Introduction

Colorectal cancer (CRC) is the second leading cause of cancer death worldwide with more than 0.9 million deaths in 2020 [1, 2]. Liver metastases occur in around 50% of CRC patients during the whole disease course [3]. Surgical resection is regarded as the only curative treatment but is only applicable to about 25% of patients with a 5-year recurrence-free survival around 30% [4, 5]. For those with nonresectable liver metastases, local ablation, and systemic chemotherapy are adopted, and around 13% of the patients can undergo hepatic resection with a 5-year survival rate of 33% [6]. However, for more than 60% of the patients with liver metastases, the progression-free survival is less than 10 months after different combined chemotherapies [7,8,9], and the patients have little chance to survive to 5 years. Further combination uses of immune checkpoint blockade (ICB) therapy and chemotherapy in untreated patients with microsatellite-instability-high (MSI-H) advanced CRC prolong the median progression-free survival to 16.5 months [10], highlighting the potential of immunotherapy in the treatment of CRC with liver metastases. However, the overall response rate is only around 40% in MSI-H patients, despite the low abundance of MSI-H patients (<10% of all CRC patients) [10, 11]. Therefore, more effective chemoimmunotherapy is required for the treatment of CRC with liver metastases.

The unsatisfactory response of CRC patients with liver metastases to chemoimmunotherapy is largely associated with the unique immunologically tolerant microenvironment in the liver [12]. Physiologically, hepatic dendritic cells (DCs) exhibit less capability to active T cells compared with residential DCs in other tissues, and Kupffer cells and liver sinusoidal endothelial cells (LSECs) can promote CD8+ T cell anergy and Treg cell priming [13,14,15,16]. Pathologically, activated Treg cell accumulation is more prominent in the liver metastases of CRC than in healthy liver tissues [17]. Recently, hepatic monocyte-derived macrophages (Mono-M) have been found to impose a systemic suppressive immune state, by inducing tumor antigen-specific CD8+ T cell apoptosis via FasL-Fas pathway in the presence of liver metastases [18]. The presence of these Mono-M in the liver with metastases is believed to negatively associate with the prognosis of ICB therapy in both patients and animal models [18], and thus an effective strategy to modulate the Mono-M is of urgent need.

Mono-M are the major source of tumor-associated macrophages, and are recruited to and differentiated in livers with metastatic cancers [19, 20]. Based on their polarization status, Mono-M can be classified into classically activated macrophages (Mono-M1) and alternatively activated macrophages (Mono-M2) [19, 21]. Mono-M2 are arguably the main cells that inhibit antitumor immunity, while Mono-M1 mainly exert antitumor activity [22]. Given the essential roles of Mono-M in regulating antitumor immunity, strategies to modulate their functions have been extensively explored [22,23,24]. For instance, antibodies that block the key molecules in macrophage recruitment, activation, and survival have been developed [25,26,27,28,29]. However, systemic treatments with current antibodies are either of limited efficacy or side effects due to their deposition in non-specific organs. To address this issue, nanocarriers that can target Mono-M in the diseased sites have been explored [23, 30], including liposomes, cationic nanoparticles, polymers, gold nanoparticles, and exosomes [23, 31,32,33,34]. These nanocarriers could efficiently target tumor-associated macrophages, and improve the efficacy of cancer immunotherapy. Despite these advances, targeted modulation of a certain subtype of Mono-M is still required to further improve the specificity of treatment. High-density lipoprotein (HDL) and its synthetic version (sHDL) have the potential to target myeloid cells as carriers for contrast agents, drugs, and neoantigens with good biocompatibility [35,36,37,38,39]. Recently, we found that sHDL was able to accumulate in subcutaneous tumors and preferentially modulate intratumoral Mono-M2 over Mono-M1 based on their differences in the expression of scavenger receptor class B type 1 (SR-B1) [40]. Currently, the cell-specificity of sHDL in targeting Mono-M2 in the diseased livers in the presence of Kupffer cells has not been investigated, which is crucial for the treatment of CRC with liver metastases.

Herein, we found that the number of Mono-M2, rather than those of Mono-M1, dendritic cells (DCs), and neutrophils, was significantly increased in the livers of mice bearing both subcutaneous tumors and liver metastases than those in the mice with only subcutaneous tumors. To preferentially modulate these hepatic Mono-M2, we prepared a gemcitabine-loaded sHDL (G-sHDL) using lipids and apolipoprotein A-1 (ApoA-1) mimetic peptides (Fig. 1a). After intravenous injection, G-sHDL could accumulate in both subcutaneous tumors and livers, where it induced immunogenic cell death (ICD) and eradicated Mono-M2, respectively. The two effects together improved the densities of tumor antigen-specific CD8+ T cells in the blood, livers with CRC metastases, subcutaneous tumors and tumor-draining lymph nodes (TDLNs), halting the growth of CRC tumors and liver metastases especially used in conjunction with anti-PD-L1 therapy (Fig. 1b).

Fig. 1: Schematic illustration of preparation and mechanism of G-sHDL.
figure 1

a Preparation of G-sHDL. b The mechanism of action of G-sHDL. The G-sHDL could target and kill hepatic Mono-M2 that might eradicate tumor antigen-specific CD8+ T cells, and could also induce ICD of tumor cells to prime the antitumor immunity. Consequently, G-sHDL would improve the number and activity of tumor antigen-specific CD8+ T cells that subsequently killed tumor cells especially at the presence of anti-PD-L1 antibody.

