Magrolimab (Hu5F9-G4) promotes macrophage M1 polarization and is associated with enhanced autophagy in colorectal cancer

Highlights

• •High M2d macrophage infiltration correlates with poor prognosis in colorectal cancer (CRC).
• •Hu5F9-G4 promotes M2d-to-M1 macrophage polarization, inhibits CRC cell proliferation and migration, and enhances autophagy in macrophages.
• •In vivo studies show that Hu5F9-G4 reduces tumor growth and modulates immune cell populations in CRC-bearing mice.
• •Integrative analyses identify CDKN1A and MAP1LC3B as potential therapeutic targets associated with the autophagy pathway.
• •Hu5F9-G4 has therapeutic potential in CRC by modulating macrophage polarization, tumor progression, and autophagy-related pathways.

Abstract

Background

The role of macrophage infiltration in colorectal cancer (CRC) and its implications for prognosis remains uncertain. While macrophage immune checkpoint inhibitors have shown efficacy in treating solid tumors, their impact on CRC progression and immune regulation is unclear.

Purpose

This study investigates the effect of the macrophage immune checkpoint inhibitor Hu5F9-G4 on the immune microenvironment, macrophage behaviour, and CRC progression to identify potential therapeutic targets.

Methods

A total of 73 CRC samples were analysed to evaluate the association between M2d macrophage infiltration and prognosis. In vitro studies examined the effect of Hu5F9-G4 on M2d-to-M1 polarization, CRC cell proliferation, and migration. In vivo experiments assessed tumor growth and macrophage polarization in CRC-bearing mice treated with Hu5F9-G4. Single-cell transcriptome analysis, bioinformatics, and clinical validations were conducted to identify potential autophagy-related therapeutic targets.

Results

High M2d macrophage infiltration correlated with poor CRC prognosis. Hu5F9-G4 treatment significantly promoted M2d-to-M1 polarization, inhibited CRC cell proliferation and migration, and increased autophagy in macrophages. In vivo, Hu5F9-G4 reduced tumor growth and modified immune cell populations. Integrative analyses identified CDKN1A and MAP1LC3B as potential therapeutic targets linked to the autophagy pathway.

Conclusion

Hu5F9-G4 demonstrates potential as a therapeutic agent in CRC by promoting M2d-to-M1 macrophage polarization, suppressing tumor progression, and influencing the autophagy pathway. These findings highlight CDKN1A and MAP1LC3B as promising targets for CRC therapy.

Introduction

It is well known that there are very few effector T cells in the tumor-infiltrating leukocytes of Colorectal cancer (CRC), and they cannot be activated. This creates a tumor microenvironment (TME) that is typically classified as a “cold tumor”. In response, researchers have shifted their focus to developing immunotherapies that target immune cells other than T cells [1]. Recently, the use of activated macrophage immune checkpoint blockade therapy has been successful in some solid tumors, which has pioneered the way for immunotherapy advancements in treating cold tumors, including CRC [2].

However, there are few reports on macrophage infiltration in CRC and the relationship between the polarization states of various macrophages and CRC prognosis remains underexplored [[3], [4], [5], [6]].

Magrolimab (Hu5F9-G4) has demonstrated promising efficacy in preclinical studies, particularly in the treatment of hematological malignancies [7,8]. Based on this, Hu5F9-G4 (Magrolimab) has also been evaluated in ongoing clinical trials for solid tumors, including colorectal cancer. These include a Phase Ib/II study of Magrolimab combined with cetuximab in advanced solid tumors and colorectal cancer (ClinicalTrials.gov identifier: NCT04313881), and a Phase I/II study of Magrolimab in combination with FOLFIRI ± bevacizumab in metastatic colorectal cancer (NCT04751383). As of October 2025, both trials remain active but without published results. From the clinical trial project registration information, the characteristics of enrolled patients (tumor immune infiltrating cells and other information) of these two projects are not limited.

The efficacy of macrophage immune checkpoint inhibitors on CRC is unknown. This study aims to explore the effect of macrophage immune checkpoint inhibitors Hu5F9-G4 on the immune microenvironment and progression of CRC.

Materials and methods

Study design

A total of 73 (380 cases were collected, and 73 cases were screened, Fig. 1A) postoperative specimens from CRC patients were collected, along with the recording of clinical-pathological factors. The study has been approved by the Ethics Committee and is exempt from obtaining patient informed consent. The mIHC technology was used to detect CD8+ T cells, NK cells, and macrophages in CRC samples, analyzing the differences in immune cell infiltration in the tumor parenchyma and stroma, as well as their correlation with prognosis. RAW264.7 cells were induced into M2d macrophages with IL-6 treatment, and the phenotypic markers of M2d macrophages were detected. A non-contact co-culture system was established using Transwell chambers to study the interaction between M2d macrophages treated with Magrolimab (Hu5F9-G4) intervention and CRC cells. HE staining was performed on CRC tissues to observe pathological changes. Gene and protein expression were detected by RT-PCR and Western blot; CCK8 was used to test cell viability, plate cloning experiments assessed cell cloning ability, Transwell experiments measured cell migration ability, and EDU detected cell proliferation capacity. An MC-38 cell subcutaneous xenograft tumor model was established in C57BL/6 N mice, and Magrolimab (Hu5F9-G4) intervention was performed to observe tumor growth. Immune cell alterations within the tumor tissue and peripheral blood of CRC-bearing mice were analyzed. The tumor tissues of the disease group and the treatment group of mouse models were analyzed to screen for differences in gene expression related to autophagy and to conduct functional analysis and cell-cell communication research. The HPA database was used to analyze the expression of autophagy-related proteins in CRC, and the GEPIA2 database was utilized to analyze the impact of their expression levels on the survival of CRC patients. Detailed experimental protocols, including reagent sources, incubation conditions, and data analysis parameters, are provided in the Supplementary Methods.

