Advances in Clinical and Experimental Medicine

Title abbreviation: Adv Clin Exp Med
JCR Impact Factor (IF) – 2.1 (5-Year IF – 2.0)
Journal Citation Indicator (JCI) (2023) – 0.4
Scopus CiteScore – 3.7 (CiteScore Tracker 3.3)
Index Copernicus  – 161.11; MNiSW – 70 pts

ISSN 1899–5276 (print)
ISSN 2451-2680 (online)
Periodicity – monthly

Download original text (EN)

Advances in Clinical and Experimental Medicine

2023, vol. 32, nr 7, July, p. 763–776

doi: 10.17219/acem/157515

Publication type: original article

Language: English

License: Creative Commons Attribution 3.0 Unported (CC BY 3.0)

Download citation:

  • BIBTEX (JabRef, Mendeley)
  • RIS (Papers, Reference Manager, RefWorks, Zotero)

Cite as:


Shang H, Xu A, Yan H, Xu D, Zhang J, Fang X. PIEZO2 promotes cell proliferation and metastasis in colon carcinoma through the SLIT2/ROBO1 /VEGFC pathway. Adv Clin Exp Med. 2023;32(7):763–776. doi:10.17219/acem/157515

PIEZO2 promotes cell proliferation and metastasis in colon carcinoma through the SLIT2/ROBO1/VEGFC pathway

Haotian Shang1,C, Aiguo Xu1,B,C, Haicui Yan1,B, Dandan Xu1,B,C, Jingyu Zhang1,A,C,D,F, Xinjian Fang1,E,F

1 People’s Hospital, Lianyungang Clinical College of Bengbu Medical College, China

Graphical abstract


Graphical abstracts

Abstract

Background. The PIEZO2 may be involved in the occurrence and development of tumors.

Objectives. To explore the potential mechanism and effect of PIEZO2 on colon cancer.

Materials and methods. We assessed the expression and prognostic role of PIEZO2 in patients with colon cancer. The role of PIEZO2 in SW480 cell proliferation, migration and invasion in vitro was investigated using cell counting kit-8 (CCK-8), wound healing, and transwell and cell invasion assays, respectively. The effect of PIEZO2 on SW480 cells in vivo was also explored. The potential mechanisms of PIEZO2 in SW480 cells were detected using quantitative reverse-transcription polymerase chain reaction (qRT-PCR) and western blot.

Results. The PIEZO2 was significantly increased in colon cancer tissues and the PIEZO2 high expression group was associated with a lower overall survival (OS) rate. Furthermore, PIEZO2 knockdown weakened the proliferation, migration and invasion of SW480 cells. The PIEZO2 knockdown was related to a lower expression of SLIT2, ROBO1, HIF-1α, and VEGFC. Finally, the tumors in control SW480 cells grew faster and larger than those in mice inoculated with si-PIEZO2 SW480 cells. Moreover, the si-PIEZO2 SW480 cell group showed a reduced expression of Ki67 and VEGFC and, at the same time, a significantly higher apoptosis index of tumor cells compared to the control group. The expression of PIEZO2 was higher in cancer-associated fibroblasts (CAFs) of colon cancer.

Conclusions. The PIEZO2 was increased in colon cancer tissues and was an unfavorable gene in patients with colon cancer, promoting colon cell proliferation, migration and invasion through the SLIT2/ROBO1/VEGFC pathway.

Key words: colon cancer, cancer-associated fibroblasts, PIEZO2, SLIT2/ROBO1/VEGFC pathway

 

Background

Because of its high mortality rate, colorectal cancer (CRC) has become a major health burden worldwide. Globally, there were more than 1.9 million new CRC cases and 935,000 deaths reported in 2020.1 If the cancer is local, 90% of patients with CRC survive 5 years after the diagnosis. However, once the disease has distant metastases, this rate reduces dramatically to 10%.2 The genetics of cancer cells is not a sole factor influencing tumor metastasis; additionally, the interaction between tumor cells and the surrounding microenvironment plays a pivotal role in tumor differentiation and progression.3 Mechanical stimulation including stiffness, shear stress, compression, and tensional forces from the surrounding microenvironment is frequent in tumor cells.4 Mechanotransduction, the conversion of physical stimuli into biochemical or electrical signals, plays a key role in biological processes in tumor cells, including development, proliferation, migration, and apoptosis.5

The PIEZO transmembrane proteins are conserved ion-channel proteins that sense external mechanical forces. The cell membrane pressure induced by external mechanical force can regulate changes in intracellular ions and exert physiological effects by changing the closure of the PIEZO transmembrane protein. The PIEZOs can regulate nociception and stem cell differentiation in Drosophila. The PIEZO transmembrane proteins mainly include 2 homologous proteins: PIEZO1 and PIEZO2. Although these proteins are homologous, their roles reported in the literature are not consistent. The PIEZO1 is mainly distributed in endothelial cells, red blood cells and hair cells in cochlea, as well as kidney, bladder and lung cancer cells. Under normal physiological conditions, PIEZO1 is primarily responsible for the regulation of intravascular balance, blood pressure and axon growth. The PIEZO2 is mainly distributed in the dorsal root nerve and trigeminal ganglion, which are responsible for regulating touch, proprioception of muscles and airway extension. It has been reported that PIEZOs are involved in the formation of human blood vessels.6 The PIEZO knockout can affect the growth and development of mice. Because of the important role of PIEZOs, the functions of the 2 homologous proteins have been increasingly studied.

In recent years, many studies have reported that PIEZOs may be involved in the occurrence and development of tumors, and may be related to angiogenesis. Gottlieb et al. reported that the reduction or deletion of PIEZO1 played an important role in lung cancer metastasis,7 while Yang et al. indicated that the decrease in PIEZO1 expression was related to the decrease in the gastric cancer metastasis.8 A study by Yang et al. showed that PIEZO2 knockout can reduce angiogenesis, vascular permeability, endothelial cell proliferation, metastasis, and tubule formation in gliomas.9 Another study showed that the expression of PIEZO1/2 was different in bladder cancer formation, suggesting that PIEZO1/2 plays a role in the proliferation and angiogenesis of bladder cancer cells.10 Chen et al. reported that the expression of PIEZO1 was inversely proportional to the stage and prognosis of glioma.11 The PIEZO1 can regulate changes in the extracellular matrix and tissue hardness as well as promote tumor progression by activating the integrin focal adhesion kinase (FAK) signaling pathway; however, PIEZO2 is not a prognostic factor. In a study by Pardo-Pastor et al., it was shown that PIEZO1 mainly forms a mechanoreceptor ion pathway in breast cancer cells, whereas PIEZO2 is involved in regulating RhoA, actin skeleton formation and cell movement.12 However, the roles of PIEZOs in colon cancer and their underlying mechanisms remain unclear and need to be further elucidated.