Full size image

Materials and methods

Materials

ApopA-1 peptide (Ac-GFAEKFKEAVKDYFAKFWD-OH, purity >95%) was purchased from GL Biochem Ltd (Shanghai, China). Gemcitabine, DiD, and DiR were purchased from Dalian Meilun Biotechnology Co., Ltd (Dalian, China). Cholesteryl oleate (CO), OVA peptide, and palmitic acid were purchased from Sigma-Aldrich (St. Louis, MO, USA). Cholesterol, hydrogenated soybean phosphatidylcholine (HSPC), and distearoyl phosphatidyl ethanolamine-polyethylene glycol (DSPE-MPEG2000) were purchased from AVT Ltd (Shanghai, China). Macrophage colony-stimulating factor (M-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), interleukin-4 (IL-4), interleukin-2 (IL-2), interleukin-13 (IL-13), interferon-γ (IFN-γ) and Fixable Viability Dye eFluor™ 455UV were purchased from Thermo Fisher Scientific (Waltham, MA, USA). Lipopolysaccharides (LPS), mouse high mobility group box 1 protein (HMGB1) ELISA kit, DNase I, and β-mercaptoethanol (β-ME) were purchased from Solarbio LIFE SCIENCES (Beijing, China). Collagenase IV, hyaluronidase, and Annexin V-PE/7-AAD apoptosis detection kits were purchased from Yeasen Biotechnology Co., Ltd (Shanghai, China). Oleic acid, N,N-diisopropylethylamine, dimyristoylphosphatidylcholine (DMPC), and N,N,N′,N′-tetramethyl-O-(7-azabenzotriazol-1-yl) uronium hexafluorophosphate (HATU) were purchased from J&K Scientific (Shanghai, China). Clodronate liposome (Cl-Lipo) was purchased from Liposoma BV (Amsterdam, Netherlands). RPMI-1640 medium and IMDM medium were purchased from Shanghai BasalMedia Technologies Co., Ltd (Shanghai, China). EasySep™ Mouse F4/80 Positive Selection Kit, EasySep™ Mouse CD8a Positive Selection Kit, and dispersase were purchased from STEMCELL Technologies (Vancouver, Canada). Collagenase I, Collagenase II, and CFSE were purchased from Shanghai Maokang Biotechnology Co., Ltd (Shanghai, China). ATP assay Kit was purchased from Beyotime Biotechnology (Shanghai, China). ELISA kits for tumor necrosis factor-α (TNF-α) and IL12p40 were purchased from Neobioscience Technology Co., Ltd (Shenzhen, China). Anti-mouse PD-L1 antibody was purchased from Bioxcell (West Lebanon, NH, USA). Zombie Red™ fixable viability kit was purchased from Biolegend (San Diego, CA, USA). BD Cytofix/CytopermTM fixation/permeabilization kit was purchased from BD Biosciences (Franklin Lake, NJ, USA). Flow antibodies were listed as follows: CD45 (cat. 103115), F4/80 (cat. 123145), CD86 (cat. 105007), CD206 (cat. 141703), CCR2 (cat. 150609), TIM-4 (cat. 130009) and CD4 (cat. 100405) were purchased from Biolegend (San Diego, CA, USA). CD11b (cat. 65-0112-U025), F4/80 (cat. 50-4801-U025), CD86 (cat. 20-0862-u025), purified anti-mouse CD16/CD32 (cat. 70-0161), CD11c (cat. 35-0114-U100), CD80 (cat. 20-0801-U025), CD3e (cat. 65-0031-U025) and Ly-6G (cat. 50-5931-U100) were purchased from TONBO Biosciences (San Diego, CA, USA). CD86 (cat. 560582), CD8a (cat. 553032), and CD90.2 (cat. 553932) were purchased from BD Biosciences (Franklin Lake, NJ, USA). CD206 (cat. 25-2061-80) was purchased from eBioscience (San Diego, CA, USA). SR-B1 (cat. NB400-104AF488) was purchased from Novus Biologicals (Littleton, CO, USA). Calreticulin (cat. 62304) and cleaved caspase-3 (cat. 9603) were purchased from Cell Signaling Technology (Danvers, MA, USA). Ep-CAM (cat. 25-5791-80) was purchased from Invitrogen (Carlsbad, CA, USA). Unless otherwise noted, all other reagents were obtained from Sinopharm Chemical Reagent Co. Ltd (Shanghai, China).

Cells and animals

Murine MC38, MC38-LUC, and MC38-OVA tumor cells were obtained from FuHeng BioLogy (Shanghai, China). The cells were cultured in RPMI-1640 containing 10% fetal bovine serum (Life Technology, Carlsbad, CA, USA) and 1% antibiotics. Cells were maintained at 37 °C in a humidified incubator containing 5% CO2.

Female C57BL/6 mice (8–10 weeks old) were purchased from the Shanghai Experimental Animal Center (Shanghai, China), and female OT-1 mice (8–10 weeks old) were purchased from Cyagen Biosciences Inc (Suzhou, China). All mice were used under guidelines approved by the Institutional Animal Care and Use Committee of the Shanghai Institute of Materia Medica, Chinese Academy of Sciences (2021-06-LYP-43 and 2022-06-LYP-44).

Tumor models

For subcutaneous tumor models (SC), MC38 cells (2 × 106) were subcutaneously inoculated. For liver metastases models, MC38 cells (5 × 105) were injected into the left lobule of the livers. For mice bearing both subcutaneous tumors and liver metastases (SC+Liver), two tumors were established at the same time. Mice were randomized into experimental groups when subcutaneous tumors reached 50 mm3.