Fig. 1.

Flow chart of case enrollment and infiltration of tumor immune cells in CRC and its correlation with clinical pathological factors and prognosis. (A) Flow chart of case enrollment. (B) Multiplex immunofluorescence staining images of tumor-associated immune cells in CRC tissue. PD-1 (green), PD-L1 (yellow), CD8 (pink), CD68 (cyan), IL-10 (red), CD3 (pink), CD4 (red), CD20 (green), CD56 (cyan), FoxP3 (yellow). The upper images are fusion images of immune cells, and the lower images are positive control images, negative control images, and HE images, respectively. (C, D) Cell counts and proportions of CD8+ T cells, macrophages, and NK cells in the IT and TR of CRC. (E, F) PFS and OS of CD8+ T cells (high and low infiltration) in IT. (G, H) PFS and OS of CD8+ T cells (high and low infiltration) in TR. (I, J) PFS and OS of M1 macrophages (high and low infiltration) in IT. (K, L) PFS and OS of M1 macrophages (high and low infiltration) in TR; (M, N) PFS and OS of M2d macrophages (high and low infiltration) in IT. (O, P) PFS and OS of M2d macrophages (high and low infiltration) in TR. (Q, R) PFS and OS of CD56bright-NK cells (high and low infiltration) in IT. (S, T) PFS and OS of CD56bright-NK cells (high and low infiltration) in TR; (U, V) PFS and OS of CD56dim-NK cells (high and low infiltration) in IT. (W, Z) PFS and OS of CD56dim-NK cells (high and low infiltration) in TR. IT: Intratumoral Region; TR: Tumor Rim.

Statistical analysis

Data were analyzed using SPSS 24.0, and graphs were plotted using the GraphPad Prism software (version 7.0). The median represented skewed data. Analysis of the relationship between various clinicopathological factors and protein expression, as well as correlation analysis between proteins, was performed using the Wilcoxon rank-sum test. The cumulative survival time and rate were calculated using the Kaplan–Meier method, and the survival test was conducted using the log-rank test. Independent sample t-tests were used to analyze differences between two groups, while a one-way analysis of variance was used to analyze differences between multiple groups' means. P < 0.05 Demonstrated a notable and statistically meaningful difference.

Results

Basic clinical information

73 CRC patients (380 cases were collected and 73 cases were screened, Fig. 1A) were included (sex: male in 40 cases, female in 33 cases); age (median, 55): >55 years: 36 cases, ≤55 years: 37 cases; Mismatch Repair (MMR) status: deficient MMR (dMMR) in 11 cases, proficient MMR (pMMR) in 62 cases; pT staging: stages T1–2: 16 cases, stages T3–4: 57 cases; TNM staging: stages I-II: 36 cases; stage III-IV: 37 cases.

All patients were followed up until June 2023 or until death, with the longest follow-up time being 187 months and the shortest follow-up time being 3 months. Progression-free survival (PFS) was defined as the time from the first day after surgery to the time of tumor progression (in any aspect) or death from any cause. The longest PFS in this study was 187 months, and the shortest was 3 months. Overall survival (OS) was defined as the time from the first day after surgery to death from any cause. This study's most prolonged OS was 187 months, and the shortest was 6 months.

Infiltration of CD8+ T cells, macrophages, and NK cells in CRC

Macrophages accounted for the highest proportion of all CRC tissues, followed by NK and CD8+ T cells. The infiltration levels of M1 and M2d macrophages in the intratumoral region were significantly lower than those in the tumor rim (P = 0.0298 and P = 0.0182, respectively) (Fig. 1CD). The infiltration levels of CD8+ T cells, macrophages, and NK cells in the intratumoral region and tumor rim were not significantly correlated with pathological factors of CRC (sex, age, MMR status, T stage, and TNM stage) (P > 0.05) (Supplementary Table 2).

Patients with CRC with low infiltration of M2d macrophages in the intratumoral region had longer PFS and OS (P = 0.0378 and P = 0.0215, respectively). There was no significant correlation between CD8+ T cells and NK cells in the intratumoral region and the tumor rim or between the PFS and OS of patients with CRC (Fig. 1E-Z).

M2d macrophage model and the polarization effect

The murine monocyte-macrophage leukemia cell line RAW264.7 is primarily round or oval and grows adherently. After induction with 40 ng/mL of IL-6 for 12 h, some RAW264.7 cells became regularly round, while others became spindle-shaped (Fig. 2A). The identification of M2d macrophage markers was detected by RT-PCR and ELISA. The RT-PCR experimental results showed that compared with RAW264.7 cells, the mRNA expression of Arg-1 and IL-10 in M2d macrophages was significantly increased (P < 0.0001, P = 0.0002) (Fig. 2B). ELISA results showed that compared with RAW264.7 cells, the secretion of IL-10 was significantly increased (P = 0.0475) (Fig. 2C), demonstrating that the M2d macrophage model was successfully established.

Fig. 2.

