Angiopoietin-like 4 promotes glucose metabolism by regulating glucose transporter expression in colorectal cancer

Angiopoietin-like 4 (ANGPTL4) was recently shown to be associated with cancer progression but little is known about its contribution to cancer metabolism. The purpose of this study was to elucidate the role of ANGPTL4 in glucose metabolism in colorectal cancer (CRC). Immunohistochemical staining of CRC specimens classified 84 patients into two groups according to ANGPTL4 expression. Clinicopathological characteristics, gene mutation status obtained by next-generation sequencing, and fluorodeoxyglucose (FDG) uptake measured by positron emission tomography/computed tomography (PET/CT) were compared between the two groups. Furthermore, the impact of ANGPTL4 expression on cancer metabolism was investigated by a subcutaneous xenograft mouse model using the ANGPTL4 knockout CRC cell line, and glucose transporter (GLUT) expression was evaluated. There were significantly more cases of T3/4 tumours (94.3% vs. 57.1%, P < 0.001) and perineural invasion (42.9% vs. 22.4%, P = 0.046) in the ANGPTL4-high group than in the low group. Genetic exploration revealed a higher frequency of KRAS mutation (54.3% vs. 22.4%, P = 0.003) in the ANGPTL4-high tumours. All the FDG uptake parameters were significantly higher in ANGPTL4-high tumours. In vivo analysis showed a significant reduction in tumour size due to ANGPTL4 knockout with lower expression of GLUT1 and GLUT3, and suppression of AKT phosphorylation. ANGPTL4 regulates the expression of GLUTs by activating the PI3K–AKT pathway and thereby promoting glucose metabolism in CRC. These findings establish a new functional role of ANGPTL4 in cancer progression and lay the foundation for developing a novel therapeutic target.


Introduction
Angiopoietin-like 4 (ANGPTL4) is a secreted member of the angiopoietin-like protein family, which includes ANGPTL1-7 (Hato et al. 2008). Structurally, it contains a C-terminal fibrinogen-like domain (cANGPTL4) and an N-terminal coiled-coil fragment (nANGPTL4). In recent years, it has been reported that cANGPTL4 participates in cancer progression by elevating adenylate energy charge or activating the ERK pathway (Baba et al. 2017;Teo et al. 2017;Zhu et al. 2016). In fact, high ANGPTL4 expression is associated with a poor prognosis in some solid cancers, such as gastric, lung, and breast cancer (Baba et al. 2017;Cai et al. 2020;Chen et al. 2018;Zhu et al. 2016). In colorectal cancer (CRC), one report using immunostaining of human CRC specimens showed that postoperative distant metastasis is significantly correlated with high ANGPTL4 expression (Nakayama et al. 2011). However, the detailed mechanism by which ANGPTL4 contributes to CRC progression has not been fully studied.
Physiologically, ANGPTL4 is known to be associated with glucose metabolism and lipid metabolism, and its overexpression has been reported to cause impaired glucose metabolism or hyperlipidaemia (Barchetta et al. 2020;Yoshida et al. 2002). Both tissue expression and serum concentration could be useful markers in clinical practice. In fact, serum ANGPTL4 values of diabetes patients are reported to be significantly higher than those of nondiabetic patients (McCulloch et al. 2020). As such, ANGPTL4 is strongly related to glucose metabolism, although little is known about its contribution to cancer metabolism.
Glucose metabolism is essential for cancer cell proliferation. Enhanced glucose metabolism contributes to the progression of various cancers (and Weinberg 2011). Some of the glucose transporter (GLUT) subtypes have been shown to be important for cancer progression and have been Hanahan demonstrated to be prognostic factors (Szablewski et al. 2013). Especially for CRCs, high expression of GLUT1 or GLUT3 has been reported as a poor prognostic factor (Feng et al. 2017;Kim et al. 2019;Yang et al. 2017). Moreover, it has recently been shown that GLUT1 expression in cancer cells is related to the mutation status of KRAS, a major driver gene involved in CRC progression (Yun et al. 2009;Wan et al. 2019).
We herein showed that ANGPTL4 expression in CRCs is associated with the expression of some GLUTs and thereby contributes to promoting glucose metabolism. Our findings will lead to a better understanding of the role of ANGPTL4 in cancer progression and indicate a new potential target for cancer treatment.