Previous studies showed that the SLIT/ROBO signaling pathway inhibited the migration of nerve cells in the central nervous system, prevented the nerve axon at the suture from passing through the middle line of the nerve tube, and controlled the accurate localization of neurite in nervous system. Studies on the expression and function of SLIT/ROBO signaling in various cancers have been conducted.13, 14, 15 Recently, the molecular mechanism of SLIT2/ROBO1 signaling in the regulation of colon cancer tumorigenesis has also been deeply studied. The SLIT2 was highly expressed in colon cancer and its expression increased in pathological stages.16 At the same time, SLIT2/ROBO1 signaling induced tumor metastasis in colon cancer partially through the activation of the transforming growth factor beta (TGF-β)/Smads pathway. Moreover, SLIT2/ROBO1 signaling recruited Src to E-cadherin for tyrosine phosphorylation, leading to E-cadherin degradation and epithelial–mesenchymal transition in colon cancer.17 Accordingly, SLIT2/ROBO1 signaling played an important role in the formation and metastasis of colon cancer.

Objectives

The above studies suggested PIEZO2 played an important role in the occurrence and progression of tumors. However, the detailed mechanism was not clear in colon cancer. Therefore, this study was designed to explore the prognostic role of PIEZO2 in colon cancer and whether the SLIT2/ROBO1 pathway is responsible for PIEZO2-induced progression of colon cancer.

Materials and methods

Patient information

A total of 74 colon cancer tissue samples were collected from January 2015 to December 2016 from patients with colon cancer who received surgery at the Department of Gastrointestinal Surgery, affiliated with Second People’s Hospital of Lianyungang (Bengbu Medical College, China). In addition, we collected 30 colon cancer tumor tissues and matched normal tissues. Clinical data were complete and the last follow-up time was December 2021. Patients included in the study did not receive anticancer therapy such as radiotherapy, chemotherapy or biological therapy before surgery. This study was approved by the Animal Care and Use Ethics Committee of the Affiliated Lianyungang Hospital (Bengbu Medical College, China; approval No. 2020-013-02).

Cell culture

The human colorectal cell line SW480 was purchased from the cell bank of the Chinese Academy of Sciences (Shanghai, China). The cell line grew in Roswell Park Memorial Institute (RPMI) 1640 medium (cat. No. 21870076; Thermo Fisher Scientific, Waltham, USA) supplemented with 10% fetal bovine serum (FBS) (item code: 13011-8611; SolelyBio, Shanghai, China) and cultured in a humidified 5% CO2 environment at 37°C.

Cell transfection

The SW480 cells were transfected with small interfering (si)RNA targeting PIEZO2 (si-PIEZO2, 5’-GGATAGTGAAGAGGAGGAAGA-3’) and the corresponding negative control (NC) using Lipofectamine 2000 Transfection Reagent (1 µL/50 µL, cat. No. 12566014; Invitrogen, Carlsbad, USA). The transfection effect was evaluated using quantitative reverse-transcription polymerase chain reaction (qRT-PCR) and the positive cell line was used for the following experiments.

RT-qPCR assay

Total RNA was extracted using TRIzol reagent (cat. No. 15596026; Invitrogen) and synthesized into cDNA using a reverse transcriptase Moloney Murine Leukemia Virus (M-MLV) Kit (cat. No. 18057018; Invitrogen). The gene expression was detected with a SYBR® Premix Ex Taq (cat. No. 11780200; Thermo Fisher Scientific). The primers used in this study are presented in Table 1. The expression of PIEZO2, HIF-1α, SLIT2, VEGFA, VEGFB, VEGFC, VEGFD, and ROBO1 was calculated using the 2−ΔΔCq method, and every value was measured thrice independently. The reaction conditions were as follows: 5 min at 95°C, followed by 40 cycles at 95°C for 30 s and 60°C for 45 s, and a final step at 72°C for 30 min.

Cell counting kit-8 assay

Cells (1×104 cells) were plated in four 96-well plates. At every time point, we added 10 μL of cell counting kit-8 (CCK-8) solution (cat. No. C0037; Beyotime Biotechnology, Beijing, China) to each well and after the solution was incubated with cells at 37°C for 30 min, the absorbance at 450 nm was measured with a microplate reader. Every experiment was repeated thrice independently.

Wound healing assay

Cells (1×105 cells) were seeded in 12-well plates. When cells reached 80% confluence, a 10-microliter sterile pipette tip was used to wound the cell monolayer. Then, a phase-contrast microscope (model IX71; Olympus Corp., Tokyo, Japan) was used to observe the cell migration at 0 h, 24 h and 48 h.

Transwell assay

Cells at a density of 6×104 were plated in the upper layer (8 μm; Corning, Lowell, USA). The bottom chamber was filled with medium supplemented with 10% FBS. After 48 h, the cells in the lower chamber were fixed. Then, a crystal violet solution was used to stain cells for further analysis using the microscope (model IX71; Olympus Corp.). Twelve randomly chosen areas per sample were used to assess the number of penetrating cells.

Cell invasion assay

The 2×105 cells were plated in the upper chamber of 24-well 8-micrometer pore well chambers (Corning). The cells were covered with diluted BD Matrigel (BD Biosciences, Franklin Lakes, USA) containing a complete medium. The lower chamber was filled with a medium containing 10% FBS. After 24 h, cells on the lower layer were fixed with 4% paraformaldehyde solution and stained with 0.1% crystal violet solution. The total number of penetrating cells was calculated and pictures were taken with a light microscope (model CKX53; Olympus Corp.; ×200 magnification).