Identification of the M2d Macrophage Model and the Polarization Effect of Hu5F9-G4 on M2d Macrophages. (A) Optical microscope images of RAW264.7 and M2d macrophages (× 100). (B) RT-PCR detection of M2d macrophage markers Arg-1 and IL-10 mRNA expression in the two cell types. (C) ELISA detection of M2d macrophage marker IL-10 expression in the two cell types. (D) ELISA detection of M2d macrophage marker IL-10 expression in the M2d macrophage blank control group, M2d macrophage + IL-2 group, and M2d macrophage + Hu5F9-G4 group. (E) RT-PCR detection of M1 macrophage markers CD86, TNF-α, and M2d macrophage markers Arg-1, IL-10 mRNA expression in the M2d macrophage blank control group, M2d macrophage + IL-2 group, and M2d macrophage + Hu5F9-G4 group. (F-H) WB detection of M2d macrophage markers Arg-1, IL-10, and M1 macrophage marker TNF-α protein expression in the M2d macrophage blank control group, M2d macrophage + IL-2 group, and M2d macrophage + Hu5F9-G4 group.***P < 0.05, **P < 0.01, ***P < 0.001.

Experimental groups were set up as follows: M2d macrophage blank control group, M2d macrophage + IL-2 group (20 ng/mL IL-2 induced M2d macrophages for 24 h), M2d macrophage + Hu5F9-G4 group (10 µg/mL Hu5F9-G4 induced M2d macrophages for 12 h). ELISA and WB detected M2d macrophage markers, showing that compared with the M2d macrophage blank control group, the secretion of IL-10 was reduced after intervention with both IL-2 and Hu5F9-G4 (P = 0.0055 and P = 0.0036, respectively), but the effect after Hu5F9-G4 intervention was more significant (Fig. 2D).

RT-PCR was used to detect the M1 and M2d macrophage markers. The experimental results showed that compared with the M2d macrophage blank control group, after Hu5F9-G4 intervention, the mRNA expression of M1 macrophage markers CD86 and TNF-α was significantly increased (P = 0.0479; P < 0.0001), while the mRNA expression of M2d macrophage markers Arg-1 and IL-10 was significantly decreased (P < 0.0001; P = 0.0008). Compared with IL-2, after Hu5F9-G4 intervention, the mRNA expression of CD86 and TNF-α was significantly increased (P = 0.0285; P < 0.0001), and the mRNA expression of Arg-1 was significantly decreased (P < 0.0001) (Fig. 2E).

WB experimental results showed that compared with the M2d macrophage blank control group, after Hu5F9-G4 intervention, the expression of Arg-1 and IL-10 was decreased (P = 0.0447; P = 0.0131), while the expression of TNF-α was increased (P < 0.0001). Compared with IL-2, after Hu5F9-G4 intervention, the expression of Arg-1 was significantly decreased (P = 0.0037), and the expression of TNF-α was increased (P = 0.0002) (Fig. 2E-H). These results suggest that Hu5F9-G4 has a more significant effect on promoting the polarization of M2d to M1 cells than IL-2.

Autophagy level of M2d macrophages induced by Hu5F9-G4

To observe the morphological effects of Hu5F9-G4 on M2d macrophages, transmission electron microscopy was used to observe the cell morphology of the M2d macrophage blank control and M2d macrophage + Hu5F9-G4 groups. The results showed that the number of autophagic lysosomes in the M2d macrophage + Hu5F9-G4 group was greater than that in the M2d macrophage blank control group (Fig. 3A-D). WB results indicated that, compared to the M2d macrophage blank control group, the expression of Beclin-1 increased and that of P62 significantly decreased after Hu5F9-G4 intervention (P < 0.0001). Compared to IL-2, the expression of Beclin-1 increased and that of P62 significantly decreased after Hu5F9-G4 intervention (P < 0.0001) (Fig. 3E). The experimental results indicated that autophagy levels in M2d macrophages increased following Hu5F9-G4 induction.

Fig. 3.

Autophagy level of M2d macrophages induced by Hu5F9-G4. (A) Pseudo-colored transmission electron micrograph of the M2d macrophage blank control group at 1000x magnification. (B) Pseudo-colored transmission electron micrograph of the M2d macrophage + Hu5F9-G4 group at 1000x magnification. (C) Pseudo-colored transmission electron micrograph of the M2d macrophage blank control group at 8000x magnification, with autophagic lysosomes marked in red. (D) Pseudo-colored transmission electron micrograph of the M2d macrophage + Hu5F9-G4 group at 8000x magnification, with autophagic lysosomes marked in red. (E) WB detection of the protein expression of autophagy markers Beclin-1 and P62 in the M2d macrophage blank control group, M2d macrophage + IL-2 group, and M2d macrophage + Hu5F9-G4 group. *P < 0.05, **P < 0.01, ***P < 0.001.

Impact of MC-38 and CT-26 cells on proliferation and migration after co-culture with Hu5F9-G4-induced by M2d macrophages

After co-culturing CRC cells MC-38 and CT-26 with conditioned medium, M2d macrophages, M2d macrophages + IL-2, and M2d macrophages + Hu5F9-G4, the results of the colony formation assay showed that compared with the conditioned medium blank control group, the number of clone formations of MC-38 and CT-26 cells significantly increased after co-culturing with M2d macrophages (P < 0.0001; P = 0.0071). Compared with the M2d macrophage control group, the number of clone formations of MC-38 and CT-26 cells significantly decreased after co-culturing with M2d macrophages + IL-2 and M2d macrophages + Hu5F9-G4 (P < 0.0001, P < 0.0001; P = 0.0092, P = 0.0019). Clone formation was lower in the M2d macrophages + Hu5F9-G4 group than in the M2d macrophages + IL-2 group (P = 0.0004; P = 0.0033) (Fig. 4A-C).