Animal treatment protocols

Gemcitabine-based treatments (Gem, G-Lipo, and G-sHDL) were given by intravenous injection (15 mg/kg, every 4 days for three times). For hepatic macrophages depletion, clodronate liposome (Cl-Lipo) was given by intraperitoneal injection (200 µL first dose, 100 µL later doses) [18]. For immune checkpoint blockade therapy, anti-mouse PD-L1 antibody was given by intraperitoneal injection (100 µg per dose), one day after gemcitabine-based treatments. The animals were monitored for tumor growth and body weight change every 2 days after the first treatment.

Flow cytometry

Single-cell suspensions were prepared from livers, lymph nodes, subcutaneous tumors, and blood as we described before [41]. The cell numbers were determined by Countess™ II automated cell counter (Thermo Fisher Scientific). Dead cells were excluded by fixable viability dye. After surface staining with corresponding fluorochrome-conjugated antibodies for 30 min, 150 µL Fixation/Permeabilization solution was added into 100 µL cell suspension for 20 min if intracellular staining was needed. After that, the cells were washed with BD Perm/Wash buffer and stained with intracellular antibodies for another 30 min. Purified anti-mouse CD16/CD32 antibody was used to block the Fc receptors on macrophages. The biomarkers of various cells were listed as follows: DCs (CD45+CD11c+), Neutrophils (CD45+CD11b+Gr-1+), MC38 tumor cells (Ep-CAM+), CD8+ T cells (CD45+CD3+CD8+), CD4+ T cells (CD45+CD3+CD4+), Macrophages (CD45+CD11b+F4/80+), Kupffer cells (CD45+CD11b+F4/80+TiM-4+), Mono-M (CD45+CD11b+F4/80+CCR2+), Mono-M1 (CD45+CD11b+F4/80+CCR2+CD86+), Mono-M2 (CD45+CD11b+F4/80+CCR2+CD206+). Data were analyzed by FlowJo v10.6.2.

Preparation and characterization of sHDLs

We used the film-hydration method to prepare sHDLs according to our previous method [40, 41]. In brief, DMPC (3.33 mg) and CO (0.13 mg) were dissolved in 12 mL solvent (methanol: dichloromethane = 1:3, v/v), and then a thin film was formed under vacuum. The film was re-hydrated with 8 mL PBS, and sonicated with probe sonicator while ApoA-1 peptide (4 mg) was added. The suspension was centrifuged (7000 × g, 10 min), and then the supernatant was concentrated using ultrafiltration tubes (COMW = 30 kDa). To prepare G-sHDL, gemcitabine-oleic acid ester (GemE), a liable prodrug of gemcitabine was added into the film [40]. To get fluorescent labeled sHDLs, DiD (D-sHDL) or DiR (R-sHDL) with palmitic acid were added at the molar ratio of 3:5 under dark [40].

We prepared liposomes using film-hydration method [42]. To get gemcitabine-loaded liposome (G-Lipo), HSPC (40 mg), CH (5 mg), GemE (8 mg), and DSPE-MPEG2000 (4 mg) were dissolved in 5 mL solvent (methanol: dichloromethane = 1:1, v/v), and then a film was obtained under vacuum. The film was re-hydrated with 1 mL PBS, and sonicated with probe sonicator. To get fluorescent labeled liposomes, DiD (D-Lipo) or DiR (R-Lipo) with palmitic acid were added at the molar ratio of 4:1 under dark.

All the sHDLs and liposomes were observed with bright-field transmission electron microscopy (TEM, Talos L120C, FEI, USA) for their size and morphology using a negative staining method with uranyl acetate. Their zeta potentials were measured by dynamic light scattering (DLS, ZS90, Marven, UK). The encapsulation efficiency (EE) and drug loading (DL) were calculated by analyzing the content of gemcitabine-oleic acid ester with HPLC after destroying the nano-structure with solvent consisting of methanol and DMSO (1:1, v/v). DL and EE were calculated as follows.

$${{{{{\rm{EE}}}}}}={W}_{{{{{{\rm{drug}}}}}}\,{{{{{\rm{in}}}}}}\,{{{{{\rm{sHDL}}}}}}}/{W}_{{{{{{\rm{drug}}}}}}\,{{{{{\rm{added}}}}}}} \times 100 \%$$
$${{{{{\rm{DL}}}}}}={W}_{{{{{{\rm{drug}}}}}}\,{{{{{\rm{in}}}}}}\,{{{{{\rm{sHDL}}}}}}}/{W}_{{{{{{\rm{sHDL}}}}}}} \times 100 \%$$

Cytotoxicity assay

MC38 cells were seeded into 96-well plates (8000 cells/well) for 24 h, and then treated with drugs for another 24 h before the measurement of cell viability with CCK-8 kit. The cells were treated by Gem, G-Lipo, or G-sHDL (from 0.001 µM to 100 µM in gemcitabine). All the experiments were performed in triplicate.

Live animal imaging and biodistribution

Mice models bearing liver MC38-LUC tumors were established as above. One dose of R-sHDL, R-Lipo, or free DiR (10 μg DiR/dose) was given by intravenous injection, respectively (n = 5). The distribution of DiR in vivo was observed at 1 h, 4 h, 8 h, 12 h, 24 h, 48 h, and 5 d after injection by an IVIS spectrum imaging system (PE IVIS SPECTRUM, Perkin Elmer, USA).

