Article Text
Abstract
Background Limited activation and infiltration of CD8+ T cells are major challenges facing T cell-based immunotherapy for most solid tumors, of which the mechanism is multilayered and not yet fully understood.
Methods Levels of CD93 expression on monocytes from paired non-tumor, peritumor and tumor tissues of human hepatocellular carcinoma (HCC) were evaluated. The underlying mechanisms mediating effects of CD93+ monocytes on the inhibition and tumor exclusion of CD8+ T cells were studied through both in vitro and in vivo experiments.
Results In this study, we found that monocytes in the peritumoral tissues of HCC significantly increased levels of CD93 expression, and these CD93+ monocytes collocated with CD8+ T cells, whose density was much higher in peritumor than intratumor areas. In vitro experiments showed that glycolytic switch mediated tumor-induced CD93 upregulation in monocytes via the Erk signaling pathway. CD93 on the one hand could enhance PD-L1 expression through the AKT-GSK3β axis, while on the other hand inducing monocytes to produce versican, a type of matrix component which interacted with hyaluronan and collagens to inhibit CD8+ T cell migration. Consistently, levels of CD93+ monocytes positively correlated with the density of peritumoral CD8+ T cells while negatively correlated with that of intratumoral CD8+ T cells. Targeting CD93 on monocytes not only increased the infiltration and activation of CD8+ T cells but also enhanced tumor sensitivity to anti-PD-1 treatment in mice in vivo.
Conclusion This study identified an important mechanism contributing to the activation and limited infiltration of CD8+ T cells in solid tumors, and CD93+ monocytes might represent a plausible immunotherapeutic target for the treatment of HCC.
- Tumor microenvironment - TME
- Hepatocellular Carcinoma
- Immunotherapy
- Macrophage
- Monocyte
Data availability statement
Data are available in a public, open access repository. All data relevant to the study are included in the article or uploaded as online supplemental information. The Bulk RNA-seq data from the TCGA LIHC dataset were obtained from the Cancer Genome Atlas (TCGA) (https://www.cancer.gov/ccg/research/genome-sequencing/tcga). The immune infiltration data of TCGA LIHC were obtained from TIMER 2.0 (http://timer.cistrome.org/). Some scRNA-seq data were downloaded from Tumor Immune Single-cell Hub (http://tisch.comp-genomics.org/), other scRNA-seq data (BioProject ID PRjCA007744) were downloaded from the Genome Sequence Archive at the National Genomics Data Center (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA001749). The Bulk RNA-seq data of siNC and siCD93 monocyte with accession number GSE276062 have been deposited at the NCBI’s Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/).The other data generated in this study are included in the article or uploaded as online supplemental materials.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Immune checkpoint blockade (ICB) therapy has a low response rate in the treatment of solid tumors such as hepatocellular carcinoma (HCC).
WHAT THIS STUDY ADDS
Glycolysis-induced CD93 expression on peritumoral monocytes inhibits the activation and tumor infiltration of CD8+ T cells by inducing PD-L1 expression and versican production, respectively, in HCC.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
CD93+ monocytes might be useful both as a marker in stratifying patients for ICB therapy and as an attractive therapeutic target in the treatment of HCC.
Introduction
Responses to T cell-based immunotherapy, such as immune checkpoint blockade (ICB), have been limited in solid tumors, and level of effector T cells infiltration in tumor tissue is an important factor determining the efficacy of immunotherapy.1–3 There are multiple mechanisms tumor might manipulate to limit T cell accumulations: reduced chemokines, downregulated adhesion molecules for immune cell extravasation, adverse metabolic microenvironment for the survival and functions of effector T cells, dense extracellular matrix to create physical barriers for effective T cell penetration, etc.4–9 It might be one of these factors, or a combination of several ones, that determined the final outcome of effector T cell levels in solid tumor tissues, and deciphering mechanisms regulating the T cell trafficking process would yield new targets for single or meaningful combinational therapies for the treatment of solid tumors.
Monocytes/macrophages are versatile cells with diverse functions.10 They accumulated in tumor tissues and processed both protumor and antitumor functions depending on local environmental signals.11–14 Our previous study found that monocytes in peritumoral tissues of human hepatocellular carcinoma (HCC) significantly upregulated PD-L1 expression to induce CD8+ T cell anergy, and targeting these PD-L1-expressing monocytes suppressed tumor progression in mice models.15 Although these two types of cells collocated and were highly accumulated in peritumoral tissues of HCC,15 the possible role of monocytes in influencing the further infiltration of CD8+ T cells into intratumoral areas has not yet been explored.