Western blot

Total protein was extracted by means of a radioimmunoprecipitation assay (RIPA) buffer supplemented with a mixture of protease and phosphatase inhibitors. The protein concentration was determined with a bicinchoninic acid (BCA) protein assay. Protein samples were separated using 10% sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene fluoride (PVDF) membrane. After blocking with 10% fat-free milk, the membrane was incubated at 4°C overnight with the following primary antibodies: HIF1α (1:6000, cat. No. 20960-1-AP), ROBO1 (1:10000, cat. No. 20219-1-AP), VEGFC (1:1000, cat. No. 22601-1-AP), and SLIT2 (1:600, cat. No. 20217-1-AP; all from ProteinTech Group). Next, the membrane was incubated with horseradish peroxidase (HRP) secondary antibody (1:3000, cat. No. HRP-60004; ProteinTech Group) for 1 h. Chemiluminescent HRP substrate and a suitable X-ray were used to detect the protein bands. Proteins were identified and quantified using the AlphaEaseFC software (Alpha Innotech, San Leandro, USA).

Animal experiments

Male athymic BALB/c mice (5–6 weeks, 16–18 g) were purchased from Nanjing University Animal Laboratory (Nanjing, China). The animals were provided with food and water ad libitum. All experiment conditions conformed to the 3R principles of animal experimentation. To evaluate the role of PIEZO2 in vivo, the mice were injected with SW480 cell suspension (2×106) in 0.1 mL of culture medium subcutaneously into the right scapular region. The tumor size was measured every 7 days and calculated with the following formula: volume = tumor length × (tumor width)2 × 0.5236. All data were recorded in millimeters. When the bearing tumors reached ~1 cm3, mice were killed and tumor samples were weighed and fixed in formalin followed by paraffin embedding.

Immunohistochemistry

In animal experiment, immunohistochemical Ki-67 staining was used to evaluate the proliferative activity, and VEGFC expression was measured to evaluate the effect of PIEZO2 on angiogenesis in tumor tissues. We also explored the expression of PIEZO2 in colon cancer samples obtained from the patients of our hospital. The anti-Ki67 (cat. No. 27309-1-AP), anti-PIEZO2 (cat. No. 26205-1-AP) and anti-VEGFC (cat. No. 22601-1-AP; all from ProteinTech Group) antibodies were diluted and used at 1:16000, 1:200 and 1:200, respectively. Goat anti-rabbit secondary antibodies were also diluted and used at 1:200, 1:500 and 1:500. The labelling indices were introduced to quantify the expression of Ki-67 and VEGFC, calculated by the ratio of Ki-67- and VEGFC-positive cells/total cells and quantified using ×200 magnification in 3 randomly selected areas in each tissue sample. We evaluated the PIEZO2 expression in colon cancer tissues using the staining intensity and staining area of positive tumor cells in each section.18 The immunohistochemical score was used to assess the expression of PIEZO2 referred to the protocol described by Tian et al.19 The median score was used as the split point. The group with values higher than or equal to the median was considered the high-expression group, and the the group with values lower than the median was considered the low-expression group.

TUNEL assay

Tumor tissues were cut into 5-μm-thick slides and the slides were covered with the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) reaction mixture (cat. No. PF00006; ProteinTech Group). After washing off the unbound enzyme conjugate, a substrate reaction was used to visualize the peroxidase (POD)-retained cells in the immune complex. Using a light microscope (model CKX53; Olympus Corp.) at ×200 magnification, 3 different visual areas of each section were randomly selected and analyzed. The positive rate of apoptotic cells was defined as the density of apoptotic cells (the apoptosis index) using ImageJ software (National Institutes of Health, Bethesda, USA).

Statistical analyses

The relationship between PIEZO2 expression and the clinical characteristics of gender or site of primary tumor in colon cancer patients was analyzed using Spearman’s correlation analysis, and between PIEZO2 protein expression and the remaining 6 variables using Kendall’s tau-b correlation analysis. Comparison between the PIEZO2 expression in tumor and paired normal tissue samples was performed using χ2 test. The time from tumor resection to the patient’s death or last follow-up was defined as the overall survival (OS) time. The impact of the prognostic role of clinical/pathological characteristics on OS was analyzed using the log-rank test and the Kaplan–Meier analysis. Subsequently, proportional hazard assumption was assessed using Schoenfeld residuals. The variables of N stage and PIEZO2 did not meet the assumption. The univariate analysis was performed using Cox proportional hazards regression model for variables that met the assumption. The time-dependent Cox regression model was used for variables that did not meet the assumption. The multivariate analysis was conducted using the time-dependent Cox regression model to exclude the influence of confounding factors. Since the data from the 2 groups met the assumption of normality, the differences between the 2 groups were evaluated using the Student’s t-test and expressed with mean ± standard deviation (M ±SD). Because the repeated measurement data from the 2 groups did not show normal distribution, the data of this part were analyzed using the Scheirer–Ray–Hare test and expressed as median ±range. The IBM SPSS v. 20.0 (IBM Corp., Armonk, USA), R v. 4.2.2 software (survminer package; R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism v. 9 (GraphPad Software, San Diego, USA) were used in this study. The value of p < 0.05 was considered statistically significant.

Results

PIEZO2 expression was increased in colon cancer and associated with poor prognosis

A total of 30 colon cancer specimens with adjacent tissues were randomly selected from patients with colon cancer operated in Second People’s Hospital of Lianyungang. The expression of PIEZO2 was analyzed using immunohistochemistry. The results indicated that PIEZO2 was mainly located on the cell membrane and cytoplasm (Figure 1A). The PIEZO2 was highly expressed in colon cancer tissues compared with normal samples (18/30 compared to 9/30, χ2 = 5.455, p = 0.020; Figure 1B). As shown in Table 2, high expression of PIEZO2 was associated with tumor differentiation (Kendall’s tau-b = 0.360, p = 0.002), T stage (Kendall’s tau-b = 0.433, p < 0.001), N stage (Kendall’s tau-b = 0.453, p < 0.001), and clinical stage (Kendall’s tau-b = 0.322, p = 0.006). In addition, the prognostic data showed that patients with high expression of PIEZO2 had shorter OS (log-rank (Mantel–Cox) = 15.701, p < 0.001; Figure 1C). Finally, we found that high expression of PIEZO2 played an independent prognostic role for worsening OS in colon cancer, as shown with univariate and multivariate analyses (p = 0.007 and p = 0.034, respectively; Table 3).

