDAR activation reversed the effect of propofol on glycolytic inhibition (Fig. 3A). Furthermore, the protein and mRNA levels of enzymes involved in glycolysis were also elevated in the co-treatment group compared with the propofol-treatment alone group (Fig. 3B and 3C).

The influx of extracellular Ca²⁺ through NMDAR is required for cell signal transduction. To determine how these signaling events contribute to the maintenance of glycolytic metabolism in HT29 and SW480 cells, we studied the effect of MK801, KN93, and rapastinel on intracellular Ca²⁺ levels and the levels of p-CAMKII and p-ERK. Flow cytometry analysis demonstrated that MK801 significantly reduced intracellular Ca²⁺ levels, as propofol did, while rapastinel attenuated the inhibitory effect of propofol (Fig. 4A). KN93 showed no effect on intracellular Ca²⁺ levels (Fig. 4A). Next, immunoblots showed that, consistent with propofol, MK801 and KN93 repressed the levels of p-CAMKII, p-ERK, and HIF1α, while rapastinel reversed the effect of propofol (Fig. 4B).

Having established an important role for propofol in glycolysis in vitro, we next examined its effect in vivo. To establish a xenograft model, we first administered propofol, MK801, KN93, rapastinel+propofol, or DMEM to murine CT26 colon cells, consistent with the in vitro treatments. Twenty-five female BALB/c mice were randomly divided into 5 groups. Then, the 5 groups of differently treated CT26 cells were injected subcutaneously into the right flanks of the 5 randomly divided groups. Compared with the control group, tumor volume and tumor growth were reduced significantly in the propofol, MK801, and KN93 groups, while tumor growth was increased in the rapastinel+propofol group compared with the propofol group (Fig. 5A). We further determined glucose metabolism in tumor-bearing mice using a small animal imaging system (Fig. 5B). Fig. 5C shows that, compared with the control group, ¹⁸F-FDG uptake was significantly decreased in the propofol, MK801, and KN93 groups, while rapastinel reversed the inhibition of ¹⁸F-FDG uptake by propofol. Subsequent qRT-PCR and immunoblot analyses of tumor tissues demonstrated that, compared with the control group, the mRNA and protein expression levels of the glycolytic enzymes GLUT1, HK2, PGK1, and LDHA were also decreased in tissues from the propofol, MK801, and KN93 groups, while they were increased in the rapastinel+propofol group, compared with propofol group (Fig. 5D and 5E).

To detect the potential clinical meaning of genes involved in glycolysis, we obtained 381 colorectal patients from TCGA and classified these patients into low or high gene expression using median expression calculated by Cox regression analysis. Consistent with our findings in vivo and in vitro, Kaplan-Meier analysis showed that the high expression of GLUT1 and CAMKIIb was significantly correlated with a poorer prognosis in colorectal patients (Fig. 6A and 6E). High PGK1 expression was also correlated with a poorer prognosis, but it did not reach the level of significance (Fig. 6C). Patients with high expression of GLUT1, PGK1, CAMKIIb, and ERK had significantly lower disease-free survival than those with low expression (Fig. 6B, 6D, 6F, and 6G). Taken together, these results suggested that the high expression of genes involved in glycolysis has a positive correlation with a poorer prognosis in colorectal patients. Propofol was suggested to inhibit the expression of genes involved in glycolysis in our previous experiments; thus, there is a high possibility that propofol might be beneficial for the treatment of patients with colorectal cancer.

In the present study, we provided evidence that propofol is involved in glucose metabolism in colorectal cancer cells, and the administration of propofol resulted in a reduced capacity for aerobic glycolysis. Moreover, Q-PCR and western blot analysis indicated that propofol inhibited the expression of the main enzymes in glycolysis, such as GLUT1, HK2, PGK1, and LDHA, at both the mRNA and protein levels. Then, we confirmed these results in an in vivo xenograft model. Using a microPET/CT imaging system, we observed that glucose metabolism was reduced in the propofol group, with decreased uptake of ¹⁸F-FDG. To elucidate further the specific inhibitory role of propofol on cancer cells, we examined its effect on NCM460, a normal human colon mucosal epithelial cell line, and HUVECs. We found that propofol had little inhibitory effect on the glucose metabolism of normal cells. Previous studies showed that propofol deceases the metastasis and progression of many cancers, including pancreatic cancer [26], breast cancer [27], colon cancer [28], gastric cancer [22], and cervical cancer [29]. However, the specific role of propofol in glucose metabolism in cancer cells remains poorly understood. Here, in the present study, we demonstrated that propofol has an inhibitory role in aerobic glycolysis.