Immunofluorescence

Mice models bearing liver MC38 tumors were established as above. One dose of D-sHDL or D-Lipo (10 μg DiD/dose) was given by intravenous injection, respectively (n = 3). Their livers were collected 24 h later and used for immunofluorescence slices. The semi-quantification was performed with ImageJ.

Hepatic and intratumoral immune cells affected by liver metastases

To investigate the impact of liver metastases on hepatic immune microenvironment, we established two different tumor models (SC and SC+Liver) as described (n = 5). The mice were sacrificed and their livers were collected when the subcutaneous tumors reached 50 mm3. The hepatic single-cell suspension was obtained to analyze hepatic DCs, neutrophils, and subtypes of macrophages by flow cytometry.

To investigate the impact of liver metastases on tumor immune microenvironment, we established two different tumor models (SC and SC+Liver) as described (n = 6) and analyzed the T cells within the SC tumors of these two models.

Hepatic immune cells affected by Cl-Lipo

To investigate the impact of Cl-Lipo on hepatic macrophages, we established SC+Liver model mice as described and divided them into two experimental groups when the subcutaneous tumors reached 50 mm3 (n = 5). They were treated with PBS or Cl-Lipo, respectively. After three days, we collected their livers and analyzed the subtypes of hepatic macrophages by flow cytometry.

Cellular uptake

SC+Liver model mice were established and sacrificed when the subcutaneous tumors reached 50 mm3 (n = 5). We collected their livers and analyzed the levels of SR-B1 on subtypes of hepatic macrophages and tumor cells via flow cytometry.

Another two experimental groups (n = 6) were established as above. One dose of D-sHDL or D-Lipo (10 μg DiD/dose) was given by intravenous injection, respectively. After treatment for 24 h, the mice were sacrificed and their livers were collected. The hepatic single-cell suspension was obtained in the dark. The uptake of DiD by various cells was examined with flow cytometry.

To obtain bone-marrow-derived macrophages (BMDMs), bone-marrow in femurs and tibias of healthy C57BL/6 mice was harvested and cultured in IMDM medium (containing 10% fetal bovine serum and 1% antibiotics) in a 75 cm2 culture flask for 48 h. Then the non-adherent cells were collected and cultured in IMDM medium, and 40 ng/mL M-CSF was added. Half medium was replaced every 2 days. On day 7, IFN-γ (50 ng/mL) and LPS (100 ng/mL) were used to induce the differentiation of Mono-M1 while IL-4 (20 ng/mL) and IL-13 (20 ng/mL) were used to induce the differentiation of Mono-M2 for 24 h [43].

BMDMs were seeded into six-well-plates and cocultured with D-sHDL or D-Lipo (0.08 µg/mL in DiD) for 30 min, and then cultured with drug-free medium for another 24 h. After that, the uptake of DiD in Mono-M1 (CD11b+F4/80+CD86+) and Mono-M2 (CD11b+F4/80+CD206+) was examined by flow cytometry. MC38 tumor cells were seeded into six-well plates (5 × 105 cells/well) and cultured for 24 h before cocultured with D-sHDL or D-Lipo (0.08 µg/mL in DiD) for 30 min. After another 24 h, the uptake of DiD in MC38 tumor cells in vitro was examined by flow cytometry. All the experiments were performed in triplicate.

Viability of BMDMs in vitro

BMDMs were seeded into 12-well-plates and cocultured with Gem, G-Lipo, G-sHDL (15 µg/mL in gemcitabine) or Cl-Lipo (1:200 dilution) for 30 min, and then cultured with drug-free medium for another 24 h. BMDMs treated with PBS served as negative control. The viability of Mono-M1 (CD11b+F4/80+CD86+) and Mono-M2 (CD11b+F4/80+CD206+) was analyzed with fixable viability kit via flow cytometry. All the experiments were performed in triplicate.

Immunogenic cell death in vitro and DC maturation

MC38 cells were seeded into 12-well plates (1 × 105 cells/well) for 24 h, and then treated with gemcitabine-based drugs (10 µg/mL in gemcitabine) for 4 h. The cells were then cultured with drug-free medium for 12 h for the measurements of ATP and HMGB1, and for 24 h for further staining with fixable viability kit and anti-Calreticulin-Alexa Flour 488 [44]. All the experiments were performed in triplicate.

To obtain bone-marrow-derived dendritic cells (BMDCs), bone-marrow in femurs and tibias of healthy C57BL/6 mice was harvested and cultured in RPMI-1640 medium, and 20 ng/mL GM-CSF and 20 ng/mL IL-4 was added. Half culture medium was replaced every 2 days. The immature BMDCs were obtained on day 6 [45]. The BMDCs were cocultured with MC38 cells treated as above for 24 h. BMDCs treated by LPS (1 µg/mL) served as positive control. Then we collected the BMDCs and culture medium, respectively. The cells were stained with anti-CD11c-FITC, anti-CD80-APC, and anti-CD86-PE antibodies, and then analyzed via flow cytometry. Besides, the concentrations of IL12p40 and TNF-α (tumor necrosis factor-α) in the culture medium were determined with ELISA kits. All the experiments were performed in triplicate.

SC+Liver model mice were established as above. The mice were randomized into four experimental groups (n = 5) when the subcutaneous tumors reached 50 mm3. One dose of PBS, Gem, G-Lipo, or G-sHDL was given by intravenous injection, respectively. After three days, the mice were sacrificed and their TDLNs were collected. The proportion of mature DCs in total DCs was analyzed by flow cytometry. To quantify the contents of IL12p40 and TNF-α in TDLNs, another four experimental groups (n = 3) were established and treated as above. Their TDLNs were harvested, weighed, and homogenized, and the supernatants were analyzed by ELISA kits.