Suppression of PIEZO2 decreased the migration and invasion of colon cancer

To explore the potential role of PIEZO2 in colon cancer, we identified 56 co-expressed genes (Figure 2A) using the Coexpedia website (http://www.coexpedia.org/index.php). Gene Ontology (GO) function and pathway enrichment analyses were performed using the DAVID database (https://david.ncifcrf.gov/). As expected, the biological processes were primarily associated with extracellular matrix organization, cellular response to reactive oxygen species (ROS), positive regulation of transcription, templated DNA, and response to hypoxia (Figure 2B). In terms of cellular components, these genes were also involved in the extracellular matrix (ECM), extracellular space, extracellular region, and receptor complexes (Figure 2B). The following biological processes were included: ECM structural constituents, integrin binding, ECM binding, collagen binding, and proteoglycan binding (Figure 2B). In addition, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis indicated that the following pathways were enriched in cancer: Ras signaling, PI3K-Akt signaling, MAPK signaling, and calcium signaling pathways (Figure 2C). As shown in Figure 2D, a stable transfection with si-PIEZO2 reduced the mRNA expression of PIEZO2 in SW480 cells. Moreover, SW480 cells with si-PIEZO2 showed weaker cell proliferation, migration and invasion functions (Figure 2E–H).

PIEZO2 is linked
to the SLIT2/ROBO1/VEGFC pathway

A previous study suggested that PIEZO2 plays a pivotal role in tumor angiogenesis.9 We explored the mRNA expression of VEGFA, VEGFB, VEGFC, and VEGFD in control and transfected SW40 cells, and the results indicated that only the expression of VEGFC was significantly decreased in si-PIEZO2 SW480 cells (Figure 3A). A recent study indicated that SLIT2/ROBO1 signaling participates in angiogenesis in a pathological context.20 We analyzed the relationship between SLIT2, ROBO1, VEGFC, HIF-1α, and PIEZO2 using the online website CBioPortal for Cancer Genomics (http://www.cbioportal.org/). The obtained results showed that the mRNA expression of PIEZO2 is positively correlated with the expression of VEGFC, SLIT2, ROBO1, and HIF-1α (Figure 3B–E). Furthermore, we assessed the protein and mRNA expression of VEGFC, SLIT2, ROBO1, and HIF-1α in control and transfected SW40 cells, and found that VEGFC, SLIT2, ROBO1, and HIF-1α were expressed at lower levels in transfected SW40 cells compared to the control SW40 cells (Figure 3F,G).

Suppression of PIEZO2 inhibits the growth of colon cancer in vivo

To determine the function of PIEZO2 in tumor growth, we injected control and transfected SW480 cells subcutaneously into nude mice. As shown in Figure 4A, tumors in control SW480 cell group grew faster and larger than those in transfected SW480 cell group. Consistent with the tumor volumes, the average tumor weight was clearly reduced (Figure 4B). Moreover, the transfected SW480 cells showed a reduced expression of Ki67 compared to the control group (Figure 4C). Additionally, the group of transfected SW480 cells had a higher apoptosis index of tumor cells than the control group (Figure 4D). Finally, we detected the expression of VEGFC in both groups. The results indicated that VEGFC expression in transfected SW480 cells was lower than in the control group (Figure 4E).

Positive correlation exists between PIEZO2 expression and cancer-associated fibroblasts

Recent studies have emphasized the role of cancer-associated fibroblasts (CAFs) in cancer progression and immune response. To explore the role of PIEZO2 in CAFs, we analyzed the correlation between PIEZO2 expression and CAFs using TIMER 2.0 (http://timer.cistrome.org/), and found that the expression of PIEZO2 was positively correlated with CAFs in colon cancer (Figure 5A). Then, we analyzed the expression markers of CAFs in colon cancer and discovered that PIEZO2 expression was statistically associated with immune subtypes, including wound healing, interferon gamma (IFN-γ) dominant, inflammatory, lymphocyte depleted, and TGF-β dominant (Figure 5B). Finally, we evaluated the role of PIEZO2 in CAFs in the tumor immune single-cell hub (TISCH). The results indicated that the PIEZO2 expression was positively correlated with CAFs in colon cancer (Figure 5C).

Discussion

Our data provide evidence that PIEZO2 expression increased in colon cancer and played a negative role in tumor survival. The results demonstrated that PIEZO2 promoted the proliferation, migration and invasion of colon cancer cells, and the detailed mechanism may be related to the SLIT2/ROBO1/VEGFC pathway.

The PIEZO channels are a new class of mechanosensitive channels, mainly responsible for the induction to mechanical forces and the entry or exit of Ca2+ in cells.21 The PIEZO channels are large transmembrane proteins composed of more than 2500 amino acids and 24–36 putative transmembrane fragments, including PIEZO1 and PIEZO2 members. The PIEZO channels play an important physiological role in various mechanotransduction processes including touch, hearing and blood pressure perception in mammals. Recent evidence also suggested that PIEZO1 was necessary for vascular development and function, and lineage choice of neural stem cells.22 The PIEZO2 plays an essential role in sensory transmission such as touch, mechanical proprioception, and gastrointestinal23, 24 and respiratory physiology. Moreover, some studies reported that PIEZO family made paramount contributions to the skeletal system, mainly in bone formation and mechanically stimulated bone homeostasis.25, 26

Mechanosensitive channels are necessary for mechanotransduction and are frequently lodged on the cell surface. Our results also confirmed that PIEZO2 is mainly expressed on the cell membranes of colon cells. The PIEZO channels play an important role in the pathogenesis of various diseases. Mechanical stress promoted stem cell differentiation by PIEZO channels in the adult Drosophila midgut, and this process was related to increased cytosolic Ca2+ levels caused by a direct mechanical stimulus.27 In fibrotic tissues, mechanical signaling regulated the interactions between cells and the ECM through discoidin domain receptor 1.28 However, several different cells sensed the stiffness of the surrounding substrate and responded to light by PIEZOs.29 Stiffness may regulate the expression of matricellular protein CCN1/CYR61 in endothelial cells after the tumor invasion, indicating that stiffness-induced changes are a potential target for impairing tumor metastasis.30

Some studies have suggested that PIEZO2 participates in the occurrence and development of malignant tumors. One study identified high expression of PIEZO2 in breast cancer and revealed that patients with elevated expression of PIEZO2 had a favorable prognosis in breast cancer.31 However, another study reported that the expression of PIEZO2 was clearly reduced in non-small-cell lung cancer (NSCLC) tissue and high expression of PIEZO2 correlated with better OS in NSCLC, especially in the adenocarcinoma subgroup.32 According to a recent study, PIEZO2 significantly increased in bladder cancer tissue, indicating that PIEZO2 dysfunction may contribute to the carcinogenesis of bladder cancer by causing proliferative changes and angiogenesis.10 Our results indicated that PIEZO2 was highly expressed in colon cancer and correlated with poor prognosis, which is consistent with previous results.

