Abstract Background Tumor-associated macrophages (TAMs) contribute significantly to immunosuppression in colorectal cancer liver metastasis (CRLM), leading to high aggressiveness and poor prognosis. However, the key molecules involved in shaping TAMs toward the pro-tumoral phenotype in CRLM remain unclear, limiting the development of macrophage-mediated immunotherapies for CRLM. Results In this study, we showed that DICER1 was highly expressed in TAMs and closely associated with M2 polarization in CRLM. Knockdown of Dicer, encoded by DICER1 in humans (or Dicer1 in mice), skewed macrophages toward an anti-tumoral M1 phenotype, with increased expression of pro-inflammatory cytokines and tumor cell phagocytosis, thereby suppressing tumor growth in mice. An M2 macrophage-targeting nanosystem was developed to deliver Dicer1 siRNA for selectively downregulating Dicer expression in M2 macrophages. In situ manipulation of TAMs with the nanoparticle exerted a significant anti-tumor effect with an improved immune microenvironment in a CRLM mouse model. Macrophage depletion experiments further suggested that this effect was largely dependent on the presence of TAMs. Mechanistically, Dicer inhibition reprogrammed M2-like macrophages through downregulation of miR-148a-3p and miR-1981-5p. Conclusion Our study uncovered the central role of Dicer in the M2 polarization of TAMs, in turn suggesting a promising therapeutic strategy for CRLM. Supplementary Information The online version contains supplementary material available at 10.1186/s12951-025-03518-4. Keywords: Colorectal cancer, Liver metastasis, Tumor-associated macrophages, Dicer, RNA-lipid nanoparticles Graphical abstract [44]graphic file with name 12951_2025_3518_Figa_HTML.jpg Supplementary Information The online version contains supplementary material available at 10.1186/s12951-025-03518-4. Introduction Metastasis is a major cause of mortality in colorectal cancer (CRC), with the liver being the most prevalent site of distant metastasis. Surgery and chemotherapy are standard treatments for patients with CRC liver metastasis (CRLM). However, curative resection is only applicable in 10–20% of cases, and various chemotherapeutic regimens have achieved poor therapeutic efficacy [[45]1, [46]2]. Immunotherapies such as programmed cell death-1 (PD-1) and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) antibodies have revolutionized cancer treatment in multiple cancers; however, they provide modest benefits to patients with CRLM [[47]3]. Therefore, identification of extended therapeutic targets in metastatic CRCs, especially in CRLM, is of utmost importance to develop effective immune-based therapeutic strategies and benefit a larger patient group. The suppressive tumor microenvironment (TME) within liver metastases plays an important role in immunotherapy resistance. Tumor-associated macrophages (TAMs) constitute the most abundant group of immune cells in the TME. TAMs have been shown to be closely associated with tumor progression and therapeutic efficacy [[48]4, [49]5]. They exhibit remarkable plasticity and can differentiate into various phenotypes in response to external environmental signals, with M1 and M2 phenotypes representing the two extremes [[50]6]. M1 macrophages are classically activated macrophages that exhibit pro-inflammatory anti-tumor effects, whereas alternatively, activated M2 macrophages produce anti-inflammatory mediators for immunosuppression, ultimately contributing to tumor progression. As the vast majority of TAMs manifest an immunosuppressive M2 phenotype and promote tumor growth and resistance to therapies, reprogramming TAMs from a pro-tumoral to an anti-tumoral phenotype has emerged as a promising strategy for immunotherapy [[51]7–[52]10]. In contrast to macrophage depletion, this strategy leverages the inherent immunostimulatory function of macrophages by utilizing their roles as key phagocytes and professional antigen-presenting cells to enhance the immune response. However, the key molecules involved in macrophage polarization in CRLM remain underexplored, limiting the development of macrophage-mediated immunotherapies for CRLM. Dicer, encoded by DICER1 in humans or Dicer1 in mice, is a ribonuclease that regulates the generation of mature microRNAs (miRNAs) from hairpin-shaped precursor miRNAs. Genetic deletion of Dicer1 in TAMs using the LysM-Cre system results in dysregulated miRNA expression profiles and skewing of immunosuppressive TAMs toward an immune-activating phenotype for tumor inhibition in multiple subcutaneous tumor models in mice [[53]11, [54]12]. Moreover, Dicer contributes to atheroprotection by facilitating alternative macrophage polarization through metabolic reprogramming by generating various miRNAs including miR-10a, let-7b, and miR-195a [[55]13]. These studies highlight the potentially critical role of Dicer in macrophage polarization. However, the mechanism by which Dicer orchestrates the miRNA network as a molecular switch to polarize TAMs at the post-transcriptional level in CRLM remains unknown. The development of an approach for in vivo intervention of the Dicer-miRNA axis to evaluate its therapeutic potential can provide opportunities for cancer immunotherapy. In this study, we show that Dicer is a key molecule that drives M2 polarization of macrophages in CRLM. Dicer promotes TAM M2 polarization for immune suppression and potentiates tumor outgrowth. Targeted inhibition of Dicer expression in TAMs with RNA interference using an M2 macrophage-targeting nanosystem induces TAM repolarization, leading to an anti-tumor effect and an improved immune microenvironment in a CRLM mouse model. Mechanistically, the delivery of Dicer1 siRNA repolarizes macrophages by regulating their miRNA expression profile, including the downregulation of miR-148a-3p and miR-1981-5p. Collectively, our findings identify Dicer and its downstream miRNA network as potential targets for immunotherapy against CRLM. Materials and methods Materials DLin-MC3-DMA, 1,2-distearoyl-3-phosphatidylcholine (DSPC) and DMG-PEG (MW 2000) were provided by AVT Pharmaceutical Co., Ltd. (Shanghai, China). Cholesterol was obtained from Sigma-Aldrich (USA). DSPE-PEG-M2pep (YEQDPWGVKWWY) was synthesized by RuixiBio (Xi’an, China). DiR, DiI, and DiD dyes were obtained from Meilun Bio (Dalian, China). D-Luciferin was purchased from MCE (Shanghai, China). Cell lines CT26, RAW264.7, and L929 murine cells were sourced from the American Type Culture Collection. CT26-luc and CT26-GFP cells were generated by infecting wild-type CT26 cells with a recombinant lentivirus containing overexpression plasmids encoding mouse luciferase and GFP, respectively. Dicer-knockdown RAW264.7 (RAW264.7^Dicer − KD) cells and Dicer-overexpression (OE) cells (RAW264.7^Dicer − OE) were generated by transfection with recombinant lentivirus containing the Lenti CRISPR V2 vector carrying sgRNA targeting mouse Dicer1 and Dicer-OE plasmids, respectively. The sgRNA sequences used in this study are listed in Table [56]S1. The coding sequences of Dicer1 ([57]NM_001411829.1) were amplified by reverse transcription-polymerase chain reaction (PCR), ligated into the GV513 vector and then subcloned into the BstXI/NheI enzymatic sites of the pCMV expression vector to generate Dicer overexpression plasmids. After transfection, the cells were cultured in an appropriate concentration of puromycin for 7 days and then analyzed by western blotting to verify the knockdown or overexpression efficiency. The lentiviruses and plasmids used in these experiments were purchased from GenePharma (Shanghai, China). All cells were cultured in Dulbecco’s modified Eagle medium (DMEM) (Gibco, Thermo Fisher, USA) containing 10% fetal bovine serum (Gibco, Thermo Fisher, USA) and 1% penicillin-streptomycin (P/S) (Beyotime, China) at 37 °C under 5% CO[2]. Cell authentication was performed before experimentation using short tandem repeat fingerprint profiling. Animal use and care All animal experiments were performed in accordance with the ethical guidelines approved by The Second Affiliated Hospital of Zhejiang University School of Medicine (No. 2024155). Mice were housed in ventilated cages under