Antitumor immunity

SC+Liver model mice were established as above. The mice were randomized into five experimental groups (n = 5) when the subcutaneous tumors reached 50 mm3. One dose of PBS, Gem, G-Lipo, or G-sHDL was given by intravenous injection, while Cl-Lipo was given by intraperitoneal injection, respectively. The hepatic macrophages and CD8+ T cells were analyzed three days later. Besides, the macrophages and T cells within SC tumors were analyzed five days later with another five experimental groups (n = 5) established and treated as above.

To investigate the tumor-antigen specific CD8+ T cells, SC+Liver model mice bearing both liver and subcutaneous MC38-OVA tumors were established. Mice were randomized into five experimental groups (n = 5) when the subcutaneous tumors reached 50 mm3. Two million CFSE-labeled OT-1 cells obtained as reported were given by intravenous injection two days after various treatments [18]. The numbers of OT-1 cells in blood, TDLNs, livers, and SC tumors were analyzed by flow cytometry five days later.

Hepatic macrophage-induced OT-1 cell apoptosis in vitro

Mice models bearing liver MC38-OVA tumors were established as above. One dose of PBS, Gem, G-Lipo, or G-sHDL was given by intravenous injection, while Cl-Lipo was given by intraperitoneal injection, respectively. After three days, hepatic F4/80+ cells were separated by EasySep™ mouse F4/80 positives selection kit and analyzed via flow cytometry. CFSE-labeled OT-1 cells (5 × 104) were cocultured with 2 × 105 F4/80+ cells and 1 × 105 MC38-OVA tumor cells in a 48-well-plate for 48 h before analyzing the apoptosis of OT-1 cells [18]. All the experiments were performed in triplicate.

Antitumor efficacy in vivo

SC+Liver model mice were established as above. The mice were randomized into five experimental groups when the subcutaneous tumors reached 50 mm3. Three doses of various drugs were given every 4 days. The subcutaneous tumor volumes and body weights of mice were monitored every two days. At the end of the experiment, we collected the SC tumors and weighed. Besides, we harvested their livers to observe the liver metastases. The major organs (heart, liver, spleen, lung, and kidney) and blood were collected to detect the drug toxicity.

To test the antitumor efficacy of combination strategy, mice models were randomly divided into four experimental groups. Three doses of G-sHDL were given by intravenous injection every four days, and anti-mouse PD-L1 antibody was given by intraperitoneal injection one day later. The subcutaneous tumor volumes and body weights were monitored every two days. At the end of the experiment, we collected the SC tumors and weighed. To detect liver metastases, we observed the bioluminescence of MC38-LUC tumors via an IVIS Spectrum imaging system [18, 46].

The SC tumor volume (V) was calculated with the major axis (L) and minor axis (W) of the tumors (V = L × W2/2).

Statistical analysis

Statistical analysis of data was performed using GraphPad Prism 8.0.1. Unpaired t test was used for the comparison of two groups. One-way ANOVA or two-way ANOVA was used for comparison across multiple experimental groups with single factor or two factors. Log-rank was used for survival rate analysis. The difference was considered statistically significant if P value was less than 0.05. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. All results were expressed as mean ± SD with sample size.

Results

Hepatic immune microenvironment in livers with CRC metastases

To investigate the impact of liver metastases on hepatic immune microenvironment, we established two different tumor models (SC and SC+Liver) and validated the formation of liver metastases with hematoxylin and eosin (H&E) stain (Fig. 2a). Flow cytometry was performed to analyze the differences between the two models in the numbers of DCs, neutrophils, and macrophages in the livers. In consistence with previous report [18], we found that liver metastases significantly increased the number of hepatic macrophages but not DCs and neutrophils (Fig. 2b). Further analysis revealed that the Mono-M and particularly the Mono-M2 fraction in the hepatic macrophages were significantly increased in SC+Liver models, while the fraction of the Kupffer cells was not different between the two models, leading to a higher Mono-M2/Mono-M1 ratio in the livers of SC+Liver models (Fig. 2c). We also studied the impact of liver metastases on tumor immune microenvironment (TIME). The results revealed that the presence of liver metastases significantly decreased the density of intratumoral T cells as well as the frequency of CD8+ T cells, which caused an obvious decrease in the CD8+ T cells/CD4+ T cells ratio, a feature of suppressive TIME (Supplementary Fig. S1).

Fig. 2: Liver metastases alter the hepatic immune microenvironment.
figure 2

a Establishment of SC and SC+Liver model mice. Typical photos and H&E staining slices of the livers were shown. The numbers of DCs, neutrophils and macrophages (b), and the percentage of Kupffer cells, Mono-M, Mono-M1, and Mono-M2 in macrophages and the ratio of Mono-M2/Mono-M1 (c) within the livers of the two types of mice models. d The effects of Cl-Lipo on hepatic macrophages of SC+Liver tumor models. e Schematic illustration of the impacts of liver CRC metastases on hepatic macrophages (indicated by blue arrows) and the effects of Cl-Lipo on hepatic macrophages (indicated by red arrows) in the livers of SC+Liver models. *P < 0.05, **P < 0.01.