CD93 is a C-type lectin transmembrane protein primarily expressed by endothelial cells (ECs) and plays important roles in vascular angiogenesis.16 17 Recent research showed that CD93 blockade can be a feasible approach for vascular normalization to facilitate cancer therapies.18 However, besides intratumoral ECs, our study found that monocytes in the peritumoral tissues of human HCC significantly upregulated CD93 expression compared with those within non-tumor and intratumor tissues. In vitro experiments showed that glycolysis-activated Erk signaling pathway upregulated CD93 expression on monocytes. CD93 could induce monocyte expression of PD-L1 and their production of an extracellular matrix component versican, thus not only inhibiting the activation of CD8+ T cells, but also remodeling the extracellular matrix to possibly inhibit the efficient trafficking of CD8+ T cells into tumor cores. Therefore, our results suggest that blockade of the CD93 pathway in monocytes represents a therapeutic strategy which reshapes tissue microenvironment to enhance T cells infiltration and antitumor immunity.
Materials and methods
Patients and specimens
Liver tissues were obtained from 138 untreated patients with pathologically confirmed HCC from the Cancer Center of Sun Yat-sen University and Sun Yat-sen Memorial Hospital between 2013 and 2023. Patients with concurrent autoimmune disease, HIV, or syphilis were excluded. Among these patients, 99 (cohort 1) who had complete follow-up data were used for immunohistochemical analysis and assessments of overall survival (OS) and tumor recurrence (TR). Another 39 (cohort 2) were used for the isolation of tumor-infiltrating, peritumor-infiltrating and non-tumor-infiltrating leucocytes. Non-tumor sites were defined as areas at least 3 cm away from the tumor sites. The clinical characteristics of all patients are summarized in online supplemental table S1. Blood samples were obtained from 55 healthy donors attending the Guangzhou Blood Center. Heparin tubes (367884, BD Biosciences, USA) were used to collect blood samples, and all blood assays were performed using fresh cells.
Supplemental material
Isolation of leucocytes from peripheral blood or tissues
Human peripheral blood samples from healthy donors were separated by ficoll density-gradient centrifugation. The leucocyte layer was collected to isolate CD14+ monocytes or CD8+ T cells using magnetic beads (130-050-201 and 130-045-201, Miltenyi Biotec, Bergisch Gladbach, Germany).
Tumor-infiltrating, peritumor-infiltrating, and non-tumor-infiltrating leucocytes were obtained from paired fresh tissue samples (online supplemental figure S1).19 Briefly, Fresh HCC biopsy specimens were cut into tiny pieces and digested in RPMI 1640 (C11875500BT, Thermo Fisher Scientific, USA) supplemented with 0.002% DNase I (DN25, Sigma-Aldrich, St. Louis, USA), 0.05% Collagenase IV (C5138, Sigma-Aldrich, St. Louis, USA), and 20% FBS (Fetal Bovine Serum) (10099-141, Gibco, NYC, USA) at 37℃ for 40 min. Dissociated cells were separated by ficoll density-gradient centrifugation. CD14+ leucocytes were isolated as described above.
For isolation of tumor-infiltrating leucocytes from mice hepatic orthotopic tumors, fresh tumors were cut into small pieces and digested in RPMI 1640 supplemented with 0.002% DNase I, 0.05% Collagenase IV, 50 µg/mL hyaluronidase (H1136, Sigma-Aldrich, St. Louis, USA), 30 µg/mL Collagenases XI (C7657, Sigma-Aldrich, St. Louis, USA), and 20% FBS at 37℃ for 40 min. Dissociated cells were separated by ficoll density-gradient centrifugation.
The purified cells were then used for direct analysis, in vitro experiments, or collection of culture supernatants.
Preparation of culture supernatants from cell lines or monocytes
Tumor cell lines (HepG2) were purchased from American Type Culture Collection. All cells were tested for mycoplasma contamination using the single-step PCR method and were maintained in DMEM (Dulbecco's Modified Eagle Medium) (C11995500BT, Thermo Fisher Scientific, USA) medium supplemented with 10% FBS. Cell supernatant was prepared by plating 5×106 tumor cells in 10 mL complete DMEM medium in 100 mm dishes for 24 hours, and thereafter changing the medium to complete DMEM medium supplemented with 10% human AB serum instead of FBS. After 2 days, the supernatants were harvested, centrifuged, and stored in aliquots at −80℃.