Human samples
Patients who consecutively underwent surgical resection of CRCs at Keio University Hospital between April 2018 and December 2019 were enrolled in this study. Our inclusion criteria were as follows: pathologically proven primary CRC, preoperative positron emission tomography/computed tomography (PET/CT), and consent to undergo comprehensive genomic testing. According to these criteria, 84 patients were included. The Ethics Committee of Keio University Hospital approved the study, which was conducted in accordance with the Helsinki Declaration of 1996 (Approval Number: 20150051).
ANGPTL4 expression in human samples was evaluated using the Allred score, which is one of the most promising semiquantitative composite scoring systems (Yunokawa et al. 2017). In brief, the proportion and intensity scores were added to obtain a total score ranging from 0 to 8 (Allred et al. 1998). In this study, we randomly selected three fields for each slide and adopted the median scores. We defined an Allred score ≥ 6 as high expression of ANGPTL4.

Next-generation sequencing
In this study, we used an in-house next-generation sequencing (NGS) test that has been previously reported (Hayashi et al. 2020;Saotome et al. 2020). Briefly, genomic DNA was extracted from 10-µm-thick formalinfixed, paraffin-embedded (FFPE) tissue sections of tumour specimens, and then targeted amplicon exome sequencing was performed using a 160 cancer-related gene panel as previously described (Hayashi et al. 2020;Saotome et al. 2020). The targeted regions of all 160 genes were specifically enriched using oligonucleotide probes. The enriched libraries were sequenced with a paired-end (150 bp × 2) sequencing method using the NextSeq sequencing platform (Illumina, San Diego, CA, USA), resulting in a mean depth of 500. The sequencing data were analysed using the GenomeJack bioinformatics pipeline (Mitsubishi Space Software Co., Ltd., Tokyo, Japan) (http:// genom ejack. net/) as previously described (Nakamura et al. 2021). In this study, an actionable gene alteration was defined as a genomic variation that has an established biological role in cancer.

PI3K inhibitor treatment in vitro
The NC that were grown to 90% confluence were treated with a PI3K inhibitor (LY294002, Selleck, Houston, TX, USA) with a concentration of 10 μM and incubated for 8 h in complete medium. Cells were then collected for Western blot analysis.

Cell proliferation assay
Cell proliferation was determined by the MTT assay method. NC and KO were seeded into 96-well plates, respectively (1 × 10 4 cells/well) and incubated for 24, 48, 72, and 96 h. Viable cells were stained using the Cell Count Reagent SF (Nacalai Tesque, Kyoto, Japan) and identified by measuring the absorbance at 450 and 600 nm with a spectrophotometer, Infinite M Plex (Tecan, Männedorf, canton of Zürich, Switzerland).

Animals
We used BALB/c nude female mice, which were purchased from CLEA Japan (Tokyo, Japan). The mice were allowed to acclimate for at least 7 days before use and were 8 weeks at the start of the experimental protocol. The mice were housed in a controlled environment in the laboratory at the Keio University School of Medicine under standard temperature and light and dark cycles. Ethics approval for the present study was provided by the Ethics Committee at the Laboratory Animal Care and Use Committee at Keio University School of Medicine (Approval Number: 20062), and the study was performed in accordance with the Care and Use of Laboratory Animals (NIH).

Subcutaneous xenograft mouse model
The mice were anaesthetized with 2.0% isoflurane, and 1 × 10 7 NC or KO in a total volume of 100 μl phosphatebuffered saline (PBS) and 100 μl Matrigel (#35423, Corning, Corning, NY, USA) were injected subcutaneously using a 26-gauge needle (n = 5). All mice were sacrificed 2 weeks after subcutaneous injection, and tumour weights between the two groups were compared.

In silico analysis
The correlation between ANGPTL4 and SLC2A mRNA expression in CRC was evaluated using data from The Cancer Genome Atlas (TCGA) database. Coexpression analysis was performed, and the results were downloaded from the cBioPortal for Cancer Genomics (http:// www. cbiop ortal. org/).

Positron emission tomography/computed tomography
All patients fasted for at least 4 h prior to PET/CT imaging. Biograph mCT (Siemens Medical Solutions, Knoxville, TN) was used for scanning. Patients were administered 3.7 MBq/ kg fluorodeoxyglucose (FDG) and received routine PET/CT 1 h after administration. For the metabolic parameters, we measured the maximum standardized uptake value (SUVmax), the mean standardized uptake value (SUVmean), metabolic tumour volume (MTV), and total lesion glycolysis (TLG). MTV was defined as the sum of the primary tumour metabolic volumes using an SUVmax of 2.5 as the threshold, and TLG was defined as (SUVmean) × (MTV). All of these parameters were measured and calculated using Advantage Workstation software (GE Healthcare, Chicago, IL, USA).

