ILT4 reprograms glucose metabolism to promote tumor progression in triple-negative breast cancer

Triple-negative breast cancer (TNBC) is the most aggressive subtype among different breast cancer (BC) subtypes. It accounts for 15–20% of BC cases and is characterized by high rates of metastasis, recurrence and poor survival (Foulkes et al., 2010; Reis-Filho and Pusztai, 2011). Owing to the aggressive behaviors and deficiency of therapeutic targets, the treatment of TNBC remains an unmet medical challenge. Despite all of the therapeutic advances in immune checkpoint inhibitor (ICI) (Marra et al., 2019; Woodward, 2020), PARP inhibitors (Fasching et al., 2018; Robson et al., 2019) and antibody drug conjugates (ADC) (Bardia et al., 2017; Bardia et al., 2019), the 5-year overall survival (OS) for TNBC is much lower compared with other subtypes (Bianchini et al., 2016). Identification of druggable targets for precision therapy based on novel tumorigenic mechanisms are essential for improvement of TNBC treatment and survival.

Immunoglobulin-like transcriber 4 (ILT4; also known as LILRB2) is a newly identified immune checkpoint molecule (Gao et al., 2018; Ravetch and Lanier, 2000). It is physiologically expressed on innate immune cells. Upon ligation or activation, ILT4 negatively regulates antigen presentation of dendritic cells (DCs), phagocytosis of neutrophils, maturation and M1-like polarization of macrophages, and induces immunosuppression (Chen et al., 2018; Liang et al., 2008). The homolog of ILT4 in mouse is the paired Ig-like receptor-B (PIR-B) (Kubagawa et al., 1997; Yamashita et al., 1998). In the past decade, we and other groups have found that ILT4 and PIR-B are also enriched in multiple tumor cells including lung cancer, esophageal cancer and pancreatic cancer (Carbone et al., 2015; Sun et al., 2008; Warnecke-Eberz et al., 2016; Zhang et al., 2015a). Tumor-derived ILT4 mediates immunosuppressive tumor microenvironment (TME) by restricting T cell immunity or recruiting M2-like tumor-associated macrophages (TAMs) (Chen et al., 2021; Gao et al., 2021; Li et al., 2020; Yang et al., 2021). Furthermore, ILT4 directly promotes the proliferation and invasion of lung cancer and colorectal cancer cells (Cai et al., 2019; Zhang et al., 2015a,b). However, the expression and function of ILT4 in different BC subtypes, especially in TNBC, is still unknown.

Metabolic reprogramming has recently emerged as a crucial cancer hallmark. Owing to the rapid proliferation, TNBC cells display discrete metabolic phenotypes to fuel cell progression and metastasis, among which aerobic glycolysis, also known as the Warburg effect (Warburg et al., 1927), is the most fundamental energy phenotype (Hanahan and Weinberg, 2011; Pavlova and Thompson, 2016). It is reported that TNBC is more dependent on glycolysis to sustain rapid proliferation, microenvironmental remodeling and distant metastasis, compared to other BC subtypes (Arundhathi et al., 2021). Abnormal activation of oncogenes like EGFR and Myc in TNBC represents one of the major causes for rewiring the metabolism towards glycolytic phenotype (Wang et al., 2020). However, as a potential tumor driver, whether ILT4 orchestrates TNBC progression via reprogrammed aerobic glycolysis is undefined.

In this study, we found that TNBC displayed the most abundant ILT4 expression among all BC subtypes. Enriched ILT4 predicted advanced diseases and poorer patient survival. Functionally, ILT4 promoted TNBC cell proliferation, migration and invasion in vitro, as well as tumor growth and metastasis in vivo. Further mechanistic exploration revealed that ILT4 reprogrammed aerobic glycolysis of tumor cells via AKT-mTOR signaling-mediated glucose transporter 3 (GLUT3; also known as SLC2A3) and pyruvate kinase muscle 2 (PKM2; an isoform encoded by PKM) overexpression. ILT4 inhibition in TNBC reduced tumor progression and GLUT3 and PKM2 expression in vivo. Our study identified a novel driver for TNBC progression and proposes a promising strategy to combat TNBC by targeting ILT4.