Full size image

Given clodronate liposome (Cl-Lipo) a widely used reagent to deplete macrophages in vivo [31, 47], we further investigated the effects of Cl-Lipo on the subtypes of macrophages. Flow cytometry analysis revealed that Cl-Lipo solely depleted the Kupffer cells in the livers of SC+Liver models without significant effects on both the density and the composition of Mono-M (including Mono-M1 and Mono-M2) (Fig. 2d). These results revealed that the formation of CRC liver metastasis dramatically elevated the fraction of Mono-M2 in the liver, fostering an immune suppressive microenvironment, which could not be relieved by Cl-Lipo that mainly affected the Kupffer cells in the livers (Fig. 2e).

Mono-M2 preference of sHDL

Previous results above demonstrated that Mono-M2 might be the target of potential treatments against CRC with liver metastases. Our previous study found that sHDL preferentially targeted intratumoral Mono-M2 through a scavenge receptor class B type 1 (SR-B1)-dependent pathway [40]. Thus, we first investigated the levels of SR-B1 on subtypes of hepatic macrophages and tumor cells. Flow cytometry revealed that Mono-M expressed higher levels of SR-B1 than Kupffer cells and tumor cells in the livers (Fig. 3a), and within the Mono-M population, Mono-M2 showed higher SR-B1 expression than Mono-M1 (Fig. 3b).

Fig. 3: Preferential uptake of sHDL by Mono-M2.
figure 3

SR-B1 expression on tumor cells, Kupffer cells and Mono-M (a) and subtypes of Mono-M (b). c The TEM images of sHDLs and liposomes loaded with DiD or DiR. d Distribution of free DiR, R-Lipo and R-sHDL in mice with only liver tumors at 8 h after injection. e Typical immunofluorescence images and semi-quantification analysis of D-sHDL or D-Lipo accumulation in livers 24 h after injection. Uptake of D-sHDL or D-Lipo by Mono-M1 and Mono-M2 (f) and MC38 tumor cells (g) in vitro. h Uptake of D-sHDL or D-Lipo by hepatic macrophages and MC38 tumor cells in SC+Liver models at 24 h after injection. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Full size image

The above findings prompted us to test the Mono-M2 preference of sHDL. Fluorescence-labeled sHDLs (~12 nm in diameter, ~–20 mV in ζ-potential) were prepared, and liposomes (~100 nm in diameter, ~–20 mV in ζ-potential), as widely applied nanocarriers, were used as control formulations (Fig. 3c). After intravenous injection, fluorescence-based live animal imaging revealed a higher accumulation of sHDL in the livers with CRC metastases than the liposome and free dye, which was the most evident at 8 h post the injection and still discernable at day 5 (Fig. 3d and Supplementary Fig. S2). Further examination of the liver slices and the semi-quantification result confirmed that the sHDL was more efficient in liver-targeted cargo delivery than the liposome (Fig. 3e).

We then tested the cell preference of sHDL in vitro on BMDMs and MC38 tumor cells. Higher cellular uptake of sHDL than liposome was observed, and only the sHDL showed a significant Mono-M2 preference (Fig. 3f). In addition, sHDL was more efficient in delivering cargos to the tumor cells (Fig. 3g). Based on these findings, we further investigated the accumulation of intravenously injected sHDL in hepatic macrophages and tumor cells. Compared with liposome, sHDL showed higher accumulation in hepatic macrophages, Mono-M and Mono-M2 (Fig. 3h). More importantly, sHDL preferentially delivered cargos into Mono-M2 than Mono-M1, while the liposome showed no selectivity between the two subtypes. In addition, the sHDL also delivered more cargo to tumor cells in the liver metastases.

The effects of G-sHDL on hepatic immune microenvironment

After proving the Mono-M2 preference of sHDLs, we encapsulated gemcitabine into the sHDL to create G-sHDL (11.2 ± 2.3 nm in diameter, −18.3 ± 0.7 mV in ζ-potential, EE = 55.7% ± 2.1%, and DL = 6.6% ± 1.1%), and gemcitabine-loaded liposome (G-Lipo, 84.2 ± 8.6 nm in diameter, −21.2 ± 0.9 mV in ζ-potential, EE = 70.7% ± 2.5%, and DL = 10.4% ± 0.3%) was used as a control formulation (Fig. 4a). We tested the efficacy of G-sHDL in modulating the hepatic macrophages on SC+Liver models. Flow cytometry analysis demonstrated that all the treatments diminished the fraction of macrophages in the total hepatic immune cells (Fig. 4b). However, only the G-sHDL significantly reduced the fraction of Mono-M in the diseased livers (Fig. 4c). Further subtype analysis showed that only the G-sHDL significantly increased the density of hepatic Mono-M1 and decreased the density and fraction of Mono-M2 among all the tested formulations (Fig. 4d–f). As a result, the lowest Mono-M2/Mono-M1 ratio was recorded in the livers of G-sHDL-treated mice (Fig. 4g). Similar trend was recorded when the same set of formulations were tested on Mono-M1 and Mono-M2 in vitro, where the G-sHDL killed most Mono-M2 while exhibited mild toxicity to Mono-M1 (Fig. 4h and Supplementary Fig. S3). The apoptosis of hepatic CD8+ T cells was then examined, and we found that all the tested formulations decreased the proportion of cleaved caspse-3+CD8+ T cells and the G-sHDL was the most efficient (Fig. 4i and Supplementary Fig. S4). These results indicated that the lower rates of hepatic CD8+ T cell apoptosis were associated with the global reduction in the hepatic macrophages, which might be partially due to the involvement of hepatic macrophages in physiologically inactivation of T cells [48].