Fig. 4.

The impact of MC-38 and CT-26 cells co-cultured with Hu5F9-G4-treated M2d macrophages on proliferation and migration ability. (A-C) MC-38 and CT-26 cells were co-cultured with the conditioned medium blank control group, M2d macrophages, M2d macrophages + IL-2 group, and M2d macrophages + Hu5F9-G4 group, and the colony formation assay was used to detect the number of clone formations in each group of cells. (D-F) Transwell migration assay was used to detect the number of migrations in each group of cells. (G-I) EdU assay was used to detect the EdU(+) cell rate in each group of cells. (J, K) CCK8 assay was used to detect the cell activity in each group of cells. *P < 0.05, **P < 0.01, ***P < 0.001.

The Transwell migration assay results showed that compared with the conditioned medium blank control group, the number of MC-38 and CT-26 cells passing through the basement membrane gel significantly increased after co-culturing with M2d macrophages (P = 0.0303; P < 0.0001). Compared with the M2d macrophage control group, the number of MC-38 and CT-26 cells passing through the basement membrane gel significantly decreased after co-culturing with M2d macrophages + IL-2 and M2d macrophages + Hu5F9-G4 (P = 0.0276, P = 0.0134; P = 0.0014, P = 0.0001). The number of MC-38 and CT-26 cells passing through the basement membrane gel was significantly lower in the CT-26 M2d macrophages + Hu5F9-G4 group than in the M2d macrophages + IL-2 group (P = 0.0186) (Fig. 4D-F).

The EdU assay results showed that compared with the conditioned medium blank control group, the EdU-positive cell rate of MC-38 and CT-26 significantly increased after co-culturing with M2d macrophages (P = 0.0002; P = 0.0031). Compared with the M2d macrophage control group, the EdU-positive cell rate of MC-38 and CT-26 significantly decreased after co-culturing with M2d macrophages + IL-2 and M2d macrophages + Hu5F9-G4 (P = 0.0428, P < 0.0001; P = 0.0272, P = 0.0024). The number of EdU-positive cells was lower in the M2d macrophages + Hu5F9-G4 group than in the M2d macrophages + IL-2 group (P = 0.0305; P = 0.0027) (Fig. 4G-I).

The CCK8 assay results showed that compared with the conditioned medium blank control group, the cell activity of MC-38 and CT-26 significantly increased after co-culturing with M2d macrophages (P = 0.0024; P = 0.0006). Compared with the M2d macrophage control group, the cell activity of MC-38 and CT-26 significantly decreased after co-culturing with M2d macrophages + Hu5F9-G4 (P = 0.0124; P = 0.0008) and was lower than that in the M2d macrophages + IL-2 group (Fig. 4J, K).

Effect of Hu5F9-G4 on subcutaneous tumor growth rate and immune cells in tumor tissues and peripheral blood in CRC mice

To observe the effect of Hu5F9-G4 on the growth rate of subcutaneous tumors in C57BL/6 N mice inoculated with MC-38 cells, physiological saline, IL-2, and Hu5F9-G4 were administered to mice via intraperitoneal injection (Fig. 5A). Subsequently, the subcutaneous tumors were dissected and photographed, and the tumor volume was recorded. The results showed that the tumor volumes in the IL-2 and Hu5F9-G4 groups were smaller than those in the physiological saline control group, and the tumor volume in the Hu5F9-G4 group was smaller than that in the IL-2 group (Fig. 4B). The experimental results indicated that the tumor growth rate in the IL-2 and Hu5F9-G4 groups was slower than that in the physiological saline control group, and that in the Hu5F9-G4 group was slower than in the IL-2 group. On the last day, the tumor volumes in the IL-2 and Hu5F9-G4 groups were significantly smaller than those in the physiological saline group (P = 0.0256 and P < 0.0001, respectively), and the tumor volume in the Hu5F9-G4 group was significantly smaller than that in the IL-2 group (P = 0.0003) (Fig. 4C). The tumor weights in the IL-2 and Hu5F9-G4 groups were lower than those in the physiological saline control group (P = 0.0001 and P < 0.0001, respectively), and the tumor weight in the Hu5F9-G4 group was lower than that in the IL-2 group (P = 0.0003) (Fig. 5D). The TGI in the IL-2 and Hu5F9-G4 groups was significantly higher than that in the physiological saline group (P = 0.0004 and P < 0.0001, respectively), and the TGI in the Hu5F9-G4 group was significantly higher than that in the IL-2 group (P = 0.0004) (Fig. 5E). The comprehensive results of the animal experiments suggest that Hu5F9-G4 has a more significant ability to inhibit tumor growth than IL-2.

Fig. 5.

The effect of Hu5F9-G4 on the growth rate of subcutaneous tumors, peripheral blood, and immune cells in tumor tissue of CRC-bearing mice. (A) Schematic diagram of treatment for CRC-bearing mice. (B) The growth of subcutaneous tumors in CRC-bearing mice in each group. (C) Comparison of subcutaneous tumor volume in CRC-bearing mice in each group. (D) Comparison of subcutaneous tumor weight in CRC-bearing mice in each group. (E) Comparison of TGI in subcutaneous tumors of CRC-bearing mice in each group. (F, G) The proportion of immune cells in the peripheral blood of CRC-bearing mice in each group. (H, I) The expression of immune cell markers iNOS, TNF-α, IL-10, Arg-1, and CD8a in the subcutaneous tumors of CRC-bearing mice in each group. *P < 0.05, **P < 0.01, ***P < 0.001.