Increased ILT4 expression has been identified in multiple cancers (Cai et al., 2019; Carbone et al., 2015; Chen et al., 2021; Gao et al., 2021; Li et al., 2020; Sun et al., 2008; Warnecke-Eberz et al., 2016; Yang et al., 2021; Zhang et al., 2015a,b). To determine whether ILT4 is a malignant biomarker in BC, first we investigate the expression of ILT4 in human breast cancer samples within the Ualcan database (http://ualcan.path.uab.edu/analysis.html). As shown in Fig. 1A, BC tissues displayed significantly higher ILT4 expression compared with normal breast tissues. Given the heterogeneous tumor biologies and clinical outcomes in different BC subtypes, we then analyzed ILT4 expression in all BC subtypes, including luminal, Her 2 positive and TNBC subtypes. We found that among all these types, TNBC, the most aggressive subtype, showed the highest ILT4 levels (Fig. 1B). We also determined patient survival based on ILT4 expression in different molecular types. As shown in Fig. 1C, in TNBC patients, high ILT4 expression predicted unfavorable OS compared with ILT4 low/medium group. We then confirmed the above conclusion in our patient cohort (see Materials and Methods for further details). We retrospectively analyzed the expression of ILT4 in primary BC tissues with luminal, Her2-positive and TNBC subtypes. We found that ILT4 expression is significantly much higher in all these subtypes compared with that in adjacent normal tissues, whereas TNBC showed the highest ILT4 levels (Fig. 1D,E). Positive ILT4 staining was mainly found in cell membrane and cytoplasm of tumor cells. These results prompted us to focus our work on TNBC. We then detected mRNA and protein expression of ILT4 in TNBC cell lines. As expected, different TNBC cell lines showed markedly higher ILT4 expression compared with that in the mammary epithelial cell line MCF10A at both mRNA (Fig. 1F) and protein (Fig. 1G) levels. To explore the role of ILT4 in the pathogenesis of TNBC development, we analyzed the correlation of ILT4 levels with patient clinicopathological parameters (Table S1). We observed that patients with high ILT4 expression showed larger tumor sizes (Fig. 1H), more lymph node involvement (Fig. 1I), frequent distant metastasis (Fig. 1J) and finally advanced TNM stages (Fig. 1K). We also found ILT4-high patients had higher Ki-67 (also known as MKI67) levels (Fig. 1L), indicating more active proliferation ability. Kaplan–Meier survival analysis showed that high ILT4 expression was associated with significantly shortened OS [hazard ratio (HR)=2.074, 95% c.i. 1.004–4.233] in TNBC patients (Fig. 1M). Altogether, TNBC displayed the highest ILT4 expression among all BC subtypes. Enriched ILT4 in tumor cells predicts progressive diseases and poor prognosis, indicating the important role of ILT4 in TNBC occurrence and development.

To investigate the functional effect of ILT4 on TNBC progression, we established ILT4 overexpression and knockdown TNBC cell lines by lentivirus infection. MDA-MB-231 and HCC1937 cells with intrinsically low ILT4 expression were selected for ILT4 overexpression, whereas BT549 and HS578 T cells with relatively higher ILT4 expression for knockdown. The overexpression (Fig. S1A,B) and knockdown (Fig. S1C,D) efficiency was first verified at both the mRNA and protein levels. EdU cell proliferation assays were used to detect the effect of ILT4 on tumor cell proliferation. We found ILT4 overexpression remarkably enhanced the proliferation ability of MDA-MB-231 and HCC1937 cells (Fig. 2A,B), whereas knockdown of ILT4 significantly decreased the proliferation rate of BT549 and HS578 T cells (Fig. 2C,D). These results were further verified by a CCK8 proliferation assay (Fig. 2E; Fig. S1E). We also detected the classical proliferation biomarker Ki-67 upon ILT4 regulation. Consistent with the proliferation ability, overexpression (Fig. 2F,G) or knockdown (Fig. 2H,I) of ILT4 markedly up- or down-regulated Ki-67 expression at both mRNA and protein levels respectively. These results suggest that ILT4 regulates the proliferation and growth of TNBC cells, and explain the correlation of ILT4 with larger tumor sizes in TNBC patients. Given that high ILT4 expression is also correlated with tumor metastasis in the patient cohort, we then want to know whether ILT4 controls the migration and invasion, two fundamental hallmarks for tumor metastasis, of TNBC cells. Using Transwell assays, we found that ILT4 overexpression increased the migration and invasion ability of MDA-MB-231 and HCC1937 cells (Fig. 2J–L). Accordingly, knockdown of ILT4 significantly decreased the migration and invasion ability of BT549 and HS578 T cells (Fig. 2M–O). To further consolidate the impact of ILT4 in tumor biologies, we used the mouse TNBC cell line E0771 to establish PIR-B overexpression and knockdown cells via lentivirus infection, and detected how PIR-B regulates the proliferation, migration and invasion of the cells. The overexpression (Fig. S2A,C) and knockdown (Fig. S2B,C) efficiency was first verified at both the mRNA and protein levels. As observed in human cell lines, overexpression of PIR-B increased the proliferation, migration and invasion of E0771 cells, whereas knockdown of PIR-B decreased these abilities (Fig. S2D–H). These findings clearly suggest that ILT4 directly regulates the proliferation, migration and invasion of TNBC cells, which are crucial hallmarks for tumor growth and metastasis.