specific pathogen-free conditions with controlled 12-hour light/dark cycles, stable temperature and humidity, access to enriched water, and ad libitum feeding. Preparation of bone marrow-derived macrophages (BMDMs) BMDMs were obtained from BALB/c mice as previously described [[58]14]. Bone marrow cells were flushed from the femurs and tibiae using DMEM containing 1× P/S. The cell suspension was filtered through a 70-µm cell strainer (Biosharp, China). Red blood cells were eliminated using a red cell lysis buffer (Beyotime, China). The remaining cells were cultured in DMEM containing 10% FBS and 30% L929-conditioned medium as a source of macrophage colony-stimulating factor. The cultures were incubated for six days and the medium was replaced every three days. The BMDMs were verified by flow cytometry (FCM) using anti-CD11b and anti-F4/80 antibodies and used for further experiments. Macrophages of the M1- or M2-like phenotypes were generated by stimulation with 100 ng/mL lipopolysaccharide (LPS) (PeproTech, USA) or 10 ng/mL IL-4 (PeproTech, USA), respectively, for 24 h. Screening of M2 polarization-associated genes in TAMs In patients with metastatic pan-cancer (MET500), the abundance of M2 macrophages in various tissues was assessed using the CIBERSORT algorithm. To standardize the M2 polarization characteristics of these macrophages, the overall numbers of monocytes/macrophages was used as a correction factor. The microenvironment cell populations-counter function was used for immune cell infiltration analysis. The ratio of M2 macrophages to the total monocyte/macrophage lineage was used as an indicator of M2 polarization levels. Single-cell (sc) sequencing data from patients with CRLM ([59]GSE164522) were then analyzed. The Seurat R package was used to merge all cells, and the MAESTRO R package annotated individual cells using the GRCh38 human reference genome, extracting a total of 24,788 monocytes/macrophages labeled as “Mono/Macro.” The reference gene set for “Mono/Macro” was sourced from the human immune system. CIBERSORT parameters in MAESTRO. From CRC liver metastases, 5,818 monocytes/macrophages were extracted, and the FindClusters function with a resolution of 0.1 was applied to classify all monocytes/macrophages. The M2 gene set was used to evaluate the M2 polarization level of the macrophages using the AddModuleScore method [[60]15]. Related genes correlated with M2 polarization in patients with metastatic pan-cancer (MET500) or CRLM ([61]GSE164522) were then determined. Finally, RNA-seq datasets from six cohorts ([62]GSE39582, [63]GSE14333, [64]GSE41258, [65]GSE72970, TCGA-COAD, and TCGA-READ) were integrated using meta-analysis to determine whether these genes were associated with the prognosis of patients with CRC. The RMA function from the metafor R package was used to pool Cox-regression results and the forestplot R package was used to show meta-analysis results. Human specimens Tissue samples were collected from patients with CRLM at The Second Affiliated Hospital of Zhejiang University School of Medicine. This study was conducted in accordance with the ethical guidelines and was approved by the ethics committee of the hospital (Approval No. 20240985). Establishment of CRLM mouse models A CRLM model was established using six–eight-week-old male BALB/c mice, as previously described [[66]16]. The mice were anesthetized by 2.5% isoflurane inhalation and placed in a supine position on a sterile surgical platform. A small vertical incision was made below the rib cage on the left side to expose the spleen. The spleen was carefully divided into two sections, each with an intact vascular pedicle, by ligation with surgical sutures. A suspension of 5 × 10^5 CT26 cells was injected into the distal section of the spleen, using a fine-gauge needle. The distal half of the spleen was resected after a 5-minute incubation period. The proximal portion of the spleen was repositioned into the abdominal cavity, and the abdominal wall and skin were closed. Preparation and characterization of lipid nanoparticles (LNPs) Conventional LNPs were prepared using a microfluidic formulation as previously described [[67]17]. M2pep-LNPs were manufactured with some modifications. Briefly, the lipids DLin-MC3-DMA, DSPC, cholesterol, DMG-PEG, and DSPE-PEG-M2pep were dissolved in ethanol at a molar ratio of 50:10:38:1:1 to achieve an ionizable lipid concentration of 2 mg/mL. siRNA was dissolved in citrate buffer (pH 4.0, 10 mM) to a final concentration of 20 µg/mL. The lipid and siRNA solutions were rapidly mixed using a microfluidic device at a 1 mL/min flow rate and a 6:1 volume ratio of aqueous phase to ethanol. The resulting nanoparticles were dialyzed against 1× phosphate-buffered saline (PBS) using the Float-A-Lyzer G2 Dialysis Device (3.5 kDa) (Spectrum Laboratories) for 6 h at room temperature (RT). The size distribution and zeta potential of the LNPs were determined by dynamic light scattering (DLS) using a Zetasizer Nano ZS particle sizer (3600; Malvern Instruments Ltd., U.K.). Transmission electron microscopy (TEM, JEOL JEM-1010) was employed after staining the LNPs with uranyl acetate to examine their morphology. The encapsulation efficiency and release kinetics of siRNA within the LNPs was assessed using the Quant-iT RiboGreen RNA Assay ([68]R11490, Thermo Fisher Scientific) as previously described [[69]18]. In vitro transfection BMDMs were seeded in 6-well plates at a density of 2 × 10^5 cells per well and transfected with nanoparticles or Lipofectamine RNAiMAX encapsulating siRNA at a final concentration of 100 nM for 24 h. Subsequently, the cells were incubated for another 24 h, and the protein and mRNA expression levels were assessed. The siRNA oligonucleotides were synthesized by GenePharma (Shanghai, China) (Table [70]S1). For the transfection of miRNA mimics and inhibitors, Lipofectamine RNAiMAX was used at final concentrations of 50 nM for mimics and 100 nM for inhibitors. RiboBio (Guangzhou, China) synthesized the miRNA mimics and inhibitors. Quantitative reverse transcription PCR (qRT-PCR) Total RNA was extracted using TRIzol reagent (Beyotime, China) according to the manufacturer’s protocol. For cDNA synthesis, HiScript II Q RT SuperMix for qPCR (+ gDNA wiper) (Vazyme, China) was used for mRNA. In contrast, the miRNA 1st Strand cDNA Synthesis Kit (by stem-loop) (Vazyme, China) was used for miRNA, according to the manufacturer’s instructions. Quantitative PCR (qPCR) was performed on a LightCycler 480 II system (Roche). ChamQ Universal SYBR qPCR Master Mix (Vazyme) was used for mRNA quantification, whereas miRNA Universal SYBR qPCR Master Mix (Vazyme) was used for miRNA analysis. β-actin and U6 were used as endogenous mRNA and miRNA quantification controls, respectively. The primer sequences used are listed in Table [71]S2-[72]S3. Tumor-suppression assay For the phagocytosis assay, IL-4 pre-treated wild type RAW264.7 (RAW264.7^WT) or RAW264.7^Dicer − KD cells were mixed with CT26-GFP cells and co-cultured in a 6-well plate at a ratio of 9:1 (RAW264.7 to CT26-GFP). After 48 h of co-culture, the cells were detached using trypsin-EDTA solution (Biosharp, China) for subsequent FCM analysis. For the indirect tumor-suppression assay, IL-4 pre-treated RAW264.7^WT or RAW264.7^Dicer − KD cells were seeded at a density of 1 × 10^4 cells per well in the upper chamber of a 24-well Transwell plate (Corning, China). CT26 cells were seeded in the lower chamber at the same density. CT26 cells were collected after 48 h of co-culture and cell proliferation assays (CCK-8) were performed at 24, 48, and 72 h. For in vivo assay, IL-4 pre-treated RAW264.7^WT or RAW264.7^Dicer − KD cells, as well as LPS pre-treated RAW264.7^WT, were mixed with CT26 cells (5 × 10^5 cells per cell type) in a total volume of 100 µL. The mixture was then injected into the distal section of the spleen of BALB/c mice to establish the CRLM model as described above. Three days post-injection, the tumor volume was monitored every 4 days. The average radiant efficiency was quantified using Living Image 4.5.4 Software (PerkinElmer). Cellular uptake siRNA-LNPs and transfection efficacy evaluation BMDMs or RAW264.7 cells were seeded