Peripheral blood was collected from the mice for immune cell detection. Flow cytometry results showed that the proportion of CD8+ T cells in the peripheral blood of the IL-2 and Hu5F9-G4 groups was significantly higher than that in the physiological saline control group (P = 0.0129 and P = 0.0003, respectively) and higher in the Hu5F9-G4 group than in the IL-2 group (P = 0.0021). The proportion of M2d macrophages in the peripheral blood of the IL-2 and Hu5F9-G4 groups was significantly lower than that in the physiological saline control group (P = 0.0019 and P = 0.0007, respectively), and that in the Hu5F9-G4 group was lower than that in the IL-2 group (P = 0.0013) (Fig. 5F, G). The proportion of immune cells in the peripheral blood indicated that Hu5F9-G4 inhibited the polarization of M2d macrophages in the peripheral blood of CRC-bearing mice.

Subcutaneous tumors from the mice were used for immune cell detection. IHC experiment results showed that the expression of the M1 macrophage marker TNF-α in the tumor tissue of the Hu5F9-G4 group was significantly higher than that in the physiological saline control group (P = 0.0424). The expression of the M2d macrophage marker IL-10 in the tumor tissue of the Hu5F9-G4 group was significantly lower than that in the physiological saline control and IL-2 groups (P = 0.0472 and P = 0.0027, respectively). The expression of the M2d macrophage marker Arg-1 in the tumor tissues of the IL-2 and Hu5F9-G4 groups was significantly lower than that in the physiological saline control group (P = 0.0488; P = 0.0004). The expression of the CD8+ T cell marker CD8a in the tumor tissues of the Hu5F9-G4 group was significantly higher than that in the physiological saline control group (P = 0.0288) (Fig. 5HI). These results indicate that Hu5F9-G4 promotes the polarization of M2d macrophages to M1 macrophages in CRC tissues.

Single-Cell transcriptome analysis

Cell population clustering and annotation and identification of key cell clusters

The 10 × genomics scRNA sequencing dataset was obtained from three control groups (disease groups A1, A2, and A3) and three Hu5F9-G4 groups (treatment groups B1, B2, and B3) from fresh tumor tissues of the mice. The R package Seurat [9] was used to filter the data, with a post-filtering cell count of 49,261 and a gene count of 22,076 (Fig. 6A). After standardizing the data, the vst method was used to extract genes with high coefficients of variation between cells, identifying the top 2000 variable genes (highly variable genes) that were used for subsequent analysis. The names of the top 10 genes are displayed (Fig. 6B).

Fig. 6.

Identification of cell clusters and key cell clusters in six tumor samples. (A) The number of cell gene expressions, gene quantity, and mitochondrial ratio after sample quality control. (B) Screening of high-variation genes. (C) PCA analysis of samples. (D) Linear dimensionality reduction analysis. (E) Non-linear dimensionality reduction combined with RunMap function for cell distribution analysis. (F) UMAP clustering map of cells. (G) Distribution of cell clusters in different samples. (H) UMAP map of cell annotations. (I) UMAP map of cell annotations for the disease group and treatment group. (J) AddModuleScore scores of immune cells based on ARGs as a gene set. *P < 0.05, **P < 0.01, ***P < 0.001.

PCA was performed on different samples, and the cells from different samples in the control and disease groups were mixed and distributed relatively concentrated; therefore, subsequent analysis could be performed normally (Fig. 6C). Based on the principal component elbow plot, it can be seen that the p-value changed significantly before the 30th principal component, and the trend of change was smaller after the 30th principal component; therefore, dims =30 (the first 30 PCs) was selected for subsequent analysis (Fig. 6D). The functions of the Seurat packages FindNeighbors and FindClusters were used for unsupervised clustering analysis of the cells. The UMAP clustering method was used to cluster cells, and the results were visualized, with all cells eventually divided into 13 clusters (Fig. 6E-G). Based on the cell clustering results, the clustering results were annotated, and marker genes were used to annotate the cell clusters. A total of six cell types were annotated, including epithelial cells, macrophages, fibroblasts, neutrophils, NK cells, and endothelial cells (Fig. 6H, I).

To obtain key immune cells related to autophagy in the disease and treatment groups, the AddModuleScore function of the Seurat software package was used based on 215 ARGs converted from the human autophagy database to mouse as a gene set for AddModuleScore analysis. The AddModuleScore scores of the immune cells in the disease and treatment groups were calculated according to the gene set, and the differences in the AddModuleScore scores of various cell types between the disease and treatment groups were compared. Cell types with p < 0.05 were selected as key cells for this study. The results showed that the AddModuleScore scores of neutrophils and macrophages in immune cells were significantly different between the disease and treatment groups; therefore, these two cell types were selected as key cells related to autophagy in the treatment group (Fig. 6J).