Tumorigenesis depends on the reprogramming of cellular metabolism to acquire necessary nutrients and maintain tumor cell viability and metastasis (Pavlova and Thompson, 2016). Of the various metabolic adaptations, the glycolytic phenotype represents the most profound one in TNBC (Arundhathi et al., 2021). These studies prompted us to determine whether tumor-derived ILT4 regulates glucose metabolism of TNBC cells. We first examined the changes of the key metabolic enzyme genes involving glucose metabolism in tumor cells upon ILT4 overexpression or knockdown in human TNBC cell lines. These molecules include glucose transporters GLUT1 (also known as SLC2A1) and GLUT3, glycolysis-related enzymes glucose-6-phosphate isomerase (GPI), lactic dehydrogenase (LDH), enolase 1 (ENO1), hexokinase 2 (HK2), and pyruvate kinase muscle 2 (PKM2) (Li et al., 2019). We found that mRNA expression of glucose metabolism enzymes was varied in different tumor cells upon ILT4 regulation. Most molecules were upregulated with ILT4 overexpression and downregulated with ILT4 knockdown, but the variation trend for metabolic molecules in each cell line was not consistent, suggesting that there is heterogeneity in metabolic characteristics between different cell lines. However, ILT4 overexpression in MDA-MB 231 and HCC1937 significantly upregulated the gene expression of GLUT3 and PKM2, the key enzymes for glucose uptake and glycolysis (Fig. 3A,B). Meanwhile, ILT4 knockdown in HS578 T and BT549 markedly downregulated the mRNA levels of GLUT3 and PKM2 (Fig. 3C,D). We then conformed ILT4-regulated GLUT3 and PKM2 expression using western blotting. As expected, ILT4 overexpression or knockdown remarkably increased or decreased the protein levels of GLUT3 and PKM2 (Fig. 3; Fig. S3A,B). Then using a glucose uptake and lactate assay, we demonstrated the direct control of ILT4 on glucose uptake and glycolysis. As shown in Fig. 3F,G, ILT4 overexpression significantly increased glucose uptake and lactate production, while ILT4 knockdown decreased glucose uptake and lactate production, suggesting that high ILT4 expression in tumor cells boosts the process of glucose metabolism. Moreover, we confirmed our findings in mouse E0771 cells. As expected, overexpression or knockdown of PIR-B in E0771 markedly upregulated or downregulated GLUT3 and PKM2 expression both at mRNA and protein levels (Fig. S3C–E). To further explore the role of ILT4-reprogrammed glucose metabolism on TNBC malignant biologies, we pretreated ILT4-overexpression tumor cells with either glucose transporter inhibitor KL-11743 or PKM2 inhibitor compound 3k and examined their proliferation, migration and invasion abilities. We found inhibition of either GLUT3 or PKM2 prevented ILT4-induced glucose uptake and lactate production (Fig. S4A,B). Consistent with our previous functional studies, overexpression of ILT4 promoted the proliferation ability of MDA-MB 231 and HCC1937, whereas inhibition of either GLUT3 or PKM2 prevented ILT4-upregulated tumor cell proliferation (Fig. 3H–J; Fig. S4C,D). Similarly, inhibition of either GLUT3 or PKM2 reversed ILT4-induced tumor cell migration and invasion (Fig. 3K,L; Fig. S4E–G). We also repeated our results in mouse E0771 cells, and similar findings were obtained – that treatment with either KL-11743 or compound 3k reversed PIR-B-upregulated tumor cell proliferation (Fig. S5A,B), migration (Fig. S5C,D) and invasion (Fig. S5E,F). Taken together, ILT4 reprograms the metabolic phenotype of tumor cells towards aerobic glycolysis via upreguated GLUT3 and PKM2, resulting in elevated proliferation, migration and invasion of TNBC cells.