As far as we know, when solid tumors grow beyond a certain size, they require angiogenesis to supply nutrients. In this process, some inflammatory and cell factors are involved.33 Yang et al. reported that PIEZO2 increased angiogenesis and vascular hyperpermeability in endothelial cells, and Ca2+/Wnt11/β-catenin signaling played an important role in this process.9 Another study suggested that intracellular Ca2+ signaling promoted angiogenesis and vasculogenesis by enhancing the functions of vascular endothelial cells (VECs) and endothelial colony-forming cells, which is linked to an increase in the expression of PIEZO2.34 Therefore, these studies indicate that PIEZO2 is involved in angiogenesis. In this study, we found that the expression of VEGFC was changing along with PIEZO2 expression, which was verified by the level of mRNA, protein and in vivo studies. The VEGFC is a member of VEGF family and plays an important role in lymphangiogenesis. The VEGFC stimulates the formation of new lymph vessels and provides a chance for detached cancer cells to metastasize to a distant site.35 The VEGFC can promote the immune escape in colon cancer by activating VEGFR3, which results in tumor growth.36 Moreover, our results suggest that PIEZO2 is linked to the SLIT2/ROBO1 pathway. Additionally, some studies reported that the SLIT2/ROBO1 signaling participates in the process of angiogenesis.20, 37 Overall, our results suggest that PIEZO2 plays an antiangiogenic role through the SLIT2/ROBO1/VEGFC pathway. However, the detailed mechanism is needed to be explored further.

Currently, the tumor microenvironment (TME) is a trending topic in tumor research. Among numerous components of TME, CAFs play an important role in regulating tumorigenesis. Affo et al. found that a pan-CAF signature was significantly associated with poor survival in cholangiocarcinoma and tumor recurrence.38 Another study found that when pancreatic cancer cells were co-cultured with CAFs, cancer stemness increased.39 A recent study suggested that CAFs are the primary source of TIMP-1, and TIMP-1 production is enhanced through cancer–stromal interaction, which promotes cancer cell migration.40 Regarding colon cancer, one study suggested that increased RAB31 expression in CAFs may contribute to tumor progression by regulating hepatocyte growth factor (HGF) secretion in the tumor stoma.41 Our study showed that the expression of PIEZO2 positively correlates with CAFs in colon cancer, indicating that PIEZO2 is a prognostic factor of the disease. Recently, 6 immune subtypes were identified by immunogenomic data from 33 different types of cancer.42 This classification method can distinguish the beneficiaries of immune targeted therapy from multiple heterogeneous tumors. In this study, PIEZO2 was significantly and aberrantly expressed in C1, C2, C3, C4, and C6 of colon cancer, suggesting that PIEZO2 might have potential in cancer immune therapy. Moreover, the positive correlation between PIEZO2 expression and CAFs was further validated using a single-cell sequencing dataset. However, the underlying mechanism of PIEZO2 regulating CAFs requires further study.

Limitations

There are some shortcomings of our study that are worth considering. First, the sample size of clinical patients was relatively small. A large sample study would be needed to validate the results of our study. Second, the detailed mechanism of PIEZO2 regulating SLIT2/ROBO1/VEGFC pathway had not been explored. This may be the direction of our further research.

Conclusions

The PIEZO2 is highly expressed and is a poor prognostic factor in colon cancer. Moreover, PIEZO2 can promote cell proliferation and mobility in colon carcinoma through the SLIT2/ROBO1/VEGFC pathway. At the same time, PIEZO2 is positively correlated with CAFs, indicating that it plays an important role in the CAF function. Therefore, PIEZO2 may be a potential target for colon cancer treatment.

Supplementary data

The supplementary files are available at https://doi.org/10.5281/zenodo.7428254. The package contains the following files:

Supplementary Table 1. Results of χ2 phi test for Figure 1B.

Supplementary Table 2. Results of the log-rank test for Figure 1C.

Supplementary Table 3. Results of the normality test for Figure 2D.

Supplementary Table 4. Results of the Student’s t-test for Figure 2D.

Supplementary Table 5. Results of the normality test for Figure 2E.

Supplementary Table 6. Results of the Scheirer–Ray–Hare test for Figure 2E.

Supplementary Table 7. Results of the normality test for Figure 2F.

Supplementary Table 8. Results of the Student’s t-test for Figure 2F.

Supplementary Table 9. Results of the normality test for Figure 2G.

Supplementary Table 10. Results of the Student’s t-test for Figure 2G.

Supplementary Table 11. Results of the normality test for Figure 2H.

Supplementary Table 12. Results of the Student’s t-test for Figure 2H.

Supplementary Table 13. Results of the normality test for Figure 3A.

Supplementary Table 14. Results of the Student’s t-test for Figure 3A.

Supplementary Table 15. Results of the normality test for Figure 3F.

Supplementary Table 16. Results of the Student’s t-test for Figure 3F.

Supplementary Table 17. Results of the normality test for Figure 3G.

Supplementary Table 18. Results of the Student’s t-test for Figure 3G.

Supplementary Table 19. Results of the normality test for Figure 4A.

Supplementary Table 20. Results of the Scheirer–Ray–Hare test Figure 4A.

Supplementary Table 21. Results of the normality test for Figure 4B.

Supplementary Table 22. Results of the Student’s t-test for Figure 4B.

Supplementary Table 23. Results of the normality test for Figure 4C.

Supplementary Table 24. Results of the Student’s t-test for Figure 4C.

Supplementary Table 25. Results of the normality test for Figure 4D.

Supplementary Table 26. Results of the Student’s t-test for Figure 4D.

Supplementary Table 27. Results of the normality test for Figure 4E.

Supplementary Table 28. Results of the Student’s t-test for Figure 4E.

Supplementary Table 29. Results of Spearman’s correlation analysis for Table 2.

Supplementary Table 30. Results of Kendall’s tau-b correlation analysis for Table 2.