CD14+ cells were purified from the peripheral blood of healthy donors as described above. These cells were left untreated or treated with HepG2 hepatoma supernatants (TSN) for 24 hours, washed, and cultured for another 48 hours before their supernatants (designated as CCM and TCM, respectively) were harvested, centrifuged, and stored at −80℃ before use.
In vitro culture of monocytes
Purified CD14+ cells were cultured in DMEM with 10% human AB serum in the presence or absence of 30% HepG2 TSN. In some experiments, CD14+ cells were left untreated or treated with TSN in the presence or absence of 2DG (20 mM, D8375, Sigma-Aldrich, St. Louis, USA), 3PO (20 µM, SML1343, Sigma-Aldrich St. Louis, USA), JSH-23 (NF-κB inhibitor, 10 µM, 481408, Merk Millipore, USA), AG490 (STAT3 inhibitor, 10 µM, HY-12000, MCE, USA), SP600125 (JNK inhibitor, 10 µM, HY-12041, MCE, USA), U0126 (Erk inhibitor, 25 µM, HY-12031A, MCE, USA), SB 203580 (p38 inhibitor, 25 µM, HY-10256, MCE, USA), MK-2206 (AKT inhibitor, 5 µM, S1078, Selleck, USA), or LiCl (20 mM, L9650, Sigma-Aldrich, USA), for indicated times. In some experiments, monocytes were transfected with control siRNA, siCD93, or siVCAN, (GenePharma, China), before being exposed to medium or HepG2 TSN.
Quantitative real-time PCR
Total RNA was extracted using TRIzol reagent (AM9738, Invitrogen, USA) and then used to synthesize cDNA with 5X All-In-One RT MasterMix (G492, abm). Sequences of the primers used for PCR analysis were listed in online supplemental table S2. Quantitative PCR was performed according to a standard protocol using the SYBR Green Real-Time PCR Mix (QPS-201, TOYOBO, Japan) in LightCycler 480 System (Roche, Basel, Switzerland). To determine the relative fold change of different genes, their levels of expression were normalized to those of β-actin.
Immunofluorescence staining
For immunofluorescence (IF) analysis of patient samples, frozen sections were processed as described previously.20 Sections were incubated with primary antibodies against human CD68 (M0876, DakoCytomation, Denmark), CD93 (HPA009300, Atlas antibodies, USA), PD-L1(ab228462, Abcam, UK), CD8 (ZA-0508, Zsbio, China), Versican (ab177480, Abcam, UK), α-SMA (ZM-0003, Zsbio, China), CD31 (ZM-0044, Zsbio, China) or antibodies against mouse Ly6C (ab15627, Abcam, UK), Versican (ab177480, Abcam, UK), F4/80 (70076, Cell Signaling Technology, USA.), CD93 (PK40533, Abmart, China), CD31 (77699, Cell Signaling Technology, USA), PD-L1 (13684, Cell Signaling Technology, USA). IF signals were amplificated by a tyramide signal amplification kit (PPK007100100, Panovue) for visualization. Nuclei were counterstained with DAPI (10236276001, Roche).
IF staining images were visualized by the ZEISS microscope (LSM780, Germany). The intensity of protein expression was evaluated using Zeiss ZEN software (LSM780, Germany), and positive cells were quantified using ImagePro Plus software (Media Cybernetics, Maryland, USA). Results were expressed as mean±SEM in high-powered fields.
Immunoblotting
Immunoblotting was performed as described previously.21 Primary antibodies used included: anti-human CD93 (HPA009300, Atlas antibodies, USA), PD-L1, phospho-FAK, FAK, phospho-p38, p38, phospho-AKT, AKT, phospho-GSK3β, GSK3β, phosphor-Erk, Erk, Versican (ab177480, Abcam, UK), anti-human β-actin (BM0627, Boster, USA). HRP-linked goat anti-rabbit/mouse IgG antibodies (7074/7076) were purchased from Cell Signaling Technology.