We next explored the underlying mechanisms for ILT4-regulated tumor cell metabolism and aggressive behaviors. Previous studies have shown that MAPKs ERK1 and ERK2 (ERK1/2, also known as MAPK3 and MAPK1, respectively) and phosphoinositide 3-kinase (PI3K)-AKT-mTOR pathways are potentially activated downstream of ILT4 (Gao et al., 2021, 2018; Zhang et al., 2015a); thus we investigated these signaling pathways further. Using western blotting, we verified that overexpression of ILT4 in MDA-MB-231 and HCC1937 cells dramatically increased phosphorylation of AKT (p-AKT; antibody recognizes all isoforms), mTOR and ERK1/2 (Fig. 4A; Fig. S6A), whereas knockdown of ILT4 in HS578 T and BT549 cells significantly downregulated phosphorylation of AKT, mTOR and ERK1/2 (Fig. 4B, Fig. S6B). To determine whether they are involved in ILT4-reprogrammed glucose metabolism, we used the specific pharmacological inhibitors MK2206, rapamycin and U0126 to block AKT, mTOR and ERK1/2, respectively, in ILT4 overexpressing MDA-MB-231 and HCC1937 cells. Using RT-qPCR analyses, we found that AKT inhibitor MK2206 and mTOR inhibitor rapamycin, rather than ERK1/2 inhibitor U0126, significantly reduced ILT4-upregulated GLUT3 and PKM2 expression (Fig. 4C,D). Then we confirmed our results by Western blot analysis, and obtained the similar results. As shown in Fig. 4E–G and Fig. S6C–E, addition of MY2206 and rapamycin, but not U0126, reversed ILT4 overexpression-induced GLUT3 and PKM2 upregulation. In addition, ILT4 overexpression enhanced glucose uptake and lactate production, whereas blockade of AKT and mTOR signaling significantly prevented these effects (Fig. 4H,I). These findings clearly reveal that ILT4 activates AKT-mTOR signaling pathway in TNBC cells, leading to increased GLUT3 and PKM2 expression, and glycolysis.

Next, we wanted to build a causative link between AKT-mTOR signaling and malignant behaviors of TNBC. We pretreated ILT4-overexpressing TNBC cells with either the AKT inhibitor MK2206 or the mTOR inhibitor rapamycin, and examined their proliferation, migration and invasion abilities. As revealed by EDU analysis, ILT4 overexpression promoted the proliferation of MDA-MB-231 and HCC1937 cells, whereas addition of MK2206 or rapamycin prevented ILT4-upregulated tumor cell proliferation (Fig. 5A–C). Consistent with this, ILT4 overexpression upregulated Ki-67 levels, whereas MK2206 or rapamycin reversed ILT4-induced Ki-67 expression in MDA-MB-231 and HCC1937 cells (Fig. 5D). In addition, addition of MK2206 or rapamycin prevented ILT4-induced tumor cell migration (Fig. 5E,F) and invasion (Fig. 5G,H). Then we confirmed our results in mouse E0771 cells and found that treatment with either MK2206 or rapamycin restored PIR-B-induced tumor proliferation (Fig. S5A,B), migration (Fig. S5C,D) and invasion (Fig. S5E,F). In summary, the ILT4-activated AKT-mTOR signaling pathway mediates metabolic reprogramming and aggressive behaviors of TNBC cells.

Our in vitro study has clearly shown that ILT4/PIR-B reprogrammed glucose metabolism via induction of GLUT3 and PKM2 expression, resulting in tumor proliferation and mobility. Next, we aim to determine whether ILT4/PIR-B is critical for tumor growth and metastasis in vivo. TNBC transplantation and metastasis models were established in 6–8-week-old immunocompetent female C57BL/6 mice to determine how PIR-B in tumor cells affects tumor development. In tumor transplantation models, each mouse was subcutaneously injected with ∼10⁶ E0771 cells infected with lentiviruses carrying PIR-B gene or control vector. Tumor volume was assessed every 4 days. At the end of the experiments, tumors were separated to weigh, and embedded in paraffin for further immunohistochemistry (IHC) staining. As shown in Fig. 6A, tumors with PIR-B overexpression grew significantly faster than the control group. Furthermore, tumor sizes (Fig. 6B) and weights (Fig. 6C) collected from the E0771 with PIR-B overexpression group on day 24 post inoculation were significantly larger than those in control group. In contrast, knockdown of PIR-B in E0771 cells slowed down tumor growth, and decreased tumor sizes and weights compared with control group (Fig. 6D–F). In tumor metastasis models, 5×10⁵ E0771 cells with PIR-B overexpression or knockdown were injected into C57BL/6 mice via tail veins. At 42 days after tumor injection, mice were killed, and their lungs were separated to calculate the metastatic nodules. As shown in Fig. 6G–I, overexpression of PIR-B in E0771 remarkably increased the number of metastatic nodules both on the surface and inside of the lungs. Consistent with this, knockdown of PIR-B in E0771 cells decreased lung metastasis (Fig. 6J–L). To verify the correlation of PIR-B with glucose metabolism, IHC was performed to elucidate the expression of GLUT3 and PKM2 in transplanted tumors from each group. As expected, tumors with PIR-B overexpression showed dramatically elevated GLUT3 and PKM2 levels (Fig. 6M,N), whereas tumors with PIR-B knockdown displayed decreased GLUT3 and PKM2 expression (Fig. 6O,P). Collectively, PIR-B facilitates TNBC growth, metastasis and glycolysis via upregulated-GLUT3 and PKM2 in vivo, whereas inhibition of PIR-B represses TNBC progression.