Key cellular biomarkers

To identify DEGs in key cells, this study utilized the R package “Seurat” FindMarker function to screen for DEGs in macrophages and neutrophils, which are key cells in the disease group and treatment group of the single-cell dataset. The screening criteria were as follows: p < 0.05, avg_log2FC > 0.26, and pct > 0.1. A total of 1164 DEGs were identified in macrophages between the disease and treatment groups, with 597 genes upregulated and 567 genes downregulated in the treatment group. Neutrophils had 299 DEGs, with 186 upregulated and 113 downregulated genes in the treatment group (Fig. 7A).

Fig. 7.

Identification of key cellular biomarkers. (A) Volcano plot of DEGs. (B) Venn diagram of differential ARGs. (C) GO enrichment analysis of differential ARGs in macrophages. (D) GO enrichment analysis of differential ARGs in neutrophils. (E) KEGG enrichment analysis of differential ARGs in macrophages. (F) KEGG enrichment analysis of differential ARGs in neutrophils. (G) PPI network of biomarkers. (H) UMAP distribution of biomarkers across cell types. (I) Expression differences of biomarkers in key cells.

To identify the autophagy-related differential genes in CRC immune cells, the differential genes obtained from the above analysis of macrophages and neutrophils were intersected with mouse ARGs. The results showed that six autophagy-related differential genes were common to macrophages and neutrophils from CRC samples: CTSD, CDKN1A, MAP1LC3B, CFLAR, EIF4EBP1, and BCL2L1 (Fig. 7B).

To explore the biological functions involved in the pathogenesis of CRC by the differential ARGs, we used the R package “clusterprofiler” [10] to perform GO pathway and KEGG enrichment analysis on the six differential ARGs obtained from the above analysis. The results were considered statistically significant at p < 0.05. The analysis revealed 582 GO pathways, including 499 biological processes, involving responses to starvation, ketones, nutrient levels, extracellular stimuli, and hepatocyte apoptosis. There were 33 cellular components, including vesicle membranes, membrane rafts, membrane microdomains, autolysosomes, and secondary lysosomes. There were 50 molecular functions, including peptidase regulatory activity, ubiquitin protein ligase binding, ubiquitin-like protein ligase binding, enzyme inhibitor activity, and amide binding (Fig. 7C, D). KEGG pathway enrichment analysis of the six differential ARGs showed that the differential ARGs were enriched in animal autophagy, apoptosis, the p53 signaling pathway, pancreatic cancer-muscle, and chronic myeloid leukemia (Fig. 7E, F).

To obtain the protein-level interaction relationships of autophagy-related genes in macrophages and neutrophils, we conducted a PPI analysis (interaction score ≥ 0.7) on the six differential ARGs based on the String database (http://www/string-db.org/). The PPI analysis identified five biomarkers: CTSD, CDKN1A, MAP1LC3B, CFLAR, and BCL2L1 (Fig. 7G).

Based on all cell types annotated in the single-cell sequencing data, a UMAP plot was used to display the expression distribution of biomarkers in different cells of the disease and treatment samples. The results showed that the biomarkers were highly expressed in both macrophages and neutrophils (Fig. 7H). The Mann–Whitney U test was used to compare the expression differences of biomarkers in key cells between the disease and treatment samples (p < 0.05), and the results indicated that the biomarkers showed significant differences in key cells between the disease group and treatment group (Fig. 7I).

Heterogeneity analysis

In all samples of the single-cell sequencing data, the key cells, macrophages, were subjected to secondary dimension reduction and clustering using the “FindNeighbors” and “FindClusters” functions in the Seurat package, based on marker genes. The cells were annotated into different sub-clusters and displayed using a UMAP plot. The results showed that macrophages could be further divided into three cell subtypes: M1 macrophages, M2 macrophages, and Pro macrophages (Fig. 8A, B).

Fig. 8.

Heterogeneity and cell communication analysis of key cells. (A) UMAP plot of macrophage re-clustering. (B) UMAP plot of macrophage sub-cluster identification. (C) GSVA analysis of different macrophage subtypes. (D) UMAP distribution of biomarkers across cell types and expression differences in macrophage subtypes. (E) Proportion and number distribution of cell types. (F) Cell communication among key cell cluster subtypes in the disease group. (G) Cell communication among key cell cluster subtypes in the treatment group.

To explore the differences in biological functions among different macrophage subtypes, GSVA was performed on all samples of single-cell sequencing data. The “msigdbr” package [11] was used to load the “HALLMARK” gene set file, and the pathway activity scores were assigned to each sub-cluster cell. Differences in the pathway activity scores between the disease group and treatment group cells were calculated (p < 0.05). The results indicated that only M1 and Pro macrophages had significantly different gene sets. In M1 macrophages, the treatment group had significantly upregulated gene sets, including oxidative phosphorylation, coagulation, and epithelial-mesenchymal transition, compared to the disease group. The treatment group had significantly downregulated gene sets compared to the disease group, including interferon alpha response, interferon gamma response, inflammatory response, TNFA signaling via NFKB, allograft rejection, IL6 JAK STAT3 signaling, and mitotic spindle. In Pro macrophages, the treatment group had significantly upregulated gene sets, including complement and coagulation, compared to the disease group (Fig. 8C).

To further clarify the expression distribution of biomarkers in macrophage subtypes, a UMAP plot was used to display the expression distribution of biomarkers in different macrophage subtypes in disease and treatment samples based on the cell types annotated in the single-cell sequencing data for macrophage subtypes. The results showed that biomarkers were highly expressed in all macrophage subtypes. The Mann–Whitney U test was used to compare the expression differences of biomarkers in key cells between the disease and treatment samples (p < 0.05). The results indicated that the biomarkers BCL2L1, MAP1LC3B, and CFLAR differed significantly in macrophage subtypes M1 and M2 between the disease and treatment groups. The biomarker Cdkn1a showed significant differences in macrophage subtypes M2 and Pro between the disease and treatment groups. Ctsd showed significant differences in all macrophage subtypes between the disease and treatment groups (Fig. 8D).