in 24-well plates and cultured overnight. DiI-labeled or FAM-siRNA-loaded nanoparticles were added to the cells and incubated for 1 and 6 h, respectively. The supernatants were then removed and cell nuclei were stained with Hoechst 33,342 (Beyotime, China). The fluorescence intensities of DiI and FAM in the cells were examined using fluorescence microscopy (Leica, Germany) and flow cytometry (FCM; Beckman, USA). For ex vivo analysis, TAMs and spleen macrophages were isolated from liver metastases and spleens of CRLM model mice. The isolated cells were incubated with DiD-labeled nanoparticles at RT for 2 h. After incubation, cells were washed thrice with PBS and labeled with fluorochrome-conjugated antibodies for FCM at the recommended dilutions for 30 min at 4 °C. After staining, the cells were washed twice by centrifugation, resuspended in ice-cold PBS, and analyzed using FCM. In vivo biodistribution of nanoparticles DiR-labelled or Cy3-siRNA-loaded nanoparticles were prepared as previously described. The nanoparticles were administered intravenously to CRLM mice at a dose of 0.75 mg/kg. To monitor their in vivo distribution, retention of DiR-labeled nanoparticles in the liver was monitored using an IVIS imaging system (PerkinElmer) at the specified time points. Imaging was performed at excitation and emission wavelengths of 748 nm and 780 nm, respectively. At the 8-h time point, a subset of mice was euthanized, and major organs were collected for further imaging analysis. The average radiant efficiency was quantified using Living Image 4.5.4 Software (PerkinElmer). For cellular distribution analysis, mice were sacrificed 6 h after injecting Cy3-siRNA-loaded nanoparticles. The fluorescence of Cy3 across different cell types in liver metastases was detected using FCM (Beckman, USA) and immunofluorescence (IF) staining. Western blotting Collected cells or tissues were lysed using RIPA buffer (Beyotime) supplemented with protease inhibitors. The lysates were centrifuged at 12,000 × g (4 °C, 5 min), and the supernatants were carefully collected. The protein concentration was determined using a bovine serum albumin (BCA) protein assay kit (Beyotime, China). Equal amounts of protein were separated by SDS-PAGE and transferred to PVDF membranes. Membranes were blocked with 5% BSA in TBST buffer for 0.5–1 h at RT and subsequently immunoblotted overnight at 4 °C with the following primary antibodies: anti-Dicer (ab259327, 1:1000 dilution, Abcam), anti-IL-10 (HA722032, 1:1000 dilution, HUABIO), anti-IL-1β (12242, 1:1000 dilution, Cell Signaling Technology), and anti-β-Actin (ET1702-52, 1:2000 dilution, HUABIO). The membranes were then incubated with horseradish peroxidase (HRP)-conjugated secondary antibody (1:5000, Beyotime, China) for 1 h at RT. Protein bands were visualized using an enhanced chemiluminescence (ECL) kit (Fdbio Science, China) and imaged using a gel documentation system. All antibodies used in this study are listed in Table [73]S4. In vivo anti-tumor studies To evaluate the in vivo anti-tumor efficacy of siRNA-loaded nanoparticles, CRLM mice were randomized into different treatment groups. Each group received intravenous injections of LNPs or M2pep-LNPs encapsulating siDicer1 or siNC at a dose of 0.75 mg/kg. Tumor growth was monitored using the IVIS at predetermined intervals. On the 12th day post-treatment, all mice were euthanized and body and liver weights were recorded. The major organs, including the heart, liver, spleen, lungs, and kidneys, were harvested and fixed in 4% paraformaldehyde solution for pathological staining. Whole blood samples were collected for biochemical analysis. For survival studies, mice were monitored daily, and the endpoint was reached if any of the following conditions were met: weight gain or loss exceeding 10% within 1 week or clinical signs of distress such as dehydration, inactivity, or severe lethargy. In macrophage depletion experiments, mice bearing CRLM were assigned to different groups and intraperitoneally injected with clodronate liposomes (1.4 mg/20 g body weight) (40335ES10, YEASEN) or an equivalent volume of PBS-liposomes twice per week. Concurrently, nanoparticles were administered as scheduled. Tumor growth and survival were monitored as previously described. Flow cytometry (FCM) and immunofluorescence (IF) analyses were performed to assess macrophage depletion efficacy and changes of immune cells in liver metastases. FCM assay Cells were isolated from liver metastases using gradient centrifugation. Tissues were finely minced and digested with collagenase IV (0.5 mg/mL) for 30 min at 37 °C under slow rotation. Leukocyte populations were enriched using 40% and 70% Percoll gradients (Biosharp, China). The cells were washed and resuspended in the staining buffer. Cells were blocked with anti-mouse CD16/CD32 antibodies (Biolegend, USA) for 30 min at 4 °C to block non-specific binding. Subsequently, fluorochrome-conjugated antibodies (Biolegend, USA) were added at the manufacturer-recommended dilutions, followed by incubation for another 30 min at 4 °C. For intracellular staining, cells were permeabilized using the FOXP3 Fix/Perm Buffer Set (Biolegend, USA) and stained for 60 min at RT. The stained cells were analyzed by FCM (Beckman, USA). All antibodies used in this study are listed in Table [74]S4. The gating strategy used for immune cells was as follows. M1 TAMs: CD11b^+ Gr-1^− F4/80^+ CD86^+. M2 TAMs: CD11b^+ Gr-1^− F4/80^+ CD206^+. CD8^+ T cells: CD3^+ CD8^+. CD4^+ T cells: CD3^+ CD4^+. Th1 cells: CD3^+ CD4^+ T-bet^+. Th2 cells: CD3^+ CD4^+ GATA3^+. Treg cells: CD3^+ CD4^+ CD25^+ Foxp3^+. MDSCs: CD11b^+ Gr-1^+. DCs: CD11b^+ CD11c^+. IF, IHC, and H&E staining Tissue staining was performed as previously described [[75]19]. For IF staining, paraffin-embedded liver sections were dewaxed, subjected to antigen repair, and blocked with 3% BSA solution for 1 h at RT. Sections were then incubated overnight at 4 °C with primary antibodies, washed thoroughly, and treated with fluorochrome-conjugated secondary antibodies (Invitrogen Co.) in the dark. For Cy3^+ cell detection in liver samples frozen in O.C.T. compound (Sakura Finetek, USA), tissues were sliced to a thickness of 8 μm. These sections were permeabilized with 0.1% Triton X-100 (Beyotime) in PBS for 30 min, followed by blocking with 3% BSA for 1 h at RT. Alexa Fluor^® 647-conjugated F4/80 antibody (BM8, Biolegend, USA) was applied for 2 h at RT, and the sections were mounted using Prolong™ Gold Antifade Mountant with DAPI (Invitrogen). Images were captured using a fluorescence microscope (Leica, Germany) and analyzed using ImageJ software. As for IHC and H&E staining, paraffin-embedded liver tissues were sectioned to 5 μm thickness, deparaffinized, and rehydrated. After blocking with 3% BSA, the sections were incubated overnight at 4 °C with the primary antibodies and for 1 h at RT with the secondary antibodies. DAB substrate liquid (S21024-2, Thermo Fisher Scientific) was used for protein staining, followed by counterstaining with hematoxylin. Sections were observed and photographed under a microscope. Protein expression was analyzed using ImageJ software and scored using the H-score method (strong expression × 3 + moderate expression × 2 + weak expression × 1). The antibodies used for immunofluorescence staining are listed in Table [76]S4. Blood chemistry analysis Serum was collected from the mice after treatment. The levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), blood urea nitrogen (BUN), and creatinine (Cre) were measured using the Vitros^® 5600 analyzer (QuidelOrtho, USA). miRNA-Seq analysis and targets prediction IL-4 pre-treated BMDMs were transfected with siNC@M2pep-LNPs or siDicer1@M2pep-LNPs for 48 h to investigate the miRNA expression profile. Total RNA was extracted from the cells using TRIzol reagent following the manufacturer’s protocol. The RNA samples were sent to Novogene (Tianjin, China) for miRNA sequencing and analysis. The generated data were processed to determine the expression levels of miRNAs, which were quantified as Fragments Per Kilobase of transcripts per million mapped reads. The potential targets of miRNA were obtained through examination