BC is the most commonly diagnosed malignancy worldwide, in which TNBC represents the most aggressive subtype (Foulkes et al., 2010; Reis-Filho and Pusztai, 2011). The absence of druggable targets like estrogen receptor (ER), progesterone receptor (PR) and HER2 expression in TNBC cells has brought a great hurdle to precision anti-tumor therapy. Development of target therapy based on novel tumorigenic mechanisms are quite essential for TNBC treatment. In the past decade, impressive advances have been made in targeted therapies for TNBC, including PD-L1 and PD-1 inhibitors, PARP inhibitors and anti-TROP-2 antibody drug conjugates (Bardia et al., 2017; Bardia et al., 2019; Fasching et al., 2018; Marra et al., 2019; Robson et al., 2019; Woodward, 2020). However, less than 20% of TNBC patients carried germline BRCA1 or BRAC2 mutations (Won and Spruck, 2020), and only 40% of TNBC tumor cells and tumor-associated immune cells expressed PD-L1 (Steiner and Tan, 2021). Therefore, the large majority of TNBC patients cannot benefit from these strategies and require novel target therapies. Herein, we found that TNBC displayed the most abundant ILT4 expression among all BC subtypes. High ILT4 expression predicts poor patient outcomes in TNBC. Enriched ILT4 promoted TNBC cell proliferation, migration and invasion in vitro, as well as tumor growth in vivo. Mechanically, ILT4 reprogrammed aerobic glycolysis of tumor cells via AKT-mTOR signaling-mediated GLUT3 and PKM2 overexpression. Using mouse tumor models, we further confirmed that ILT4 inhibition in TNBC reduced tumor growth and GLUT3 and PKM2 expression in vivo.

ILT4 is an inhibitory myeloid molecule which negatively regulates their activation and immune response (Chen et al., 2018; Gao et al., 2018; Liang et al., 2008; Ravetch and Lanier, 2000). Recently, accumulated evidence has shown that ILT4 is overexpressed or highly induced in various solid tumors, including non-small cell lung cancer (NSCLC), colorectal cancer, esophageal carcinoma and pancreatic cancer (Cai et al., 2019; Carbone et al., 2015; Li et al., 2020; Sun et al., 2008; Warnecke-Eberz et al., 2016; Zhang et al., 2015a,b). Tumor-derived ILT4 acts as a pluripotent factor in tumor promotion, either directly via controlling malignant tumor cell biologies, or indirectly via creating an immunosuppressive TME by interaction with T cells or tumor-associated macrophages (TAMs) (Chen et al., 2018, 2021; Gao et al., 2021; Li et al., 2020; Yang et al., 2021). However, the expression and function of ILT4 in TNBC remains unclear. In this study, we first observed the highest expression level of ILT4 in TNBC among all BC subtypes, which prompt us to focus our study on TNBC. Then we discovered a positive correlation between ILT4 levels and poor patient outcomes both in our patient cohort and in public database. Using gain-of-function and loss-of-function strategies, we demonstrated the pro-tumoral role of ILT4 through induction of malignant tumor biologies. These findings indicates that ILT4 is an important driver for TNBC progression and could be a prognosis biomarker for TNBC patients. More importantly, our in vivo results clearly revealed the potential of ILT4 inhibition on TNBC treatment. Our study identified a novel driver for TNBC progression and proposed a promising strategy to combat TNBC by targeting ILT4.