Statistical analysis
All results are expressed as the median value (range). All statistical analyses were performed using Stata software (Stata Corporation, College Station, TX, USA). P values < 0.05 were considered to indicate statistical significance. Nonparametric Mann-Whitney U tests and χ 2 tests were used to compare two groups as appropriate.

High ANGPTL4 expression is associated with advanced pathological factors and KRAS mutation
As shown in Fig. 1, tumour samples were classified into ANGPTL4 high or low expression groups by IHC staining. Of the 84 samples, 49 (58%) samples were classified as ANGPTL4 low, and 35 (42%) samples were classified as ANGPTL4 high. The pathological characteristics of the two classified groups are detailed in Table 1. The comparison of Union for International Cancer Control (UICC) tumour-node-metastasis (TNM) classification showed that high ANGPTL4 expression was significantly associated with advanced stage. Especially for tumour invasion, a significantly higher number of T3/4 cases was found in the ANGPTL4-high group than in the ANGPTL4-low group (94.3% vs. 57.1%, P < 0.001). Both lymph node metastasis and distant metastasis tended to be more advanced in the ANGPTL4-high group, but the difference was not statistically significant. Of note, there was also a significant difference in perineural invasion (42.9% vs. 22.4%, P = 0.046). There were no significant differences in tumour location, histology, lymphatic invasion or venous invasion between the groups.
We next aimed to identify any genetic features of ANGPTL4-high tumours using NGS data. The comparison of the mutation status of extracted major driver genes is shown in Table 2. There were significantly more KRAS mutations in the ANGPTL4-high group than in the ANGPTL4-low group (54.3% vs. 22.4%, P = 0.003). Similarly, there were significantly more FBXW7 mutations in the ANGPTL4-high group (17.1% vs. 4.1%, P = 0.044). On the other hand, there were no significant differences in TP53, APC, SMAD4, PIC3CA, and BRAF mutations between the two groups.

ANGPTL4 knock-out attenuates tumour progression
As more advanced tumours were found in ANGPTL4-high cases in terms of tumour invasion and perineural invasion, we hypothesized that ANGPTL4 positively contributes to tumour progression. To address this hypothesis, we generated an ANGPTL4 knockout cell line using CRISPR/Cas9 gene editing ( Fig. 2A). We used the HCT116 CRC cell line, which expresses significantly higher levels of ANGPTL4 than other cell lines (Fig. S1). Of note, HCT116 is also known as a KRAS mutant cell line, which is considered to be better for characterizing human ANGPTL4-high tumours (Buck et al. 2007).
When the ANGPTL4 negative control cell line (NC) and knockout cell line (KO) were xenografted, the sizes of tumours generated from the KO were significantly reduced compared with those of the NC (Fig. 2B, C). The mean total volumes of the tumours showed an approximately 50% reduction (NC: 430 mg vs. KO: 196 mg, P = 0.032) (Fig. 2D). Notably, ANGPTL4 expression in KO tumours was diminished (Fig. 2E). Evaluation of tumour angiogenesis by CD31 staining showed that the KO tumours were more hypovascular than the NC tumours (Fig. 2F). Furthermore, the proliferation assay revealed that the proliferation of the KO was significantly suppressed when compared with that of the NC (Fig. 2G).

ANGPTL4 contributes to tumour progression by activating the PI3K-AKT pathway
ANGPTL4 is reported to activate the PI3K-AKT pathway in gastric cancer (Baba et al. 2017). To explore more details of the role of ANGPTL4 in CRC progression, we further analysed the cell lines we generated. As shown in Fig. 3A, phosphorylation of AKT was significantly downregulated in the KO, whereas phosphorylation of PI3K was not significant. We next validated these findings by immunohistochemical staining of xenograft tumours. As shown in Fig. 3B, phospho-AKT expression was markedly reduced in the KO tumours, whereas phospho-PI3K expression did not change. Taken together, these findings suggest that ANGPTL4 promotes tumour progression by activating the PI3K-AKT pathway via AKT phosphorylation.