Fig. 4: G-sHDL reverses the immunosuppressive microenvironment in the livers and reduces the apoptosis of tumor antigen-specific CD8+ T cells.
figure 4

a TEM images of the G-sHDL and G-Lipo. The percentage of macrophages in CD45+ cells (b) and Mono-M in macrophages (c). The densities of Mono-M1 (d) and Mono-M2 (e) in the livers with CRC metastases. f The percentage of Mono-M2 in macrophages. g The ratio of Mono-M2/Mono-M1. h Cell viability of Mono-M1 and Mono-M2 after various treatments in vitro. i The proportion of apoptotic (cleaved caspase-3+) T cells in CD8+CD90+ T cells. j Schematic illustration of the experimental procedure of hepatic F4/80+ cell-mediated OT-1 cell apoptosis in vitro. k Flow cytometry analysis of apoptotic (7-AAD+Annexin V+) tumor antigen-specific (OT-1) cells after coculturing with hepatic F4/80+ cells and MC38-OVA cells for 48 h. l OT-1 cell numbers in different tissues of SC+Liver model mice bearing MC38-OVA tumors after different treatments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Full size image

It was reported that the Mono-M in the livers with liver metastases mainly induced eradication of the tumor antigen-specific CD8+ T cells [18]. To investigate whether the G-sHDL was the most efficient in protecting the tumor antigen-specific CD8+ T cells, we performed a 48 h co-culture study by incubating hepatic F4/80+ cells (harvested from the livers of mice bearing hepatic MC38-OVA tumors), OT-1 cells and MC38-OVA tumor cells (Fig. 4j). In consistence with the previous result (Fig. 4f), the fraction of Mono-M2 in the isolated F4/80+ cells was the lowest in mice receiving the G-sHDL (Supplementary Fig. S5). Flow cytometry analysis revealed that only the G-sHDL significantly reduced the apoptosis of OT-1 cells (Fig. 4k). We further injected CFSE-labeled OT-1 cells into the mice bearing both subcutaneous and liver MC38-OVA tumors. Flow cytometry data revealed that only the G-sHDL significantly increased the numbers of OT-1 cells in the blood, livers, TDLNs and the subcutaneous tumors (Fig. 4l), further confirming that the G-sHDL could protect against Mono-M2-mediated eradication of tumor antigen-specific CD8+ T cells.

Antitumor immunity induced by G-sHDL

The previous results prompted us to investigate the antitumor immunity induced by the G-sHDL. We first tested the capability of G-sHDL in inducing ICD of tumor cells in vitro, as the G-sHDL could effectively kill the MC38 tumor cells (Supplementary Fig. S6). We found that all the gemcitabine-based drugs (Gem, G-Lipo, and G-sHDL) induced cell surface display of calreticulin and the release of HMGB1 and ATP (Fig. 5a–c). Then the DC maturation was tested in vitro using a co-incubation method (Fig. 5d). Flow cytometry data revealed that all the gemcitabine-based formulations promoted the maturation of DCs, while G-sHDL significantly induced the secretion of proinflammatory cytokines such as IL12p40 and TNF-α in vitro (Fig. 5e–g and Supplementary Fig. S7). In consistence with the results in vitro, all the gemcitabine-based formulations enhanced the maturation of DCs in vivo, and the G-sHDL significantly increased the concentrations of IL12p40 and TNF-α within TDLNs 3 days after the treatments (Fig. 5h–j). These results demonstrated that the G-sHDL could also prime the antitumor immunity by triggering ICD of tumor cells and promoting DC maturation.

Fig. 5: Antitumor immunity of the G-sHDL.
figure 5

a MFI of CRT on MC38 tumor cells after different treatments. Concentrations of extracellular ATP (b) and HMGB1 (c) in the supernatant of treated MC38 cells. d Schematic illustration of DC maturation in vitro. e Percentage of mature DCs. Concentrations of IL12p40 (f) and TNF-α (g) in the supernatant of the DCs. Percentage of mature DCs (h), contents of IL12p40 (i), and TNF-α (j) within TDLNs three days after treatments. Percentage of Mono-M1 (k) and Mono-M2 (l) in macrophages and the ratio of Mono-M2/Mono-M1 (m) within SC tumors five days after treatments. Densities of CD8+ T cells (n) and CD8+IFN-γ+ T cells (o) in the SC tumors five days after treatments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Full size image

Given that G-sHDL could regulate hepatic macrophages, we next sought to examine the impact of G-sHDL on intratumoral monocyte-derived macrophages. We found that the G-sHDL significantly elevated the frequency of Mono-M1 and diminished the fraction of Mono-M2 in the subcutaneous tumors, showing the lowest Mono-M2/Mono-M1 ratio (Fig. 5k–m and Supplementary Fig. S8). G-sHDL also elevated the intratumoral densities of CD3+ T cells and CD8+ T cells (Fig. 5n and Supplementary Fig. S9), as more importantly the density of active (IFN-γ+) CD8+ T cells in the SC tumors (Fig. 5o). These data highlighted the capabilities of the G-sHDL in priming antitumor immunity and reversing the immunosuppressive microenvironment in the subcutaneous tumors.