Metabolic reprogramming is an emerging hallmark of TNBC cells, which enable them to fuel cell proliferation and biosynthetic (Wang et al., 2020). Of the various metabolic adaptations, the most profound is rewiring the metabolism towards the glycolytic phenotype (Arundhathi et al., 2021). It is reported that TNBC cells have a higher glycolytic gene expression signature than non-TNBC cells (Lanning et al., 2017), and suppression of glycolysis has a greater inhibitory effect on TNBC (O'Neill et al., 2019), suggesting the higher demand for glucose utilization in TNBC. In our effort to explore the mechanism for ILT4-controlled tumor biologies, we identified increased glucose uptake and lactate (glycolysis metabolite) production by ILT4 overexpression, supporting the crucial role of reprogrammed glycolysis in ILT4-induced proliferation and mobility. We also screened out the key enzymes mediating uptake and glycolysis of glucose, GLUT3 and PKM2. Tumor cells often upregulate glucose transporters to get more glucose molecules, among which GLUT3 has a higher affinity for glucose absorption and transport capacity (Ali et al., 2019). Overexpression of GLUT3 was observed to promote tumor progression and distant metastasis (Ali et al., 2019; Peng et al., 2021). Decreased glucose uptake by cancer cells due to GLUT3 inhibition might be a potential therapeutic method in cancer diseases. Whereas glucose uptake is the premise of Warburg effect, glycolysis represents the core step for energy production. PKM2 is the most important rate-limiting enzyme of glycolysis. Its expression and low enzyme activity endow cancer cells with the glycolytic phenotype, promoting rapid energy generation and flow of glycolytic intermediates into collateral pathways to synthesize nucleic acids, amino acids and lipids (Anastasiou et al., 2012; Hsu and Hung, 2018). Hence, glycolysis has emerged as a novel anti-metabolic target in tumor progression. Upregulation of GLUT3 and PKM2 by ILT4 a provide mechanistic explanation for ILT4-remodeled glucose metabolism and tumorigenesis.

In addition to GLUT3 and PKM2, we also demonstrated a causative link between AKT-mTOR signaling and ILT4-induced glycolysis. The PI3K-AKT-mTOR pathway has been implicated as a key regulator of metabolic functions, especially that of glucose uptake and glycolysis (Holloway and Marignani, 2021). AKT and mTOR inhibitors have shown promising activity in some tumors (Hay, 2005; Janku et al., 2018; Tewari et al., 2022), but their value in TNBC treatment has not been clearly defined. Currently, several phase I clinical trials are investigating the feasibility of combined chemotherapy regimens with different AKT-mTOR inhibitors including Ipatasertib (Kim et al., 2017; Oliveira et al., 2019), Capivasertib (Kalinsky et al., 2021), Everolimus (Gonzalez-Angulo et al., 2014) and Temsirolimus (Basho et al., 2017) in TNBC. Our previous study has indicated that tumor-derived ILT4 activates mTOR signaling and contributes to the upregulation of B7-H3 (also known as CD276) and immunosuppression (Zhang et al., 2015b). Consistent with that, here we confirmed that ILT4 activated the AKT-mTOR pathway and clarified its role in TNBC metabolism and tumorigenesis. This finding might help to determine the patient subpopulation who could benefit from ILT4 inhibition and provide potential combination treatment strategies.

In conclusion, we identified that ILT4 is enriched in human TNBC, predicting poor clinical prognosis. Tumor-derived ILT4 induces malignant behaviors of TNBC cells, including proliferation, migration and invasion. Mechanically, ILT4 reprograms tumor metabolism towards the glycolysis phenotype via activation of AKT-mTOR signaling and subsequent overexpression of GLUT3 and PKM2. ILT4 inhibition suppresses TNBC growth in vivo and shows promising anti-tumor activity. Our study identified ILT4 as an unfavorable prognostic biomarker as well as a novel driver for TNBC progression and proposes a promising strategy to combat TNBC by targeting ILT4.

A total of 136 TNBC patient specimens were collected, including 35 from Jinan Central Hospital affiliated to Shandong First Medical University from 2014 to 2020, and 101 from Shandong Cancer Hospital during 2015. The median age at diagnosis was 56 (27–86). Of these 136 patients, 50 TNBC tissues and their paired normal breast tissues, as well as another 10 luminal and 10 Her2-positive BC tissues (from Jinan Central Hospital affiliated to Shandong First Medical University), were tested by IHC to compare ILT4 expression level. All human tissue experiments were approved by the Medical Ethics Committee of Jinan Central Hospital (approval number: GZR2020-006-01) and Shandong Cancer Hospital (approval number: SDTHEC: 2021003138). All clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki. All patients provided written informed consent for tissue usage. All patients were followed up by telephone, and the last censor date was March 12, 2022.

Tumor tissues from mice and patients were fixed with 4% paraformaldehyde at room temperature for 24 h, and then embedded in paraffin. All tissues were sectioned into 4 µm slices. IHC was performed according to the manufacturer's instructions as we previously described (Li et al., 2021). The following primary antibodies were used: anti-PIRB antibody (1:100; Immunoway; cat. no. YN1914), anti-ILT4 antibody (1:50; Origene; cat. no. TA323297), anti-GLUT3 antibody (1:200; Proteintech; cat. no. 20403-1-AP), anti-PKM2 antibody (1:800; Cell Signaling Technology; cat. no. 4053). At least five fields were simultaneously reviewed by two independent pathologists at 400× magnification in a randomized manner, with the pathologists being unaware of the identity of the tissues. The percentage of microscopically positive cells and staining intensity were recorded and multiplied to obtain the immunoreactivity score. Staining intensity scores were as follows: 0, none; 1, weak; 2, intermediate; 3, strong. Positive cell proportion scores were as follows: 0, none; 1, ≤25%; 2, 26–50%; 3, 51–75%; 4, >75%. The cut-off scores for high and low expression were≥or<the median.