Supplementary Table 31. Results of Schoenfeld residuals for Table 3.

Supplementary Table 32. Results of the univariate analysis for Table 3.

Supplementary Table 33. Results of the multivariate analysis for Table 3.

Tables


Table 1. Primer sequences used in the study

Gene symbol

Primer sequence (5’3’)

PIEZO2

forward: TTCCGATACAATGGGCTCTC

reverse: GAGCTTCAAGGCTCACCAAC

HIF-1α

forward: AGCACAGTTACAGTATTCCAGCAG

reverse: AGTGGTGGCAGTGGTAGTGG

SLIT2

forward: TCCCCACAAATCTTCCAGAG

reverse: AGCGTAGTCCTTGGAAAGCA

VEGFA

forward: GAGCGAGAAAGCATTTGTTTG

reverse: GCAACGCGAGTCTGTGTGTTTT

VEGFB

forward: CAAGGCTGCCATCCAACAA

reverse: AGGTGGTTAAAATCTGAGTCGGG

VEGFC

forward: ACCAAACAAGGAGCTGGATG

reverse: TCCCCACAAATCTTCCAGAG

VEGFD

forward: TGTACAGACAGTGGGCAGTGGT

reverse: AACCTGAAGCTGCCCTGATCT

ROBO1

forward: ACTCCTTACGCCACCACTCAG

reverse: CCTGTGTTCTGGTCGTATGATTGG

h-actin

forward: TGGACTTCGAGCAAGAGATG

reverse: GAAGGAAGGCTGGAAGAGTG

Table 2. The relationship between PIEZO2 protein expression and clinical pathological features in colon cancer patients

Characteristic

Total, n

Expression level of PIEZO2

p-value

high

low

Age [years]

<65

≥65

30

44

18

27

12

17

0.907

Gender

male

female

44

30

25

20

19

10

0.401

Site of primary tumor

left

right

31

43

17

28

14

15

0.378

Differentiation

moderate and well

poor

37

37

16

29

21

8

0.002

T stage

T1+T2

T3+T4

29

45

10

35

19

10

<0.001

N stage

N0

N1+N2

26

48

8

37

18

11

<0.001

Clinical stage

I

II+III

20

54

12

33

8

21

0.006

CEA [ng/mL]

<6

≥6

19

55

9

36

10

19

0.931

CEA – carcinoembryonic antigen. Spearman’s correlation analysis was used to analyze the relationship between the protein expression of PIEZO2 and the clinical characteristics of gender or site of primary tumor. Kendall’s tau-b correlation analysis was applied to explore the relationship between the protein expression of PIEZO2 and the remaining 6 variables. The value of p < 0.05 was considered statistically significant.
Table 3. Univariate analysis and multivariate analysis of the overall survival (OS) in colon cancer patients

Characteristic

Univariate analysis

Multivariate analysis

HR

95% CI

p-value

HR

95% CI

p-value

Age

0.849

0.517–1.393

0.516

0.768

0.416–1.420

0.400

Gender

1.006

0.612–1.653

0.982

0.607

0.318–1.160

0.131

Tumor location

0.806

0.488–1.330

0.398

0.587

0.301–1.144

0.118

Differentiation

2.130

1.299–3.494

0.003

2.652

1.191–5.905

0.017

T stage

1.658

1.004–2.740

0.048

0.367

0.138–0.977

0.045

N stage (time-dependent variable)

3612.754

5.558–2348302.323

0.013

51.028

0.049–52836.422

0.267

Clinical stage

2.350

1.319–4.187

0.004

2.086

1.036–4.202

0.040

PIEZO2 (time-dependent variable)

4261.402

9.758–1861054.007

0.007

1884.388

1.797–1976563.451

0.034

CEA

1.179

0.669–2.077

0.570

1.443

0.729–2.857

0.293

HR – hazard ratio; 95% CI – 95% confidence interval; CEA – carcinoembryonic antigen. Proportional hazard assumption was examined using Schoenfeld residuals. The variables of N stage and PIEZO2 did not meet the assumption. Subsequently, the univariate analysis was performed using the Cox proportional hazards regression model for the variables that met the assumption. The time-dependent Cox regression model was used for variables that did not meet the assumption. At last, the multivariate analysis was conducted using the time-dependent Cox regression model to exclude the influence of confounding factors. The value of p < 0.05 was considered statistically significant.