ANGPTL4 regulates GLUT expression
According to the result that KRAS mutations were more common in the ANGPTL4-high group and the fact that ANGPTL4 physiologically contributes to glucose metabolism, we next aimed to investigate the correlation between ANGPTL4 and GLUT expression. As shown in Fig. 4A, analysis of the TCGA data set showed that the mRNA expression of ANGPTL4 was positively correlated with that of SLC2A1 and SLC2A3 (SLC2A1: r = 0.23, P < 0.001, SLC2A3: r = 0.49, P < 0.001). We next validated these results by evaluating the expression of GLUT1 and GLUT3 in our human samples. As shown in Fig. 4B, ANGPTL4high tumours showed higher expression of both GLUT1 and GLUT3 than ANGPTL4-low tumours. The results from TCGA data sets regarding other SLC2A subtypes are shown in Fig. S2. Of note, there were also significant positive To investigate whether ANGPTL4 expression directly impacts GLUT1 and GLUT3 expression, we utilized a knockout cell line. Western blot analysis showed that the expression of both GLUT1 and GLUT3 was significantly suppressed in the KO compared with the NC (Fig. 4C). Moreover, immunostaining of GLUT1 and GLUT3 in xenografted tumours showed a reduction in these GLUTs in the KO tumours (Fig. 4D). We next investigated whether the PI3K-AKT pathway activated by ANGPTL4 was involved in the upregulation of GLUTs. When the NC was treated with a PI3K inhibitor (LY294002), AKT phosphorylation was downregulated as indicated (Fig. S3A). Notably, at the same time, GLUT1 and GLUT3 expression was also downregulated (Fig. S3B). and knockout (KO) HCT116 cells. β-actin was analysed as a loading control. B Images of the mice xenografted with either NC or KO. C Representative images of the tumours generated from NC and KO mice. D Comparison of tumour weight between NC and KO mice (n = 5 each). Data are represented as the mean ± SD. *P < 0.05 E Rep-resentative images of ANGPTL4 expression in the tumours generated from NC and KO mice. F Representative images of CD31 expression in the tumours generated from NC and KO mice. The bar graph shows the proportion of the CD31-positive area in a single field, comparing NC and KO tumours. Data are shown as the mean ± SD from three independent experiments. ***P < 0.001. G Time course of cell proliferation for NC and KO. Data are shown as the mean ± SD from three independent experiments. ***P < 0.001 Taken together, these results indicate that ANGPTL4 expression is correlated with several GLUTs and directly influences the expression of GLUT1 and GLUT3 via the PI3K-AKT pathway.

ANGPTL4 expression is correlated with FDG uptake in human CRCs
To validate our in vitro and in vivo results, we next aimed to evaluate actual glucose uptake in CRC patients. We  Figure 5 shows the comparison of a variety of metabolic parameters between ANGPTL4-high and ANGPTL4-low tumours. The details are as follows: SUVmax (ANGPTL4 high vs. low = 13.28 vs. 7.56, P < 0.001), SUVmean (ANGPTL4 high vs. low = 5.33 vs. 3.85, P < 0.001),

Fig. 4 ANGPTL4 expression regulates GLUT proteins in CRC.
A Correlation of mRNA expression between ANGPTL4 and SLC2A1 and SLC2A3 according to TCGA database analysis. B Representative images of IHC staining of GLUT1 and GLUT3 in ANGPTL4high and ANGPTL4-low human CRCs. C Western blot analysis comparing the protein expression of GLUT1 and GLUT3 between ANGPTL4-negative control (NC) and knockout (KO) cells. β-actin was analysed as a loading control. D Representative images of GLUT1 and GLUT3 expression in the tumours generated from NC and KO mice MTV (ANGPTL4 high vs. low = 39.32 ml vs. 11.60 ml, P < 0.001), and TLG (ANGPTL4 high vs. low = 214.25 ml vs. 44.67 ml, P < 0.001). All parameters were significantly higher in the ANGPTL4-high group, strongly suggesting that ANGPTL4 expression is correlated with glucose metabolism in human CRCs.