The online transcriptome database Ualcan (http://ualcan.path.uab.edu/analysis.html) was used to analyze the expression of ILT4 in normal tissues and different subtypes of breast invasive carcinoma and its effect on survival.

Human BC cell lines MDA-MB-231, HCC1937, BT549, HS578 T, human mammary epithelial cell line MCF10A, and murine BC cell line E0771 were purchased from the Cell Resource Center of the Chinese Academy of Sciences (Beijing, China). All cell lines were tested and authenticated for absence of Mycoplasma, and the expected genotypes, drug response and morphology in the laboratory. MDA-MB-231, BT549 and HCC1937 were cultured with RPMI-1640 (Gibco; cat. no. C11875500BT) containing 10% fetal bovine serum (FBS) (Gibco; cat. no. 10270-106). HS578 T and E0771 were cultured with DMEM (Gibco; cat. no. C11995500BT) containing 10% FBS. HS578 T and BT549 requires an additional 10 µg/ml insulin (Beyotime; cat. no. P3376). MCF10A cells were cultured in DMEM/F12 supplemented with 10 µg/ml insulin, 20 ng/ml epidermal growth factor, 100 ng/ml cholera virus, 0.5 µg/ml hydrocortisone and 5% horse serum (MEpiCGS; ScienCell; cat. no. 7652). All cultures were maintained at 37°C in a humidified incubator supplied with 5% CO₂. For inhibition experiments, tumor cells were treated with the GLUT3 inhibitor KL-11743 (MCE; cat. no. HY-145597/CS-0376607) at 1 μM, PKM2 inhibitor (compound 3k) (Selleck; cat. no. S8616) at 500 nM, the ERK1/2 inhibitor U0126 (MCE; cat. no. HY-12031A) at 10 μM, the AKT1–AKT3 inhibitor MK2206 (Beyotime; cat. no. SF2712) at 2 μM or the mTOR inhibitor rapamycin (MCE; cat. no. 53123-88-9) at 100 nM.

Total RNA was extracted from TNBC cell lines using the total RNA extraction kit (Fastagen; cat. no. RNAfast200), and the purified total RNA was then reversely transcribed to cDNA using a RT kit (Accurate Biology; cat. no. AG11706). The mRNA expression of ILT4, PIR-B or key enzymes related to glycolysis in each sample were determined by qRT-PCR using SYBR Green Premix Pro Taq HS qPCR Kit (Accurate Biology; cat. no. AG11701) according to the manufacturer's instructions. The mRNA expression was normalized to the relative quantity of β-actin or GAPDH using the 2^–ΔΔCt method. All experiments were performed in triplicate. The specific primers used are listed as in Table S2.

Whole-cell lysates of ILT4-overexpressing or downregulated tumor cells were prepared for western blotting as we previously described (Li et al., 2021). The protein bands were visualized using Immobilon western chemiluminescent HRP substrate. The primary antibodies used in western blotting were as follows: anti-ILT4 (1:1000; Immunoway; cat. no. YT5595), anti-PIR-B (1:1000; Immunoway; cat. no. YN1914), anti-ERK1/2 (1:1000; Cell Signaling Technology; cat. no. 4695T), anti-phospho-ERK1/2 (1:1000; Santa Cruz Biotcehnology; cat. no. SC-7383), anti-AKT (1:1000; Abcam; cat. no. Ab185633), anti-phospho-AKT (Ser473) (1:1000; Cell Signaling Technology; cat. no. 4060T), anti-mTOR (1:1000; Cell Signaling Technology; cat. no. 2983T), anti-phospho-mTOR (1:1000; Cell Signaling Technology; cat. no. 5536T), anti-Ki67 antibody (1:1000; Abcam; cat. no. ab16667), anti-GLUT3 (1:1000; PTG; cat. no. 20403-1-AP and 1:500; Santa Cruz Biotcehnology; cat. no. Sc-74399), anti-PKM2 (1:1000; Cell Signaling Technology; cat. no. 4053T).

ILT4/PIR-B overexpression and knockdown lentiviruses were purchased from Genechem Inc. Human/mouse TNBC cells were infected with lentivirus when the cell density reached 30–50% according to the manufacturer's instructions. The infection efficiency was evaluated using a fluorescence microscope 72 h after lentivirus infection. The stable cell lines were selected using 2 μg/ml puromycin for 48–72 h and then cultured in 1 μg/ml puromycin for 10 days. Extraction of total mRNA and protein and detection of over-expression/knockdown efficiency was then performed.