Cell communication analysis

The distribution of the number of different cell types between the macrophage subtypes and other major cell types in the disease and treatment groups showed that the proportions of almost all cell types differed between the disease and treatment groups (Fig. 8E). These changes suggested that CRC treatment altered the interactions between various cell types. Therefore, a cell–cell communication analysis was performed between the macrophage subtypes M1 macrophages, M2 macrophages, Pro macrophages, and other major cell populations using the R package CellChat [12]. The aggregated cell–cell communication network was calculated, and the signals from each cell cluster were visualized. The results indicated that the interaction between M2 and Pro macrophages was enhanced in the treatment group compared to that in the disease group (Fig. 8F, G).

Expression and prognosis of autophagy-related markers in CRC

Analysis of the HPA database and IHC results of CRC clinical samples revealed that five autophagy-related markers (CTSD, CDKN1A, MAP1LC3B, CFLAR, and BCL2L1) were expressed in CRC (Fig. 9A). The GEPIA2 database showed that patients with low CDKN1A expression had poor OS, whereas patients with high MAP1LC3B expression had poor OS (P = 0.043, P = 0.025) (Fig. 9B).

Fig. 9.

Expression and prognosis of autophagy-related markers in colorectal cancer. (A) Expression of autophagy-related markers in the HPA database and CRC clinical samples. (B) Prognosis of autophagy-related markers in colorectal cancer in the GEPIA2 database.

Discussion

The macrophage immune checkpoint inhibitor Hu5F9-G4 promoted the polarization of M2d macrophages to M1 and inhibited the malignant evolution of CRC, its mechanism may be related to the autophagy pathway mediated by the autophagy-related markers CDKN1A and MAP1LC3B.

The tumor immune microenvironment, in which macrophages are considered the key components of immune cells, plays a crucial role in the occurrence and development of CRC. Their number and phenotypic transformations are closely related to the occurrence, development, and clinical treatment of CRC [13]. In CRC, TME often promotes the transformation of macrophages into an immunosuppressive phenotype, thereby inhibiting the immune response of T cells [14]. Therefore, finding a way to reverse the immunosuppressive effect of macrophages in the tumor microenvironment to activate the immune response of T cells and improve the response of patients with CRC to immune checkpoint inhibitors is of great significance. In recent years, activated macrophage immune checkpoint blockade therapy has been successful in the treatment of some recurrent or refractory tumors [[15], [16], [17]].

Macrophages are the main immune cells in the tumor immune microenvironment. Because of their phenotypic and functional plasticity, macrophages can be divided into classically activated M1 and alternatively activated M2 types [18]. M1 type macrophages kill tumor cells, inhibit tumor progression, and promote immunity, while M2 type macrophages promote tumor occurrence, development, and immunosuppression [19]. In an in-depth study, M2 type macrophages were further subdivided into M2a, M2b, M2c, M2d, and other subtypes [20]. In the TME, tumor-associated macrophages usually exhibit a tumor-promoting M2-like phenotype, like M2d macrophages, which plays a role in promoting tumor growth and metastasis. Studies suggest that tumor cells use macrophage plasticity to exert different effects at different stages of tumor progression [20,21]. Therefore, studying the functions and regulatory mechanisms of macrophages will aid in the development of new strategies for tumor immunotherapy. We found that Hu5F9-G4 can polarize M2d macrophages in CRC to the antitumor state of the M1 type, indicating that the macrophage immune checkpoint inhibitor Hu5F9-G4 can be used for antitumor treatment by polarizing macrophages.

Autophagy is a cellular self-digestion process that plays an important role in maintaining cellular homeostasis and responding to stressful conditions [22]. Autophagy promotes CRC progression, providing new markers and targets for CRC treatment [23]. Furthermore, autophagy plays an important role in the regulation of macrophage polarization, with the NF-κB pathway, AMPK/mTOR pathway, NLRP3 inflammasome, and miRNAs directly or indirectly participating in the regulation of autophagy during macrophage polarization [[24], [25], [26]]. In this study, the increased autophagy induced by Hu5F9-G4 in M2d macrophages showed that macrophage polarization and autophagy were not isolated during the development of CRC. In summary, combining the results of this study, Hu5F9-G4 may enhance the antitumor immune response in CRC by regulating macrophage autophagy and promoting the polarization of M2d macrophages to M1.

By treating CRC-bearing mice with Hu5F9-G4 and performing a single-cell transcriptome analysis of their tumor tissues, we identified six cell types: epithelial cells, macrophages, fibroblasts, neutrophils, NK cells, and endothelial cells. Further analysis of the mouse ARGs revealed significant differences between macrophages and neutrophils. Macrophages, as major components of the TME, interact with tumor cells through the secretion of cytokines and participate in the progression of CRC [27]. Neutrophils play a dual role in the occurrence and development of CRC; they may promote tumor growth by secreting inflammatory mediators and enzymes and can directly kill tumor cells by releasing cytotoxic substances [28,29]. The infiltration of neutrophils in CRC is associated with high invasiveness and metastasis of tumors and affects tumor progression by releasing tumor-promoting growth factors and participating in immune suppression [30]. The mechanisms of these cells in CRC are intertwined and jointly affect disease occurrence and progression. Understanding these mechanisms is important for the development of effective treatment strategies.