Discussion
In the current study, we found that ANGPTL4 expression is related to the progression of CRC, as suggested by the increased levels of tumour invasion and perineural invasion in ANGPTL4-high tumours. Genetic exploration Black horizontal lines show the median values. ***P < 0.001. SUVmax maximum standardized uptake value, SUVmean mean standardized uptake value, MTV metabolic tumour volume, TLG total lesion glycolysis revealed that a higher rate of KRAS mutation was found in ANGPTL4-high tumours and was considered to contribute to cancer progression. Furthermore, the expression of several GLUT subtypes, such as GLUT1 and GLUT3, was positively correlated with ANGPTL4 expression, suggesting a pivotal role of ANGPTL4 in glucose metabolism. In fact, all the parameters of tumour FDG uptake measured by PET/CT, which reflect glucose metabolism, were significantly higher in ANGPTL4-high tumours.
In some solid cancers, a relationship between ANGPTL4 expression and cancer progression has been suggested. In scirrhous gastric cancer, ANGPTL4 is uniquely induced by hypoxia, and its expression is essential for tumour growth and resistance to anoikis through the activation of the prosurvival FAK/Src/PI3K-AKT/ERK pathway (Baba et al. 2017). The relationship of ANGPTL4 with the ERK pathway was also demonstrated in a non-small-cell lung cancer study; the study showed that ANGPTL4 contributes to epithelial mesenchymal transition (EMT) induction via the ERK pathway (Zhu et al. 2016). In triple-negative breast cancer, the upregulation of ANGPTL4 reduces the mRNA levels of extracellular matrix (ECM)-related genes, indicating that ANGPTL4 contributes to cancer progression by suppressing ECM-related proteins (Cai et al. 2020). Our study of CRC is consistent with these past reports; our study showed ANGPTL4 as a cancer progression factor and provides a novel finding that ANGPTL4 impacts glucose metabolism in cancers.
The PI3K-AKT signalling pathway plays an important role in diverse cellular functions, such as cell proliferation, differentiation, angiogenesis and autophagy (Noorolyai et al. 2019). Among the genes related to the pathway, RAS genes, including KRAS, were the first confirmed human protooncogenes that can transform into oncogenes (Luo et al. 2020). In CRC, KRAS mutation has been documented as a negative prognostic factor for both overall survival and recurrence-free survival and is now an important biomarker in clinical practice (Tsilimigars et al. 2018;Wan et al. 2019). According to our NGS data, high ANGPTL4 expression is significantly associated with KRAS mutations. Taken together, these findings indicate that a highly activated PI3K-AKT pathway induced by both ANGPTL4 overexpression and KRAS mutation results in tumour progression. This is compatible with the result of high GLUT expression in ANGPTL4-high tumours, since the PI3K-AKT pathway also activates GLUT expression (Papa et al. 2019).
We revealed that ANGPTL4 expression was strongly associated with GLUT1 and GLUT3 expression. Among the several subtypes of GLUTs, these two subtypes are known to be correlated with the progression of CRC, which is consistent with our results (Feng et al. 2017;Kim et al. 2019;Yang et al. 2017). According to TCGA database analysis in the current study, other GLUT coding genes, such as SLC2A5, SLC2A6, and SLC2A14, were also suggested to correlate with ANGPTL4 expression. As little is known about these GLUT proteins in cancers, further study is expected to clarify their roles.
To validate the relationship between ANGPTL4 expression and glucose metabolism from another perspective, we evaluated some measurement values from PET/CT. FDG uptake in tumours is based on the elevated metabolic activity of cancer cells and reflects increased glucose uptake (Watanabe et al. 2020). In fact, positive relationships between FDG uptake and GLUT1 expression in CRC and GLUT3 expression in malignant melanoma have been observed (Gu et al. 2006;Park et al. 2012). Thus, our results strongly suggest that glucose metabolism is highly activated in ANGPTL4-high tumours.
Since knockout of ANGPTL4 in cancer cells attenuated tumour progression in this study, ANGPTL4 is expected to be a novel target for cancer treatment. In fact, a neutralizing antibody against ANGPTL4 has been shown to be effective as a breast cancer treatment in a mouse xenograft model (Kolb et al. 2019). However, there are issues to be elucidated, such as adverse effects when physiological metabolism is suppressed by treatment. In recent years, genetic screening has become widely used as a practice of precision medicine, and the measurement of biomarkers has become easier. Therefore, by evaluating PI3K-AKT pathway activation status and KRAS mutation status as biomarkers, we may be able to select patients who are expected to be highly responsive to anti-ANGPTL4 treatment. Similarly, excluding patients who are expected to have adverse effects by genetic screening may be feasible in the near future.
In conclusion, ANGPTL4 expression correlates with glucose metabolism activity and is associated with CRC progression. In particular, GLUT1 and GLUT3 are considered to be regulated by ANGPTL4.

Declarations
Competing interests The authors have no relevant financial or nonfinancial interests to disclose.
Ethics approval This study was approved by the Ethics Committees of Keio University (ID 20150051), in accordance with the Helsinki Declaration.
Consent to participate Informed consent was obtained from all individual participants included in the study.
Consent for publication Written informed consent was obtained from the patients for publication.