Figures


Fig. 1. A. Immunohistochemical staining images of PIEZO2 in colon cancer and adjacent colon tissues; B. Difference in expression levels of PIEZO2 in colon cancer and normal colon samples analyzed using 30 colon cancer patient dataset and χ2 test; C. The Kaplan–Meier analysis and the log-rank test were used to determine the relationship between the expression of PIEZO2 and the clinical outcome in 74 colon cancer patients
* p < 0.05.
Fig. 2. Inhibition of PIEZO2 suppresses the proliferation, migration and invasion in colon cancer cells. A. Fifty-six co-expressed genes of PIEZO2 identified using the Coexpedia website; B. Enrichment analysis of co-expressed genes performed using the DAVID database; C. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of co-expressed genes conducted using the DAVID database; D. After the SW480 cells were transfected with small interfering (si)RNA targeting PIEZO2, the expression of PIEZO2 in negative control (NC) and si-PIEZO2 SW480 cells was tested using quantitative reverse-transcription polymerase chain reaction (qRT-PCR). The results were analyzed using the Student’s t-test and expressed with mean ± standard deviation (M ±SD); E. Cell counting kit-8 (CCK-8) assay was performed to evaluate the cell viability of NC and si-PIEZO2 SW480 cells. The results were analyzed using the Scheirer–Ray–Hare test and expressed as median (range); F,G. Wound healing and cell invasion assays were performed to evaluate the migration ability of NC and si-PIEZO2 SW480 cells. The results were analyzed using the Student’s t-test and expressed as M ±SD; H. Matrigel™ invasion assay was performed to evaluate the invasion ability of NC and si-PIEZO2 SW480 cells. The results were analyzed using the Student’s t-test and expressed as M ±SD. All experiment results were representative of 3 independent experiments
** p < 0.01; *** p < 0.001; OD – optical density.
Fig. 3. PIEZO2 is correlated with the SLIT2/ROBO1/VEGFC pathway. A. The expression of VEGFA, VEGFB, VEGFC, and VEGFD in negative control (NC) and si-PIEZO2 SW480 cells was tested using quantitative reverse-transcription polymerase chain reaction (qRT-PCR). The differences were analyzed using the Student’s t-test and expressed with mean ± standard deviation (M ±SD); B–E. Correlation between PIEZO2 and VEGFC, SLIT2, ROBO1, and HIF1A was explored using the website CBioPortal for Cancer Genomics; F,G. The expression of VEGFC, SLIT2, ROBO1, and HIF1A in NC and si-PIEZO2 SW480 cells were tested using western blot and qRT-PCR
** p < 0.01; *** p < 0.001.
Fig. 4. Suppression of PIEZO2 significantly inhibits the growth of colon cancer in vivo. The negative control (NC) or si-PIEZO2 SW480 cells (2×106) in 0.1 mL of culture medium were injected subcutaneously into the right scapular region of nude mice to evaluate the antitumor activity of PIEZO2 in vivo (3 mice per group). The tumor size was measured every 7 days using the formula A×B2×0.5236, where A is the length and B is the width of the tumor; all measurements were recorded in millimeters. When the bearing tumors were approx. 1 cm3 in size, all mice were sacrificed, and tissue specimens were weighed and fixed in formalin for paraffin embedding. A. Tumor volumes were measured every 7 days using caliper measurements. The differences between the 2 groups were compared using the Scheirer–Ray–Hare test and expressed as median (range); B. The difference in tumor weight between the 2 groups was analyzed using the Student’s t-test and expressed as mean ± standard deviation (M ±SD); C. Proliferation analysis of tumor tissues was performed using Ki67 immunohistochemical staining, and the difference between the 2 groups was evaluated using the Student’s t-test and expressed as M ±SD; D. Apoptosis analysis of tumor tissues was performed using the terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay. The differences between the 2 groups were compared using the Student’s t-test and expressed as M ±SD; E. VEGFC expression was tested with immunohistochemical staining. The differences between the 2 groups were compared using the Student’s t-test and expressed as M ±SD
* p < 0.05; *** p < 0.001.
Fig. 5. Positive correlation between PIEZO2 expression and cancer-associated fibroblasts (CAFs). A. PIEZO2 expression level in immune subtypes in colon adenocarcinoma (COAD) based on tumor-immune system interactions database (TISIDB) (http://cis.hku.hk/TISIDB/); C. Correlation between PIEZO2 expression in single-cell sequencing and CAFs using the tumor immune single-cell hub (TISCH)

References (42)