In this study, DEGs in macrophages and neutrophils were intersected with ARGs and subjected to PPI analysis, bioinformatics analysis, and clinical sample validation. Two CRC biomarkers related to macrophage and neutrophil autophagy were identified: CDKN1A and MAP1LC3B. The Cyclin-Dependent Kinase Inhibitor 1A (CDKN1A) gene encodes p21, an important cell-cycle inhibitor. CDKN1A is involved in cell-cycle regulation, DNA repair, and apoptosis and inhibits the proliferation of CRC cells by inhibiting cyclin-dependent kinases and preventing the transition from G1 to S phase [31]. The downregulation of p21 expression in CRC cells is associated with excessive proliferation and enhanced metastasis. Restoring the function of p21 can effectively slow the proliferation of CRC cells. p21 is a downstream effector of the p53 signaling pathway, and p53 regulates the cell cycle and apoptosis by inducing the expression of p21, indicating that CDKN1A may regulate cell autophagy by affecting autophagy-related cell death pathways, such as the cell cycle and DNA damage response [32]. The MAP1LC3B gene encodes the LC3B protein, which plays a key role in cellular autophagy and is involved in the formation of autophagosomes and regulation of the autophagy process. The expression level of LC3B is usually upregulated in CRC tissues, which may be related to the increased dependence of cancer cells on autophagy [33]. LC3B is also involved in the regulation of the sensitivity of cells to drugs. Furthermore, inhibiting the function of LC3B can enhance the sensitivity of CRC cells to chemotherapeutic drugs, thereby improving their therapeutic effects [34]. The MAP1LC3B gene and its encoded LC3B protein play complex roles in the occurrence, development, and treatment response of CRC; however, their specific mechanisms require further research. Future research and treatment strategies may utilize CDKN1A and MAP1LC3B as targets for improving the therapeutic effects of CRC.

To clarify the heterogeneity of the key cell cluster macrophages, they were divided into three subtypes: M1, M2, and Pro. To explore the functional and pathway differences of different macrophage subtypes between the disease and treatment groups, we performed GSVA analysis on the data. In M1 macrophages, the treatment group had significantly upregulated gene sets, including oxidative phosphorylation, coagulation, and epithelial-mesenchymal transition, compared to the disease group. Oxidative phosphorylation is a key process in the cellular respiratory chain and occurs primarily in the inner mitochondrial membrane. The role of the oxidative phosphorylation pathway in CRC is complex and diverse, reflected in changes in energy metabolism, increased oxidative stress, regulation of signaling pathways, and the potential for targeted therapy [35]. Cell interaction analysis revealed that, compared to the disease group, the treatment group enhanced the interaction between M2 and Pro macrophages. Autophagy-related markers were differentially expressed in macrophages and their subtypes between the disease and treatment groups, providing new insights into the potential molecular mechanisms of related genes.

Combined with the results of this study, Magrolimab (Hu5F9-G4) treatment was associated with increased autophagy levels in macrophages, accompanied by a shift from the M2d to M1 phenotype and inhibition of colorectal cancer cell growth. These findings suggest that macrophage immune checkpoint blockade may represent a potential immunotherapeutic strategy for colorectal cancer. The autophagy-related markers CDKN1A and MAP1LC3B identified in our analysis may participate in this process, although the present data demonstrate correlation rather than direct causation. Further mechanistic studies are warranted to clarify whether autophagy activation drives macrophage reprogramming or results from it, and to explore the role of CD47–SIRPα signaling in this context. In addition, the single-cell transcriptomic analysis revealed significant changes in autophagy-related pathways in neutrophils, indicating that innate immune remodeling under Magrolimab treatment may involve multiple cell types. Future work will integrate functional assays and pathway perturbation experiments to validate these hypotheses and expand our understanding of macrophage–neutrophil interactions in colorectal cancer immunotherapy.

Ethics statement

This study was approved by the Ethics Committee of the First Affiliated Hospital of Kunming Medical University [Ethics Approval Number: (2023) Ethics L No 150], and a waiver of patient informed consent was granted. The animal experiment was reviewed and approved by the Animal Ethics Committee of Kunming Medical University [Ethics Approval Number: (2023) Ethics L No 150].

Funding

This work was supported by National Natural Science Foundation of China (No. 82560569), Key Laboratory of Cell Therapy Technology Transformation Medicine of Yunnan Province (No. 2015DG034), Basic Research Program of Yunnan Province (No. 202501AY070001-015), Yunnan Province Foreign Experts Program (No. 202505AP120005).Yunnan Revitalization Talent Support Program (No. RLMY20220013).

CRediT authorship contribution statement

Wan Rong Lin: Writing – review & editing, Writing – original draft, Software, Methodology, Formal analysis, Data curation. Wei Qing Liu: Writing – review & editing, Funding acquisition, Data curation. Tawfik Ali Hamood Alburiahi: Methodology, Formal analysis. Ruo Bing Chen: Formal analysis. Hao Peng Lu: Formal analysis. Jian Dong: Writing – review & editing, Funding acquisition. Jun Yang: Writing – review & editing, Writing – original draft, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The author is an Editorial Board Member/Editor-in-Chief/Associate Editor/Guest Editor for this journal and was not involved in the editorial review or the decision to publish this article.