of the overlapped intersection from two databases (miRmap and microT). To further investigate the signaling pathways that miRNA was involved in, we performed pathway analysis using KEGG pathway enrichment analysis. Statistical analysis All statistical analyses were performed using GraphPad Prism v7.0 software. Data are presented as mean ± standard deviation (SD). Student’s t-test was used for comparisons between two groups. For multiple comparisons, either a multiple t-test with Bonferroni correction or one-way analysis of variance (ANOVA) with Tukey’s post hoc test was used depending on the data distribution. Survival was analyzed using the log-rank test. Statistical significance was set at P < 0.05, with the following significance levels: ^*P < 0.05, ^**P < 0.01, ^***P < 0.001. Results Dicer is selectively upregulated in M2-like TAMs in the liver metastases of cancers Several studies have demonstrated abnormal infiltration of TAMs in cancers, which is closely related to tumor progression and therapeutic efficacy [[77]20, [78]21]. A bulk transcriptome sequencing dataset (MET500) from pan-cancer patients was analyzed using CIBERSORT and MCPcount algorithms to investigate the phenotypic polarization of macrophages in cancer liver metastasis. The results showed that M2 polarization levels of macrophages in liver metastases were significantly higher than those in non-liver metastases (P < 0.001) (Fig. [79]1A). A scRNA dataset ([80]GSE164522) from patients with CRLM was thus analyzed to screen the key regulatory molecules related to M2 polarization of macrophages in CRLM. As shown in Figs. [81]1B, 5 and 818 macrophages were analyzed and divided into five clusters (C1-C5) based on different resolutions. All macrophages were scored using gene set variation analysis (GSVA) according to the M2 polarization gene sets. As shown in Fig. [82]1C and Figure [83]S1A (Supporting Information), the GSVA score of the C5 cluster was significantly lower than that of the other four clusters (C1-C4). Accordingly, the macrophages were divided into two groups: high M2 score and low M2 scores. Differentially expressed genes (DEGs) between the two groups were determined, and 516 genes were identified to be positively correlated with M2 scores based on the criteria of min.pct = 0.15, logfc.threshold = 0.15, and P-value < 0.01 (Figure [84]S1B, Supporting Information). Fig. 1. [85]Fig. 1 [86]Open in a new tab DICER1 expression correlates with the M2 polarization of macrophages in CRLM and predicts poor prognosis. A) Comparison of M2 polarization levels in macrophages between liver metastases (LM, n = 179) and non-liver metastases (Non-LM, n = 143) in patients with pan-cancer. B) Annotation and classification of single-cell data from patients with CRLM (n = 5818). The UMAP plot shows five macrophage clusters (C1-C5). C) The distribution of M2 polarization states across different macrophage clusters. D) Venn diagram depicting the intersection of macrophage M2 polarization-related genes between the bulk transcriptome sequencing dataset (MET-500) from patients with pan-cancer and the single-cell sequencing dataset ([87]GSE164522) from patients with CRLM. E) Meta-analysis across six cohorts ([88]GSE39582, [89]GSE14333, [90]GSE41258, [91]GSE72970, TCGA-COAD, TCGA-READ) was employed to explore the correlation of 13 genes with CRC recurrence survival prognosis. F) DICER1 expression across different macrophage clusters in patients with CRLM. G-H) Relationship between DICER1 expression and CRC recurrence survival in two representative CRC cohorts ([92]GSE72970 and [93]GSE14333). CRLM: colorectal cancer liver metastasis; LM: liver metastasis, Non-LM: non-liver metastasis, Mac: macrophage, scRNA: single-cell RNA, UMAP, uniform manifold approximation and projection. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 The above sequencing dataset from pan-cancer liver metastasis samples (MET500) was further analyzed, and 1,355 genes were identified to be related to M2 polarization using the criteria of P-value < 0.01 & cor > 0.15 (Figure [94]S1C, Supporting Information). By intersecting the results from both bulk and scRNA data, we identified 13 genes associated with M2 polarization of macrophages: RNF13, CD9, ELL2, PGD, B3GNT5, ABHD5, GNPDA1, TMEM251, WIPI1, CREG1, TFRC, DICER1, and SLIRP (Fig. [95]1D). After standardizing and adjusting the expression of each gene based on the monocyte/macrophage content, we conducted a survival analysis of these 13 genes in six CRC cohorts (Table [96]S5, Supporting Information). The results were further pooled by meta-analysis to minimize potential heterogeneity across cohorts (Fig. [97]1E). Among these genes, DICER1 showed the strongest association with CRC recurrence (hazard ratio (HR) = 1.497, 95% confidence interval (CI) [1.216–1.843], P < 0.001). Furthermore, scRNA analysis showed that DICER1 was highly expressed in macrophages compared to that in other immune cells in liver metastases (Figure [98]S1D, Supporting Information). In terms of the proportion of DICER1-positive cells, macrophages accounted for the highest percentage (> 30%) among different myeloid cell populations (Figure [99]S1E, Supporting Information). The correlation between DICER1 expression and M2 polarization levels was observed at both the bulk and scRNA levels (Figs. [100]1 F and S1F, Supporting Information). Additionally, two representative cohorts ([101]GSE72970 and [102]GSE14333) showed a correlation between standardized DICER1 expression and CRC recurrence survival curves (Fig. [103]1G and H). DICER1 (also known as Dicer1 in mice) encodes Dicer, a highly conserved ribonuclease that cleaves pre-miRNAs (precursor miRNAs) into mature miRNAs. To further verify Dicer expression in TAMs, liver tissues from patients with CRLM and mice with CRLM were collected and analyzed using IF staining. As shown in Figures [104]S2A and [105]S2D (Supporting Information), TAMs were enriched in liver metastatic lesions, with the vast majority (> 95%) being an M2-like phenotype (CD163^+ cells), as supported by the bulk transcriptome sequencing dataset (MET500). In contrast, more M1-like macrophages (CD86^+ cells) were distributed in adjacent normal tissues. Within metastatic lesions, Dicer expression in M2-like TAMs was significantly higher than that in M1-like TAMs (P < 0.05, in patients; P < 0.01 in mice) (Figures [106]S2B-C and [107]S2E-F, Supporting Information). These results indicate that Dicer expression is closely correlated with the M2 polarization of TAMs in CRLM and predicts poor prognosis. Dicer knockdown promotes macrophage M1 polarization and exerts anti-tumoral effects As shown in Fig. [108]2A, LPS significantly downregulated Dicer expression, whereas IL-4 upregulated Dicer expression in the RAW264.7 macrophage cell line and primary BMDMs, indicating a close relationship between Dicer expression and macrophage phenotypes. Stable knockdown (KD) of Dicer in RAW264.7 (RAW264.7^Dicer − KD) cells was achieved using the CRISPR/Cas9 method. As shown in Figure [109]S3A (Supporting Information), the proliferation ability of RAW264.7 cells was inhibited when Dicer was down-regulated. However, Dicer downregulation did not induce apoptosis, as confirmed by the comparable expression of the apoptosis-associated protein, cleaved caspase-3 (Figure [110]S3B, Supporting Information) and FCM analysis (Figure [111]S3C, Supporting Information). Therefore, we speculated that Dicer influences the cell cycle. Indeed, FCM analysis showed that RAW264.7^Dicer − KD cells exhibited G0/G1 phase arrest compared to the control (WT), characterized by an increased proportion of cells in the G0/G1 phase (Figure [112]S3D-E, Supporting Information). Fig. 3. [113]Fig. 3 [114]Open in a new tab Enhanced targeting of M2-like macrophages by M2pep-modified LNPs in vitro.A) Particle size measurement using DLS and cryo-electron microscopy. Scale bar: 100 nm. B) Schematic representation for assessing the targeting ability of nanoparticles in vitro. Created with BioRender.com. C-D) Uptake of nanoparticles by BMDMs observed via fluorescence microscopy. Fluorescence intensity analysis (C) and representative images (D) are shown. Scale bar: 50 μm. E) Schematic representation