Glucose uptake in ILT4 overexpression and knockdown TNBC cells was detected using a glucose uptake colorimetric assay kit (BioVision; cat. no. K676-100), and lactate production was detected using a glycolysis cell-based assay kit (Cayman; cat. no. 600450) according to the manufacturer's instructions. For inhibition experiments, the cells were pretreated with the AKT inhibitor MK2206 or the mTOR inhibitor rapamycin for 48 h before glucose uptake and lactate assays. All experiments were repeated three times.

A total of 5×10⁴ ILT4 overexpression or knockdown TNBC cells (with and without signaling inhibitors pretreatment) in serum-free medium were inoculated into the upper chamber of a Transwell plate with 8-µm pore polycarbonate membranes, and medium with 20% FBS was added in the lower chamber as a chemoattractant. Matrigel matrix (BD Science) was coated on the membranes for the invasion assay. After incubation at 37°C for 24 h for the migration assay and 36 h for the invasion assay, the cells were fixed with methanol for 30 min and stained with 0.1% Crystal Violet for 20 min, and then counted under a light optic microscope from five representative fields.

The cell proliferation was determined using EdU cell proliferation kit with Alexa Fluor 594 (Beyotime Biotechnology; cat. no. C0078S). 2×10³–4×10³ ILT4 overexpression or knockdown TNBC cells (with/without signaling inhibitors pretreatment) were inoculated in 96-well plates. After 24 h, the cells were incubated with 10 µm EdU for 2 h at 37°C, and then fixed with paraformaldehyde and permeabilized with 0.5% Triton X-100 at room temperature. Subsequently, the cells were incubated in Click Additive Solution and stained with DAPI under dark condition. The fluorescence images from five random fields were then photographed under a fluorescence microscope. The cell proliferation rate was the ratio of EdU incorporation cells to the total number of cells.

For the CCK8 assay (Genechem; cat. no. GCPE0293559), 2×10³ cells per well were seeded into the 96-well plates and cultured for the indicated times. Then, 10 μl of CCK8 solution was added into each plate and incubated for 2 h. The absorbance at 450 nm was measured by a microplate reader.

Animal studies were approved by the Institutional Animal Care Committee at Jinan central hospital affiliated to Shandong First Medical University and conducted according to the NIH animal usage guidelines and Chinese regulations and standards for laboratory animal usage (approval number: GZR2020-006-01).

6–8-week-old female C57BL/6 mice were purchased from SPF (Beijing) Biotechnology Company and housed under specific pathogen-free (SPF) conditions. 10⁶ PIR-B-overexpression or knockdown E0771 cells were subcutaneously injected into the right axilla of C57BL/6 mice (n=6). Tumor volumes were measured every 4 days using a digital caliper and calculated as length×width² /2. Mice were killed (by cervical vertebral dislocation) and tumors were isolated and weighed after 24 or 28 days when tumors had grown to a size limit of 2 cm. Tumor tissues were embedded in paraffin and dissected for IHC staining. In TNBC metastasis models, mice were injected with 5×10⁵ tumor cells via the tail veins. At 42 days after injection, the mice were killed, and their lungs were separated for further study. Visible lung surface macrometastatic spots were counted using a dissecting microscope (Nikon). Then, the lung tissues were embedded in paraffin and sectioned. Hematoxylin-eosin (HE) staining was performed to determine the number of metastatic lesions.

GraphPad Prism 9.0 software was used for statistical analysis. Data were expressed as mean±s.d. or mean±s.e.m. Pearson's χ-squared test and a Student's unpaired two-tailed t-test were used to analyze the associations between ILT4 expression and clinicopathological variables. OS was determined as the duration from the date of initial diagnosis until death or last follow-up. Survival curves were drawn using Kaplan–Meier method and compared using un-adjusted log-rank test. Comparisons between two groups were performed using unpaired two-tailed Student's t-test, and P<0.05 was considered statistically significant (*P<0.05; **P<0.01; ***P<0.001; ****P<0.0001).

We thank Jianbo Zhang of the Department of Pathology, Shandong Tumor hospital affiliated to Shandong First Medical University and Fei Yang of the Department of Pathology, Jinan central hospital affiliated to Shandong First Medical University, for helping with the collection of TNBC samples and preparing tissue sections. We acknowledge Xiaowen Xia of Experimental Animal Center, Jinan Central Hospital Affiliated to Shandong First Medical University for conducting the injection of cancer cells.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260964.reviewer-comments.pdf