  1. Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209–249. doi:10.3322/caac.21660
  2. Lee RM, Cardona K, Russell MC. Historical perspective: Two decades of progress in treating metastatic colorectal cancer. J Surg Oncol. 2019;119(5):549–563. doi:10.1002/jso.25431
  3. Willyard C. Cancer therapy: An evolved approach. Nature. 2016;532(7598):166–168. doi:10.1038/532166a
  4. De Felice D, Alaimo A. Mechanosensitive piezo channels in cancer: Focus on altered calcium signaling in cancer cells and in tumor progression. Cancers (Basel). 2020;12(7):1780. doi:10.3390/cancers12071780
  5. Martino F, Perestrelo AR, Vinarský V, Pagliari S, Forte G. Cellular mechanotransduction: From tension to function. Front Physiol. 2018;9:824. doi:10.3389/fphys.2018.00824
  6. Coste B, Xiao B, Santos JS, et al. Piezo proteins are pore-forming subunits of mechanically activated channels. Nature. 2012;483(7388):176–181. doi:10.1038/nature10812
  7. Gottlieb PA, Bae C, Sachs F. Gating the mechanical channel Piezo1: A comparison between whole-cell and patch recording. Channels. 2012;6(4):282–289. doi:10.4161/chan.21064
  8. Yang XN, Lu YP, Liu JJ, et al. Piezo1 is as a novel trefoil factor family 1 binding protein that promotes gastric cancer cell mobility in vitro. Dig Dis Sci. 2014;59(7):1428–1435. doi:10.1007/s10620-014-3044-3
  9. Yang H, Liu C, Zhou RM, et al. Piezo2 protein: A novel regulator of tumor angiogenesis and hyperpermeability. Oncotarget. 2016;7(28):44630–44643. doi:10.18632/oncotarget.10134
  10. Etem E, Ceylan G, Özaydın S, Ceylan C, Özercan I, Kuloğlu T. The increased expression of Piezo1 and Piezo2 ion channels in human and mouse bladder carcinoma. Adv Clin Exp Med. 2018;27(8):1025–1031. doi:10.17219/acem/71080
  11. Chen X, Wanggou S, Bodalia A, et al. A feedforward mechanism mediated by mechanosensitive ion channel PIEZO1 and tissue mechanics promotes glioma aggression. Neuron. 2018;100(4):799.e7–815.e7. doi:10.1016/j.neuron.2018.09.046
  12. Pardo-Pastor C, Rubio-Moscardo F, Vogel-González M, et al. Piezo2 channel regulates RhoA and actin cytoskeleton to promote cell mechanobiological responses. Proc Natl Acad Sci U S A. 2018;115(8):1925–1930. doi:10.1073/pnas.1718177115
  13. Mertsch S, Schmitz N, Jeibmann A, Geng JG, Paulus W, Senner V. Slit2 involvement in glioma cell migration is mediated by Robo1 receptor. J Neurooncol. 2008;87(1):1–7. doi:10.1007/s11060-007-9484-2
  14. Xia Y, Wang L, Xu Z, et al. Reduced USP33 expression in gastric cancer decreases inhibitory effects of Slit2‐Robo1 signaling on cell migration and EMT. Cell Prolif. 2019;52(3):e12606. doi:10.1111/cpr.12606
  15. Ahirwar DK, Peng B, Charan M, et al. Slit2/Robo1 signaling inhibits small cell lung cancer by targeting β‐catenin signaling in tumor cells and macrophages [published online as ahead of print on July 15, 2022]. Mol Oncol. 2022. doi:10.1002/1878-0261.13289
  16. Yao Y, Zhou Z, Li L, et al. Activation of Slit2/Robo1 signaling promotes tumor metastasis in colorectal carcinoma through activation of the TGF-β/Smads pathway. Cells. 2019;8(6):635. doi:10.3390/cells8060635
  17. Tan Q, Liang XJ, Lin SM, et al. Engagement of Robo1 by Slit2 induces formation of a trimeric complex consisting of Src-Robo1-E-cadherin for E-cadherin phosphorylation and epithelial-mesenchymal transition. Biochem Biophys Res Commun. 2020;522(3):757–762. doi:10.1016/j.bbrc.2019.11.150
  18. Gu Y, Li X, Bi Y, et al. CCL14 is a prognostic biomarker and correlates with immune infiltrates in hepatocellular carcinoma. Aging. 2020;12(1):784–807. doi:10.18632/aging.102656
  19. Tian Y, Zhou Y, Liu J, et al. Correlation of SIDT1 with poor prognosis and immune infiltration in patients with non-small cell lung cancer. Int J Gen Med. 2022;15:803–816. doi:10.2147/IJGM.S347171
  20. Liu J, Hou W, Guan T, et al. Slit2/Robo1 signaling is involved in angiogenesis of glomerular endothelial cells exposed to a diabetic-like environment. Angiogenesis. 2018;21(2):237–249. doi:10.1007/s10456-017-9592-3
  21. Xu XZS. Demystifying mechanosensitive piezo ion channels. Neurosci Bull. 2016;32(3):307–309. doi:10.1007/s12264-016-0033-x
  22. Qin L, He T, Chen S, et al. Roles of mechanosensitive channel Piezo1/2 proteins in skeleton and other tissues. Bone Res. 2021;9(1):44. doi:10.1038/s41413-021-00168-8
  23. Bai T, Li Y, Xia J, et al. Piezo2: A candidate biomarker for visceral hypersensitivity in irritable bowel syndrome? J Neurogastroenterol Motil. 2017;23(3):453–463. doi:10.5056/jnm16114
  24. Wang F, Knutson K, Alcaino C, et al. Mechanosensitive ion channel Piezo2 is important for enterochromaffin cell response to mechanical forces: Piezo2 in EC cells. J Physiol. 2017;595(1):79–91. doi:10.1113/JP272718
  25. Zhou T, Gao B, Fan Y, et al. Piezo1/2 mediate mechanotransduction essential for bone formation through concerted activation of NFAT-YAP1-ß-catenin. eLife. 2020;9:e52779. doi:10.7554/eLife.52779
  26. Wang L, You X, Lotinun S, Zhang L, Wu N, Zou W. Mechanical sensing protein PIEZO1 regulates bone homeostasis via osteoblast-osteoclast crosstalk. Nat Commun. 2020;11(1):282. doi:10.1038/s41467-019-14146-6
  27. He L, Si G, Huang J, Samuel ADT, Perrimon N. Mechanical regulation of stem-cell differentiation by the stretch-activated Piezo channel. Nature. 2018;555(7694):103–106. doi:10.1038/nature25744
  28. Coelho NM, McCulloch CA. Mechanical signaling through the discoidin domain receptor 1 plays a central role in tissue fibrosis. Cell Adh Migr. 2018;12(4):348–362. doi:10.1080/19336918.2018.1448353
  29. Gottlieb PA. A tour de force: The discovery, properties and function of piezo channels. Curr Top Membr. 2017;79:1–36. doi:10.1016/bs.ctm.2016.11.007
  30. Reid SE, Kay EJ, Neilson LJ, et al. Tumor matrix stiffness promotes metastatic cancer cell interaction with the endothelium. EMBO J. 2017;36(16):2373–2389. doi:10.15252/embj.201694912
  31. Lou W, Liu J, Ding B, et al. Five miRNAs-mediated PIEZO2 downregulation, accompanied with activation of Hedgehog signaling pathway, predicts poor prognosis of breast cancer. Aging. 2019;11(9):2628–2652. doi:10.18632/aging.101934
  32. Huang Z, Sun Z, Zhang X, et al. Loss of stretch-activated channels, PIEZOs, accelerates non-small cell lung cancer progression and cell migration. Biosci Rep. 2019;39(3):BSR20181679. doi:10.1042/BSR20181679
  33. Carmeliet P, Jain RK. Molecular mechanisms and clinical applications of angiogenesis. Nature. 2011;473(7347):298–307. doi:10.1038/nature10144
  34. Moccia F. Endothelial Ca2+ signaling and the resistance to anticancer treatments: Partners in crime. Int J Mol Sci. 2018;19(1):217. doi:10.3390/ijms19010217
  35. Ellis LM, Hicklin DJ. VEGF-targeted therapy: Mechanisms of anti-tumour activity. Nat Rev Cancer. 2008;8(8):579–591. doi:10.1038/nrc2403
  36. Tacconi C, Ungaro F, Correale C, et al. Activation of the VEGFC/VEGFR3 pathway induces tumor immune escape in colorectal cancer. Cancer Res. 2019;79(16):4196–4210. doi:10.1158/0008-5472.CAN-18-3657
  37. Romano E, Manetti M, Rosa I, et al. Slit2/Robo4 axis may contribute to endothelial cell dysfunction and angiogenesis disturbance in systemic sclerosis. Ann Rheum Dis. 2018;77(11):1665–1674. doi:10.1136/annrheumdis-2018-213239
  38. Affo S, Nair A, Brundu F, et al. Promotion of cholangiocarcinoma growth by diverse cancer-associated fibroblast subpopulations. Cancer Cell. 2021;39(6):883. doi:10.1016/j.ccell.2021.03.012
  39. Nallasamy P, Nimmakayala RK, Karmakar S, et al. Pancreatic tumor microenvironment factor promotes cancer stemness via SPP1–CD44 axis. Gastroenterology. 2021;161(6):1998.e7–2013.e7. doi:10.1053/j.gastro.2021.08.023
  40. Nakai N, Hara M, Takahashi H, et al. Cancer cell induced tissue inhibitor of metalloproteinase 1 secretion by cancer associated fibroblasts promotes cancer cell migration. Oncol Rep. 2022;47(6):112. doi:10.3892/or.2022.8323
  41. Yang T, Zhiheng H, Zhanhuai W, et al. Increased RAB31 expression in cancer-associated fibroblasts promotes colon cancer progression through HGF-MET signaling. Front Oncol. 2020;10:1747. doi:10.3389/fonc.2020.01747
  42. Thorsson V, Gibbs DL, Brown SD, et al. The immune landscape of cancer. Immunity. 2018;48(4):812.e14–830.e14. doi:10.1016/j.immuni.2018.03.023