for investigating nanoparticle uptake by ex vivo M2-like TAMs and spleen-derived macrophages from CRLM-bearing mice. Created using BioRender.com. F) Representative FCM images (left) and MFI analysis (right) of DiD in M2-like TAMs between two groups. G) Representative FCM images (left) and MFI analysis (right) of DiD in M2-like spleen-derived macrophages between two groups. H-K) Comparison of siRNA delivery efficiency to BMDMs between the two nanoparticle types, assessed using fluorescence microscopy (H-I) and FCM (J-K). H: representative fluorescence images; I: fluorescence intensity analysis of FAM; J: representative FCM plots; K: MFI analysis of FAM. Scale bar: 50 μm. L) mRNA expression levels of Dicer1 in M2-like BMDMs transfected with siDicer1-loaded nanoparticles, detected using qRT-PCR. M) Protein expression levels of Dicer and IL-10 in M2-like BMDMs transfected with siDicer1-loaded nanoparticles, detected using western blotting. Data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001,, as calculated using Student’s t test, multiple t test with Bonferroni post hoc tests, or one-way ANOVA with Tukey’s correction as appropriate. DLS: Dynamic light scattering; BMDMs: Bone marrow-derived macrophages; IntDen: Integrated Density; TAMs: Tumor-associated macrophages; FCM: Flow cytometry; MFI: Mean fluorescence intensity To confirm whether Dicer knockdown induces M2-like macrophage repolarization in vitro, Dicer-KD RAW264.7 cells and BMDM were pre-treated with IL-4 to mimic the M2-like TAM phenotype as described in previous studies [[115]22, [116]23]. Then, we examined the protein and mRNA expression of macrophage-related markers in RAW264.7^Dicer − KD cells. We observed that Dicer knockdown upregulated M1-like markers (IL-1β, TNF-α, CXCL9, and CXCL10) and downregulated M2-like markers (IL-10, TGF-β1, and Fizz1) as well as the novel pro-tumoral marker SPP1 (Fig. [117]2B and C). FCM further confirmed that Dicer knockdown upregulated the M1-like marker CD86 and downregulated the M2-like marker CD206 (Fig. [118]2E). Similar results were obtained in BMDMs transfected with Dicer1 siRNA (BMDM^Dicer−KD) (Fig. [119]2B, D and F). Conversely, Dicer overexpression in macrophages resulted in enhanced expression of IL-10 and CD206, along with reduced expression of CD86 (Figure [120]S3F-G, Supporting Information). These results suggest that Dicer knockdown promotes the repolarization of M2-like macrophages toward the M1-like phenotype. Fig. 2. [121]Fig. 2 [122]Open in a new tab Dicer knockdown skews macrophages toward an anti-tumoral M1-like phenotype. A) Macrophages (RAW264.7 and BMDM) were polarized to M1 and M2 types using LPS and IL-4, respectively. Dicer protein expression levels were assessed using western blotting in different macrophage polarization states. B-D) Macrophages (RAW264.7 and BMDM) were treated with IL-4 for 24 h, followed by Dicer knockdown using CRISPR/Cas9 or Dicer1 siRNA transfection, respectively. The protein (B) and mRNA (C-D) expression levels of macrophage-related cytokines/chemokines were measured by western blotting and qPCR, respectively. E-F) Macrophages (RAW264.7 and BMDM) were treated with IL-4 for 24 h. The macrophage M2 marker CD206 and M1 marker CD86 expression levels were assessed using FCM. MFI values for CD206 and CD86 in macrophages were analyzed. G) Schematic overview of the exploration of the anti-tumoral effects of Dicer-KD macrophages. Created using BioRender.com. H) RAW264.7^WT or RAW264.7^Dicer − KD cells were treated with IL-4 for 24 h and then co-cultured with GFP-CT26 cells for 48 h. Phagocytosis of tumor cells by RAW264.7 cells (GFP^+ F4/80^+ cells among GFP^+ cells) was quantified using FCM. I) Expression levels of CD206 in RAW264.7 cells co-cultured with GFP-CT26 cells were quantified using FCM. J) RAW264.7^WT or RAW264.7^Dicer − KD cells were treated with IL-4 for 24 h and then added to the upper chamber of a Transwell system, with wild-type CT26 cells in the lower chamber. After 48 h of co-culture, the proliferation ability of CT26 cells was assessed using CCK-8 at 2 h (D0), 24 h (D1), 48 h (D2), and 72 h (D3). K-L) CRLM models were generated by splenic co-injection of CT26 tumor cells with RAW264.7 macrophages of defined phenotypes (n = 3 per group). RAW264.7^WT and RAW264.7^Dicer − KD cells were polarized to an M2-like phenotype by IL-4 treatment, whereas M1-like polarization was induced in RAW264.7^WT cells by LPS treatment. Tumor growth was monitored every 4 days Data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, as calculated using Student’s t test, multiple t test with Bonferroni post hoc tests, or one-way ANOVA with Tukey’s correction as appropriate. FCM: flow cytometry; MFI: mean fluorescence intensity; KD: knockdown Given that M1-like macrophages possess potent anti-tumoral effects, such as direct phagocytosis of tumor cells, we further explored whether this phenotypic switch mediated by Dicer knockdown in macrophages could inhibit tumor growth. As shown in Fig. [123]2G, IL4-treated RAW264.7^Dicer − KD cells were directly or indirectly co-cultured with CT26 cells. Compared to the control, RAW264.7^Dicer − KD cells exhibited a 1.6-fold increase in tumor cell phagocytosis (Fig. [124]2H). Moreover, these RAW264.7^Dicer − KD cells expressed lower levels of CD206 (Fig. [125]2I). Furthermore, RAW264.7^Dicer − KD cells also indirectly inhibited CT26 cell proliferation in vitro (Fig. [126]2J). In CRLM models, we found that RAW264.7^Dicer − KD cells significantly suppressed tumor growth in vivo compared with that in the control, similar to the effect of M1-like macrophages (Fig. [127]2K-L). Taken together, these results indicate that the phenotypic switch in macrophages mediated by Dicer knockdown can suppress CRLM. Development of M2pep-modifed LNP for M2 macrophage-targeted delivery of Dicer1 SiRNA The absence of Dicer-specific small-molecule inhibitors and antibodies has hindered the development of therapies targeting Dicer. However, siRNA drugs can directly regulate target genes through base complementarity upon entering the cells. Although LNPs can effectively deliver siRNAs in vivo, their lack of targeting specificity limits their application and target sites. We constructed an RNA delivery system with M2-like macrophage-targeting specificity (siRNA@M2pep-LNP) to address this. siRNA-loaded LNPs and M2pep-LNPs were prepared through a microfluidic formulation, and the final nanoparticles obtained were approximately 170 nm and 180 nm in diameter, respectively, as determined using DLS analysis and TEM (Fig. [128]3A). The zeta potentials were approximately − 3.78 and − 12.73 mV, respectively (Fig. [129]3A and Figure [130]S4A). Both nanoparticle formulations exhibited excellent stability, with minimal changes in particle size, polydispersity index (PDI), and zeta potential over a 3-day period under 4 ℃ storage conditions (Figures [131]S4A, Supporting Information). The siRNA encapsulation efficiency was approximately 95% of that of the two nanoparticles. The release of siRNA was evaluated under physiological (pH 7.4) and acidic (pH 5.0) conditions, the latter simulating the acidic endosomal environment. As can be seen in Figures [132]S4B (Supporting Information), both nanoparticles exhibited enhanced siRNA release under acidic conditions, and their release kinetics did not differ significantly. Collectively, these results indicate the potential of these nanoparticles as effective siRNA delivery vectors in vivo. To verify the targeting ability of M2-like macrophages, DiI-labeled nanoparticles were added to macrophages (RAW264.7 and BMDM) with different polarization states, and the uptake of nanoparticles was assessed using fluorescence microscopy and FCM (Fig. [133]3B). The results showed that M2pep-LNPs were internalized by M2-like macrophages nearly twice as much as LNPs (BMDM: 12,865.0 ± 2,176.0 vs. 6,014.0 ± 716.0, P < 0.001; RAW264.7:17,834.0 ± 604.3 vs. 10,004.0 ± 998.3, P < 0.01) (Figs. [134]3C-D and Figures [135]S4A-D, Supporting Information). Single-cell suspensions were prepared from liver metastases and spleens of CRLM-bearing mice and then co-incubated with DiD-labeled nanoparticles (Fig. [136]3E). FCM revealed that