Antitumor efficacy of the G-sHDL

Finally, we evaluated the antitumor efficacy induced by G-sHDL. Obvious inhibition of SC tumor growth (54.68%) was observed in SC+Liver model mice treated with G-sHDL after three doses, while all the other treatments did not show significant therapeutic effects (Fig. 6a). The mice treated with the G-sHDL had the lowest tumor burden and liver metastases growth compared with the rest of the treatments without obvious effect on the body weight of the mice (Fig. 6b and Supplementary Fig. S10). Further analysis of the serum creatinine and alanine transaminase (ALT), however, revealed that Cl-Lipo significantly elevated serum ALT (Fig. 6c, d). This finding was confirmed histologically, as slight structure deformation was observed in the livers of mice receiving Cl-Lipo (Fig. 6e and Supplementary Fig. S11). Based on these findings, we monitored the survival of animals after receiving different treatments. The G-sHDL showed a significant but mild effect (6 days) on prolonging the median survival of the mice without significantly affecting the body weights (Fig. 6f and Supplementary Fig. S12).

Fig. 6: Antitumor efficacy induced by sHDL.
figure 6

The SC tumor growth profiles (a) and tumor weights (b) of SC+Liver tumor models after different treatments (n = 6). Concentrations of creatinine (c) and ALT (d) in serum of tumor models after different treatments (n = 5 or 6). e Typical images of H&E staining slices of kidneys and livers collected from SC+Liver model mice receiving different treatments (scale bar = 50 µm). f Survival of SC+Liver tumor model mice after different treatments (n = 9). The SC tumor growth profiles (g) and tumor weights (h) of SC+Liver tumor model mice after various treatments (n = 6). i Typical bioluminescence images and semi-quantification analysis of liver tumors bioluminescence (n = 5). j Survival of SC+Liver tumor model mice after different treatments (n = 9). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Full size image

Since our G-sHDL mainly worked by protecting CD8+ T cells from apoptosis, we further tested the efficacy of combination therapy using G-sHDL and anti-PD-L1 antibody. As expected, combination therapy was the most effective treatment, retarding the growth of tumors by 67% (Fig. 6g). At the end of the experiment, the combination therapy reduced the tumor burden by 72% without significant effect on mice body weights (Fig. 6h and Supplementary Fig. S13). Based on these findings, we further evaluated the efficacy of combination therapy in the treatment of CRC liver metastases and prolonging animal survival. While the anti-PD-L1 antibody alone was not effective, its combination with the G-sHDL significantly inhibited the growth of CRC liver metastases (Fig. 6i and Supplementary Fig. S14). Further survival studies showed that the combination therapy prolonged the survival of animals by 16 days without obvious effect on mice body weights (Fig. 6j and Supplementary Fig. S15).

Discussion

Liver metastasis of CRC is the major cause of CRC-associated death because the metastatic tumor would not only cause malfunction of the liver but also impose a systemic immunosuppressive state failing the antitumor immunity [18, 49]. In this study, we found hepatic Mono-M2 the key cell subtype in shaping the hepatic and systemic suppressive immune states, and explored the use of G-sHDL in reversing the Mono-M2-mediated immunosuppressive state. The G-sHDL preferentially affected Mono-M2 due to their higher expression of SR-B1 receptor and susceptibility to gemcitabine. By eradicating the Mono-M2, the G-sHDL reduced the apoptosis of tumor antigen-specific CD8+ T cells in the livers and elevated their populations in the subcutaneous CRC tumors. The G-sHDL showed significant therapeutic efficacy against CRC tumors especially when used in conjunction with anti-PD-L1 antibody.

Given the potent efficacy of the G-sHDL in protecting tumor antigen-specific CD8+ T cells from Mono-M2-mediated eradication, the therapeutic efficacy was still far from satisfactory even when the G-sHDL was used with ICB therapy. Because of the low cytotoxicity (half maximal inhibitory concentration at 687 nM) of the G-sHDL, we believe the passable therapeutic effect of the G-sHDL may be associated with an insufficient eradication of the tumor cells and thus limited ICD. Therefore, a potential way to further improve the efficacy of the G-sHDL may be the introduction of other clinically approved local therapies such as radiotherapy and local ablation. These local therapies especially radiotherapy have been proven to exert robust immunostimulatory effects, and at the same time could significantly remodel the TIME for better infiltration and activation of effector cells [50, 51]. Another strategy could be combinatorial therapy using sHDL loaded with two drugs of different mechanism of actions, such as co-delivery of prodrugs of gemcitabine and oxaliplatin.

There are two limitations of the current work. First, in this proof-of-concept study, an SC+Liver tumor model of CRC is used to explore the subtypes of macrophages that are involved in systemic immunosuppression and the strategy to specific modulate the identified subtype. An orthotopic CRC model with spontaneous liver metastases is more clinic relevant. However, the low incidence rate of the orthotopic CRC tumors after local inoculation, difficulty in examining the incidence of liver metastases and the heterogeneity among the animals make the validation of our G-sHDL on this model extremely challenging. It will be interesting to test the G-sHDL on mice model with both liver metastases and orthotopic CRC, since the tumor microenvironments between subcutaneous and orthotopic CRC are very different. Second, our sHDL is currently prepared using film-hydration method, which is challenging in large-scale production. Industrial-compatible production techniques such as microfluidics may be explored for further sHDL preparation.

Nevertheless, our findings clearly showed that hepatic Mono-M2 modulation could be a promising strategy to overcome the systemic immunosuppression imposed by liver metastases and other diseases where abnormal Mono-M2 are involved. We envision that our G-sHDL is an effective treatment for chemoimmunotherapy of colorectal cancer with liver metastases, due to its biodegradability, biocompatibility, and receptor-mediated cell selectivity. Our study will facilitate the clinical translation of sHDL and inspire the design of other biomimetic carriers for cell-type specific drug delivery.