It is well established that hypoxia exists in many tumors, especially solid tumors, for rapidly proliferating cancer cells. Metabolic changes in tumor cells in response to hypoxia enhance glycolytic capacity and decrease mitochondrial respiration. HIF1α activation is one of the principal mechanisms driving glycolysis in tumor cells, by directly or indirectly targeting the glucose transporters and enzymes involved in glycolysis. As reported previously, propofol can inhibit the activity of HIF1α and thus attenuate the invasive ability of HIF1α-promoting malignancies [30]. Our data also confirmed that propofol reduced HIF1α expression in HT29 and SW480 colorectal cancer cells in a dose-dependent manner. A number of studies have demonstrated that HIF1α expression is regulated by major signaling pathways, including the ERK pathways [25]; therefore, we examined the effect of propofol on ERK activity. As expected, propofol significantly repressed the phosphorylation of ERK. The Ca²⁺ signaling pathway is also reportedly involved in tumorigenesis. Alterations of Ca²⁺ levels in cancer cells help to sustain the hallmarks of cancer by remodeling the proliferation, metastasis, and vascularization of tumor cells [31]. As a direct effector of Ca²⁺ signaling, CAMKII in turn regulates the activation of ERK pathways to facilitate tumor progression [31, 32]. Consistent with previous studies, we demonstrated that propofol reduced Ca²⁺ levels in cancer cells and reduced the levels of p-CAMKII and p-ERK.

There is accumulating evidence suggesting that surgical stress, volatile anesthetics, and blood transfusion may facilitate tumor growth. Moreover, investigations on intravenous anesthetics have discovered that propofol has an opposite effect to inhaled anesthetics, i.e., suppressing metastasis and promoting apoptosis of tumor cells [30, 33‒35]. A recent study found that propofol inhibited the activation of NMDAR, a Ca²⁺ channel [36]. NMDAR, with high permeability to Ca²⁺, plays a crucial role in synaptic transmission and synaptic plasticity in the nervous system [37, 38]. A growing body of evidence indicates that NMDAR is also expressed in a wide range of non-neuronal cells, including cancer cells [39, 40]. To determine whether propofol decreased the levels of cellular Ca²⁺ through NMDAR inactivation, we used MK801, an antagonist of NMDAR. Our data showed that MK801 mimicked the inhibitory effect of propofol. KN93, an antagonist of CAMKII, also had the same effect as propofol. As a target of Ca²⁺, CAMKII inhibition through KN93 had no effect on intracellular Ca²⁺ levels. To investigate further the role of propofol on NMDAR, we used rapastinel, an agonist of NMDAR. As expected, rapastinel reversed the inhibitory effect of propofol on glycolytic metabolism. To examine further the clinical correlation of glycolysis, we used 381 patients obtained from TCGA, and classified them into low and high gene expression groups. Kaplan-Meier curve analysis indicated that the high expression of genes involved in glycolysis was positively correlated with a poorer prognosis and disease-free survival in colorectal patients. These results suggested a beneficial role for propofol in the treatment of patients with colorectal cancer.

The current study has some limitations that should be discussed. We confirmed that propofol inhibited aerobic glycolysis in HT29 and SW480 colorectal cancer cells through the suppression of the NMDAR-CAMKII-ERK pathway. Though the physiological roles of NMDAR in the central nervous system have been elucidated extensively, its role as a channel in cancer cells is not well understood. In our study, we used an antagonist of NMDAR, MK801, to verify our hypothesis that propofol inhibited the activity of NMDAR, leading to reduced intracellular Ca²⁺ levels, which is not an ideal approach. In addition, NMDAR contain several subunits, and further studies are needed to investigate the specific subunits that propofol may affect. Propofol is used widely during general anesthesia in surgery. However, in the context of surgery, multiple anesthetics are used. Here, it is difficult to study the mechanisms of propofol when the patients have been exposed to other agents. However, the results of our study may still influence the understanding of anesthetics in some way, although further experiments are needed to understand the co-effect of anesthetics used together.

Our study revealed that propofol has an inhibitory role in the glycolysis of colorectal cancer cells. Our results suggested that propofol inhibited the expression of glucose transporters and enzymes (GLUT1, HK2, LDHA, and PDK1) that are crucial in glucose metabolism. It further supported the hypothesis that HIF1α, a master transcription factor controlling glucose metabolism, is also reduced by propofol in a dose-dependent manner. In addition, we found that propofol repressed the NMDAR-CAMKII-ERK pathway to inactivate HIF1α and therefore reduce glycolysis (Fig. 5F). The xenograft model also confirmed the reduction of glucose metabolism and tumor growth by propofol. Patients derived from TCGA with high expression of genes involved in glycolysis had a lower overall survival and disease-free survival than those with low gene expression, suggesting a potential role for propofol in the treatment of cancer patients. Therefore, the results of our study may contribute to the use of anesthetics in the clinical setting.

This work was supported by the Shanghai Charity Cancer Center (No. HYXH1407). We thank Yi Qin, Ph.D. (Pancreatic Cancer Institute, Fudan University Shanghai Cancer Center, Shanghai, China) for his assistance with technical aspects of performing the experiments, and Yingjian Zhang (Head of Nuclear Medicine Department, Fudan University Shanghai Cancer Center, Shanghai, China) for his assistance in microPET/CT imaging and analysis.

No conflict of interests exists.

X. Chen, Q. Wu and P. Sun contributed equally to this work.