the fluorescence intensity of DiD in M2-like macrophages from liver metastases and spleen was higher in the M2pep-LNP group compared to that in the LNP group (Liver metastases: 6.0 ± 2.3 vs. 3.7 ± 1.1, P < 0.05; spleen: 23.2 ± 7.5 vs. 14.9 ± 3.3, P < 0.05) (Figs. [137]3F-G). To further verify the differential efficiency of siRNA delivery between LNPs and M2pep-LNPs, LPS or IL-4 pre-treated BMDMs were transfected with nanoparticles carrying FAM-labeled siRNAs. As shown in Figs. [138]3H-I, FAM fluorescence intensity of the M2pep-LNP group in M2-like BMDMs was significantly higher than that of the LNP group, increasing nearly fourfold (52.9 ± 13.8 vs. 12.9 ± 2.9, P < 0.01). These findings were supported by the FCM results (Figs. [139]3J-K). Regarding the silencing efficiency of Dicer1 in M2-like macrophages, siDicer1@M2pep-LNP showed significantly higher knockdown efficiency and reprogramming ability compared to that of siDicer1@LNP (P < 0.001) (Figs. [140]3L-M). Collectively, these results suggest the superior targeting capability of M2pep-modified LNPs. The organ and cellular biodistributions of the two nanoparticles were then assessed in vivo (Fig. [141]4A). According to the IVIS results, the average DiR fluorescence intensity in the livers of the M2pep-LNP group remained consistently higher than that in the LNP group at 24 h post-injection (8 h, P < 0.05) (Figs. [142]4B-C). Major organs were collected 8 h after injection, and IVIS revealed that the M2pep-LNP group accumulated more nanoparticles in the CRLM-bearing liver than that in the LNP group (Figs. [143]4D-E). In normal mice, no significant difference was found in the fluorescence signal between the LNP and M2pep-LNP groups 24 h after nanoparticle administration (Figures [144]S5A-D, Supporting Information). To further determine whether M2pep-LNP could specifically target M2-like TAMs in CRLM-bearing mice, Cy3-siRNA-loaded nanoparticles were intravenously injected into the mice. FCM revealed that 28.80 ± 4.93% of the total TAMs were Cy3 positive in the LNP group, which was significantly lower than that in the M2pep-LNP group (53.09 ± 12.88%, P < 0.05) (Figs. [145]4F-G). Among these TAMs, 34.39 ± 3.88% of M2-like TAMs were Cy3 positive in the LNP group, whereas this percentage was as high as 61.44 ± 12.85% in the M2pep-LNP group (P < 0.05) (Figs. [146]4F-G). Additionally, few DCs and T cells were Cy3 positive, with no significant difference between the two groups (P > 0.05) (Figs. [147]4F-G). IF analysis of liver tissue supported these findings, showing that the proportion of Cy3^+ TAMs in the M2pep-LNP group was significantly higher than that in the LNP group (55.63 ± 12.61% vs. 29.10 ± 10.09%, P < 0.05) (Fig. [148]4H). These results suggested that M2pep modification enhanced the targeting specificity of LNPs to M2-like TAMs in CRLM-bearing livers. Fig. 4. [149]Fig. 4 [150]Open in a new tab M2pep-modified LNPs preferentially accumulated in the liver metastases and showed superior targeting of M2-like TAMs in vivo. A) Schematic representation to assess the targeting ability of nanoparticles in vivo. Created using BioRender.com. B-C) Representative images (B) and quantitative results (C) of the nanoparticle biodistribution at different time points after systemic injection in mice bearing liver metastasis (n = 3 per group). Average radiant efficiency [(p/s/cm^2/sr) /(µW/cm^2)] was analyzed. D-E) Representative images (D) and quantitative analysis (E) of the nanoparticle biodistribution after 8 h of injection (n = 3 per group). Average radiant efficiency [(p/s/cm^2/sr) /(µW/cm^2)] was analyzed. F-G) FCM analysis of Cy3-expressing cells in liver metastases. F: Representative FCM images of total TAMs and M2-like TAMs. G: Quantitative results of Cy3-expressing cells, including CD45^+ CD3^− CD11b^+ F4/80^+ TAMs, CD45^+ CD3^− CD11b^+ F4/80^+ CD206^+ M2-like TAMs, CD45^+ CD3^− CD11b^+ F4/80^− CD11c^+ DCs, CD45^+CD11b^−CD3^+ T cells, and CD45^+ CD3^− CD11b^− non-myeloid and non-T immune cells. H) IF analysis of Cy3-expressing M2-like TAMs in liver metastases. Representative images (left) and quantitative results (right) are shown. Scale bar: 20 μm. Data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, as calculated using Student’s t test, multiple t test with Bonferroni post hoc tests, or one-way ANOVA with Tukey’s correction as appropriate. TAMs: Tumor-associated macrophages; FCM: Flow cytometry; IF: Immunofluorescence staining LNP-mediated TAM-selective Inhibition of Dicer delays CRLM growth and remodels the immunosuppressive tumor immune microenvironment (TIME) Based on these findings, different therapeutic modalities (siNC@LNP, siNC@M2pep-LNP, siDicer1@LNP, and siDicer1@M2pep-LNP) were administered to the CRLM mouse model, as scheduled (Fig. [151]5A). According to the in vivo bioluminescence analysis (Fig. [152]5B) and quantified tumor growth curves (Fig. [153]5C), intravenous injection of siDicer1-loaded nanoparticles, especially siDicer1@M2pep-LNP, effectively attenuated CRLM growth, resulting in prolonged survival compared to that with siNC@LNP treatment (increase 18.5% and 40.7% survive time, respectively) (Fig. [154]5D). At the end of the observation period, the mice were euthanized, and the weights of the mice and their livers were measured. Compared to other treatment groups, the siDicer1@M2pep-LNP group had the lowest relative liver/body weight (10.96 ± 2.61%) (Fig. [155]5E). HE staining of liver tissue also showed the smallest tumor area in the siDicer1@M2pep-LNP group (8.25 ± 5.71%) compared to that in the siNC@LNP and siNC@M2pep-LNP groups, which occupied 54.59 ± 22.64% and 56.12 ± 20.84% of the liver, respectively (Figs. [156]5F-G). Moreover, siDicer1-loaded nanoparticles significantly reduced the expression of Dicer and the pro-tumor cytokine IL-10 in tumor tissues, with the siDicer1@M2pep-LNP group showing the most pronounced effect (Fig. [157]5H). Fig. 5. [158]Fig. 5 [159]Open in a new tab TAM-targeted Dicer treatment delays CRLM progression and remodels the immunosuppressive TIME. A) Schematic diagram illustrating the modeling, drug administration, and IVIS imaging. Created using BioRender.com. B-C) Bioluminescence imaging (B) and quantitative results of mice bearing liver metastasis receiving different treatments (n = 4 for siNC@LNP and siNC@M2pep-LNP, n = 7 for siDicer1@LNP and siDicer1@M2pep-LNP). D) Kaplan-Meier survival curves in different treatment groups (n = 6 per group). E) Relative liver weight of mice bearing liver metastasis receiving different treatments (n = 4 for siNC@LNP and siNC@M2pep-LNP, n = 7 for siDicer1@LNP and siDicer1@M2pep-LNP). F-G) HE staining of liver sections from mice bearing liver metastasis receiving different treatments (n = 4 per group). F: quantitative analysis of tumor area. G: representative images. Scale bar: 1 mm. H) Protein expression levels of Dicer and IL-10 in metastatic liver tissues of mice receiving different treatments (n = 3 per group). I) FCM detection of immune cell infiltration in the liver metastases (n = 7). J-K) IF detection of various immune cell infiltration in liver metastases, including CD8^+ T cells, CD86^+ TAMs, and CD206^+ TAMs. J: quantitative analysis, K: representative images. Scale bar: 50 μm. Data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, as calculated using Student’s t test, multiple t-test with Bonferroni post hoc tests, or one-way ANOVA with Tukey’s correction as appropriate. CRLM: Colorectal cancer liver metastasis; TIME: Tumor immune microenvironment; TAMs: Tumor-associated macrophages; FCM: Flow cytometry, IF: Immunofluorescence staining To further investigate whether siDicer1-loaded nanoparticles inhibited CRLM growth by reprogramming TAMs, thereby reshaping the immunosuppressive TME, we used FCM to detect the proportion of various immune cells in the TME. The gating strategy used for the FCM analysis is shown in Figure [160]S6A (Supplementary Information). FCM results revealed that the proportion of M2-like TAMs in liver metastases significantly decreased, while the numbers of M1-like TAMs and CD8^+ T cells significantly increased after treatment with siDicer1-loaded nanoparticles, especially siDicer1@M2pep-LNP (Fig. [161]5I). However, no significant changes were found in the total numbers of TAMs, DCs, MDSCs, or CD4^+ T cell subpopulations (Figure [162]S6B, Supplementary Information). IF staining of tumor tissue also indicated that siDicer1@M2pep-LNP reduced the number of M2-like TAMs in liver metastases while increasing the number of M1-like TAMs and CD8^+ T cell infiltration, consistent with the FCM results (Figs. [163]5J-K). H&E staining and blood chemistry analysis revealed no significant toxic side effects of this RNAi drug (Figures [164]S7A-E, Supplementary Information). The anti-tumoral effect of Dicer Inhibition in CRLM is largely dependent on macrophages To investigate the importance of macrophages in Dicer-based cancer therapy, CRLM-bearing mice were treated with clodronate liposomes twice a week to deplete macrophages in vivo and received different treatments as scheduled (Fig. [165]6A). In vivo imaging and survival curve analysis revealed that the anti-tumoral effect mediated by siDicer1@M2pep-LNP was largely attenuated in mice treated with clodronate liposomes in CRLM models, characterized by increased tumor growth and shortened survival compared to those in the PBS + siDicer1@M2pep-LNP group (Figs. [166]6B-D). Liver tissue/body weight and HE staining also revealed that clodronate liposome intervention increased the tumor size and extent in mice bearing CRLM during siDicer1@M2pep-LNP treatment (Figs. [167]6E-G). FCM and IF staining results confirmed that mice treated with clodronate liposomes showed a decrease in F4/80^+ macrophages in liver metastases (Fig. [168]6H-J and M). Furthermore, we observed fewer M1-like TAMs and less CD8^+ T cell infiltration in the clodronate + siDicer1@M2pep-LNP group than in the PBS + siDicer1@M2pep-LNP group (Figs. [169]6K-M). Overall, these results demonstrate the potential of RNAi drugs targeting Dicer in TAMs in CRLM. Fig. 6. [170]Fig. 6 [171]Open in a new tab The anti-tumoral effect of Dicer treatment in CRLM is largely dependent on macrophages. A) Schematic diagram illustrating the modeling, drug administration, and IVIS imaging. Created using BioRender.com. B-C) Bioluminescence imaging (B) and quantitative results (C) of mice bearing liver metastasis receiving different treatments (n = 3 for PBS + siNC@M2pep-LNP, n = 4 for PBS + siDicer1@M2pep-LNP and Clodro + siDicer1@M2pep-LNP). D) Kaplan-Meier survival curves in different treatment groups (n = 5 per group). E) Relative liver weight of mice bearing liver metastasis receiving different treatments (n = 3 for PBS + siNC@M2pep-LNP, n = 4 for PBS + siDicer1@M2pep-LNP and Clodro + siDicer1@M2pep-LNP). F-G) HE staining of the liver sections of mice bearing liver metastasis receiving different treatments (n = 3 for PBS + siNC@M2pep-LNP, n = 4 for PBS + siDicer1@M2pep-LNP and Clodro + siDicer1@M2pep-LNP). F: Quantitative analysis of the tumor area. G: Representative images. Scale bar: 1 mm. H-I) FCM detection of macrophage infiltration in liver metastases. H: Representative FCM images, I: Quantitative analysis of TAM infiltration in liver metastases (n = 3). J-M) IF detection of immune cell infiltration in liver metastases. Quantitative analysis (J-L) and representative images (M) of various immune cells, including F4/80^+ TAMs, CD86^+ TAMs, and CD8^+ T cells. Scale bar: 50 μm. Data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, as calculated using Student’s t test, multiple t test with Bonferroni post hoc tests, or one-way ANOVA with Tukey’s correction as appropriate. TAMs: Tumor-associated macrophages; FCM: Flow cytometry; IF: Immunofluorescence staining Delivery of Dicer1 SiRNA reprograms M2-like macrophages through downregulation of miR-148a-3p and miR-1981-5p Previous studies have confirmed that Dicer plays a key role in miRNA maturation [[172]24]. To explore the potential mechanism by which siDicer1@M2pep-LNP reprograms M2-like macrophages, miRNA sequencing was performed. As shown in the volcano plot (Fig. [173]7A), a total of 74 upregulated and 55 downregulated genes were identified based on the criteria of |log2(FC)| > 0.5 and P < 0.05 (Table [174]S6). These 55 downregulated miRNAs were cross-referenced with miRNA clusters highly expressed in macrophages according to the MiRbase database (ER0000000235). This analysis identified five candidate miRNAs (miR-148a-3p, miR-146a-5p, miR-30a-5p, miR-30d-5p, and miR-1981-5p) (Fig. [175]7B). Subsequently, the expression levels of these five miRNAs were validated in IL-4 pre-treated BMDMs and RAW264.7 cells followed by Dicer knockdown. The results showed that miR-148a-3p and miR-1981-5p were significantly decreased in both BMDMs and RAW264.7 cells (Figs. [176]7C-D), suggesting that these two miRNAs are potential targets affected by Dicer1 silencing. Fig. 7. [177]Fig. 7 [178]Open in a new tab Dicer down-regulation reprograms M2-like macrophages through downregulation of miR-148a-3p and miR-1981-5p. A) IL-4 pre-treated BMDMs were transfected with siDicer1@M2pep-LNP for 48 h followed by miRNA sequencing. Volcano plot of differentially expressed miRNAs.|log2(FC)| > 0.5 and P < 0.05. B) Intersection of differentially expressed miRNAs with miRNAs highly expressed in macrophages (ER0000000235).|log2(FC)| > 1 and P < 0.05. C-D) Validation of selected candidate miRNAs in Dicer-KD M2-like BMDMs and RAW264.7 cells, detected using miRNA qRT-PCR. E-G) IL-4 pre-treated BMDMs were transfected with miRNA mimics or inhibitors for 48 h. E-F) The expression levels of corresponding miRNA- and macrophage-related cytokines were measured using qRT-PCR. G: The MFI of CD86 and CD206 detected using FCM. H-J) IL-4 pre-treated BMDM cells were treated with Dicer1 siRNA for 24 h, followed by transfection with miR-148a-3p mimics or miR-1981-5p mimics for 48 h. Representative images (H) and quantitative analysis (I) of CD86 and CD206 were detected by FCM. J) The expression levels of macrophage-related cytokines were detected using qRT-PCR. K) Protein expression levels of Dicer, total AKT and Phospho-AKT in M2-like BMDMs transfected with Dicer1 siRNA or miRNA inhibitors, detected using western blotting. β-actin and U6 were used as endogenous mRNA and miRNA quantification controls, respectively. Data are presented as the mean ± SD. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, as calculated using Student’s t test, multiple t test with Bonferroni post hoc tests, or one-way ANOVA with Tukey’s correction as appropriate. BMDMs: Bone marrow-derived macrophages, FCM: Flow cytometry, MFI: Mean fluorescence intensity Next, corresponding mimics and inhibitors were designed for the two miRNA sequences. As shown in Figs. [179]7E-F, inhibitors of miR-148a-3p and miR-1981-5p significantly reduced the mRNA expressions of anti-inflammatory markers IL-10, TGFβ1, and Fizz1, while increasing the mRNA expressions of pro-inflammatory markers IL-1β, TNF-α, CXCL9, and CXCL10. FCM revealed that the miR-148a-3p inhibitor downregulated CD206 expression, whereas the miR-1981-5p inhibitor upregulated CD86 expression in M2-like BMDMs (Figs. [180]7G). In M2-like BMDMs with Dicer knockdown, miR-148a-3p and miR-1981-5p mimics restored the M2-like phenotype, characterized by decreased expression of M1 markers such as CD86, IL-1β, and TNF-α, and increased expressions of M2 markers including CD206, IL-10, and TGFβ1 (Figs. [181]7H-J). Finally, the potential targets of miR-148a-3p and miR-1981-5p were then examined, identifying 158 and 143 potential candidates through databases miRmap and microT (Figure [182]S9, Supplementary Information). Subsequent KEGG pathway enrichment analysis of the results intersected from both databases revealed that the Akt signaling pathway was the most significantly enriched pathway for both miRNAs. Western blot analysis showed that inhibition of Dicer or these two miRNA led to a marked suppression of the Akt signaling pathway, as evidenced by decreased levels of phosphorylated AKT (Fig. [183]7K). These findings suggest that Dicer knockdown may promote M1 macrophage polarization through miR-148a-3p and miR-1981-5p-mediated regulation of the Akt pathway. Discussion The liver microenvironment of CRLM harbors abundant immunosuppressive macrophages, which provide a protective niche to support CRC growth and form metastatic lesions in the liver. However, inhibiting macrophages recruitment or directly depleting them would indiscriminately weak their intrinsic capability of tumor-phagocytosis and antigen presentation. Thus, the alternative strategies such as TAMs-repolarization could be good choices for macrophage-based immunotherapy [[184]25]. Unfortunately, the driver regulators of macrophage phenotypic polarization in CRLM remain incompletely elucidated. This study identified Dicer as a critical regulator that drives M2 polarization of TAMs in CRLM. Inhibition of Dicer skews macrophages toward an anti-tumoral M1 phenotype, thereby suppressing tumor growth in vivo. These studies highlight the possibility of targeting Dicer to manipulate an anti-tumoral immune microenvironment in CRLM. However, the development of a therapy that targets Dicer in TAMs remains challenging. First, small-molecule inhibitors specific to Dicer have not been reported, and second, Dicer is an intracellular protein. Although antibodies hold enormous potential for drugging specific targets, they are largely restricted to extracellular targets owing to the cell membrane barrier. Therefore, we used RNA interference (RNAi) with a clinically available LNP platform to inhibit macrophage Dicer expression and explored the feasibility of Dicer deficiency for CRLM treatment in animal experiments. To improve the targeted delivery of siDicer1 to M2 TAMs, we further optimized LNP with M2pep, a peptide that can specifically target M2-polarized macrophages [[185]26] engineered onto the surface. The generated nanosystem enabled in situ repolarization of M2 TAMs toward the M1 state in the metastatic liver, which showed profound anti-tumor efficacy in mice bearing CRLM. Moreover, depletion of TAMs in mice bearing CRLM using Clodronate liposomes attenuated the anti-tumor efficacy of siDicer1@M2pep-LNPs, providing proof-of-concept evidence that targeting Dicer in TAMs in vivo can suppress metastatic tumor growth. Liver metastasis in CRC is often characterized by high resistance to current immunotherapies, largely due to the immunosuppressive microenvironment. In our study, we showed that LNP-mediated siDicer1 delivery increased the M1/M2 ratio of TAMs as well as facilitated CD8^+ T cell infiltration in the metastatic niche in the liver, suggesting the establishment of an immunostimulatory microenvironment. However, whether the anti-tumor effects of Dicer knockdown in macrophages are solely attributable to macrophage re-polarization, or also involve additional mechanisms, remains to be further explored. Macrophages are key professional phagocytes within the innate immune system, responsible for the direct engulfment of tumor cells and the clearance of apoptotic cells through efferocytosis-processes that critically shape the immune landscape within the tumor microenvironment [[186]25]. Therapeutic strategies that simultaneously enhance the macrophage-mediated phagocytosis of tumor cells while inhibiting efferocytosis have shown considerable promise in cancer treatment [[187]17, [188]27]. For example, blocking efferocytosis can induce immunogenic cell death (ICD) in dying or apoptotic tumor cells that are not eliminated, promoting the release of damage-associated molecular patterns (DAMPs) and subsequent activation of adaptive immune responses. This effect can be further augmented when combined with checkpoint inhibitors. Recent studies have shown that Dicer deletion not only suppresses macrophages polarization towards the M2 phenotype but also impairs their efferocytosis capacity, implicating Dicer in the pathogenesis of systemic lupus erythematosus and other inflammatory diseases [[189]28, [190]29]. Consistent with these findings, our study also demonstrated that Dicer knockdown enhances macrophage-mediated phagocytosis of tumor cells. These findings indicate that Dicer exerts a broader regulatory role in macrophage function than previously appreciated. Thus, targeting Dicer in macrophages, particularly in combination with clinically approved checkpoint inhibitors, may represent a promising strategy to overcome immunotherapy resistance and improve treatment outcomes. Post-transcriptional regulation plays a vital role in macrophage activation and polarization. miRNAs are considered one of the most important regulators at the post-transcriptional level. Emerging studies have focused on investigating the impact of miRNA dysregulation on macrophage polarization. Several miRNAs have been implicated in promoting M2 macrophage polarization. For example, miR-511-3p, miR-223, and Let-7c, have been reported to regulate M2 macrophage activation [[191]30]. However, how Dicer orchestrates the downstream miRNA network to regulate macrophage polarization remains elusive. In this study, we used RNAi to inhibit Dicer, and identified five miRNAs as potential regulators of M2 macrophage activation based on miRNA sequencing. Further analysis showed that inhibition of miR-148a-3p and miR-1981-5p increased the expression of M1 markers and decreased the expression of M2 markers via inhibiting Akt pathway, suggesting their roles in driving macrophage M2 polarization. Several studies confirmed that the Akt pathway converges inflammatory and metabolic signals to regulate macrophage phenotype [[192]31]. Our results are consistent with those of previous studies showing that knockout of miR-148a induces a phenotypic shift in macrophages toward a pro-inflammatory state [[193]32, [194]33]. However, Zhu et al.. found that NF-κB and STAT3 signaling were activated in miR-148a knockout macrophages when stimulated with LPS [[195]33]. Besides, based on the genetic Dicer1 deletion method, Let-7-5p is a positive regulator of M2 polarization in TAMs [[196]11]. Using the same approach as that used for Dicer1 deletion, Wei et al.. revealed that miR-10a, let-7b, and miR-195a are responsible for generating alternatively activated M2 macrophages to prevent atherosclerosis [[197]13]. These studies suggest that the key miRNAs and downstream pathways regulated by Dicer for macrophage polarization may vary between different diseases and cancer types. Our findings provide insights into the regulation of macrophage polarization by Dicer and the downstream miRNA network, presenting initial clues for the design of future Dicer- or miRNA-targeted immunotherapies for CRLM. What is worth mention is that Dicer expression is not restricted to TAMs in the liver microenvironment; other cell types, including tumor-associated fibroblasts, also express Dicer [[198]34]. A previous study demonstrated that loss of Dicer in tumor-associated fibroblasts hampered their role in supporting tumor growth in a xenograft model of ovarian cancer [[199]34]. Whether siDicer1@M2pep-LNPs decrease Dicer expression in tumor-associated fibroblasts in the microenvironment of CRLM and contribute to overall anti-tumor efficacy remains to be investigated. Although LNP mediated RNA delivery is a promising approach for Dicer intervention. The development of alternative approaches to target Dicer using cutting-edge technologies, such as artificial intelligence (AI)-powered discovery of small-molecule inhibitors [[200]35]proteolysis-targeting chimeras (PROTAC) for targeted Dicer degradation [[201]36]and TAM-specific gene editing, also holds great promise for establishing Dicer-based immunotherapies. Conclusion Taken together, we identified Dicer as a critical regulator of M2 polarization of TAMs in CRLM. We further fabricated a TAM-targeting LNP nanosystem to deliver siDicer1 for in situ manipulation of TAMs, which reversed the immunosuppressive microenvironment and displayed remarkable anti-tumor activity in mice bearing CRLM. Our study also elucidated the regulatory mechanism of siDicer1 delivery in reprogramming macrophages from the M2 to M1 phenotype. Targeting Dicer or specific miRNAs in the macrophages may thus provide a new treatment strategy for patients with CRLM. Electronic supplementary material Below is the link to the electronic supplementary material. [202]Supplementary Material 1^ (45.3KB, docx) [203]Supplementary Material 2^ (330.7KB, pdf) [204]Supplementary Material 3^ (4.4MB, docx) Acknowledgements