Flow cytometry
Cell surface markers were determined by direct staining with anti-human CD14-AF700 (557923, BD Biosciences, USA), anti-human CD93-APC (336119, Biolegend, USA), anti-human HLA-DR-BV421 (562804, BD Biosciences, USA), anti-human CD86-APC (555660, BD Biosciences, USA), anti-human CD204-PC7 (371908, Biolegend, USA), anti-human PD-L1-PC7 (558017, BD Biosciences, USA), anti-mouse CD45-BV570 (103136, Biolegend, U.S.), anti-mouse CD11b- AF700 (101222, Biolegend, U.S.), anti-mouse Ly6C-BV421 (128031, Biolegend, USA), anti-mouse Ly6G-PE-cf594(562700, BD Biosciences, USA.) anti-mouse CD93-APC (136509, Biolegend, USA), anti-mouse CD3-PE (100206, Biolegend, USA), or anti-mouse CD8-ef450 (48-0081-82, eBioscience, USA) antibodies. In some experiments, to measure intracellular IFN-γ and TNF-α production, lymphocytes were cultured at 37℃ for 12 hours, treated with Leukocyte Activation Cocktail (550583, BD Sciences, USA) for 4 hours, stained with surface markers, fixed, permeabilized with IntraPre Reagent (A07803, Beckman), and further stained with anti-mouse IFN-γ-FITC (53-7311-82, Thermo Fisher Scientific) or anti-mouse TNF-α-PE-cf594 (506346, Biolegend). Data were acquired with CytoFLEXS flow cytometer (Beckman Coulter, Brea, USA) and evaluated with FlowJo software version V.10 (Tree Star, Ashland, USA).
Immunohistochemistry staining
Paraffin-embedded and formalin-fixed samples were cut into 5 µm sections, followed by procedures for multiplex immunohistochemistry (IHC) as previously described.21 Briefly, after incubation with primary antibody, sections were stained with corresponding secondary antibody and visualized with 3-amino-9-ethylcarbazole in an Envision System. Then the sections were washed, incubated with another primary antibody, and the procedure went on for several rounds. All the images were then overlaid via software (ImageJ and Photoshop). Primary antibodies used included: anti-human human CD68 (M0876, DakoCytomation, Denmark), anti-human CD93 (HPA009300, Atlas antibodies, USA), and anti-human CD8 (ZA-0508, Zsbio, China).
Evaluation of immunohistochemical variables
Analysis was performed by two independent observers who were blinded to the clinical outcome. At low-power field (×100), the tissue sections were screened, and the five most representative fields were selected using an Eclipse Ni-U highly versatile upright microscope combining system (Nikon Instruments, Japan). For evaluating the density of tissue-infiltrating CD68+, CD93+CD68+, CD8+ cells, or CD93+ cells, the respective areas of HCC tissues were then scanned at ×400 magnification (0.146 mm2 per field). The number of nucleated cells was then counted manually and expressed as cells per field.
RNA interference
Purified monocytes were left untreated or transfected with 300 nM control siRNA, siCD93 or siVCAN using P3 primary cell 4D-Nucleofector X kit (V4XP-3024, Lonza, Switzerland) with Lonza 4D Nucleofector (Lonza). All siRNA duplexes were purchased from GenePharma, China, and their sequences are listed in online supplemental table S2.
ELISA
Cytokine concentrations were detected by ELISA kits according to the manufacturer’s instructions (TNF-α, 88-7346-86, eBioscience, USA; IL-6, 88-7066-88, eBioscience, USA; IL-1β, 88-7261-88, eBioscience, USA).
Cell migration assay
Migration assays for CD8+ T cells purified from healthy peripheral blood were performed using the 24-well Boyden chamber with 8 µm polycarbonate membrane (3422, Corning, USA). Briefly, the membrane was coated with collagen I (354236, Corning, USA) plus hyaluronan (GLR004, R&D Systems, USA), in the presence or absence of Versican (230-00833, RayBiotech, USA), or the CCM or TCM from siNC or siVCAN -treated monocytes. CD8+ T cells (2.5×105) in 100 µL of serum-free DMEM were added to the upper compartment of the chamber, while the lower compartment was filled with 600 µL of DMEM containing 10% FBS. After 24 hours of incubation, supernatants from the basal chamber were collected and the number of CD8+ T cells was manually counted.
Animals
Wild-type male C57BL/6 mice were purchased from Guangdong Medical Laboratory Animal Center (Guangzhou, China). All mice were maintained under specific pathogen-free conditions and were used between 6 and 8 weeks of age in accordance with the experimental animal guidelines set by the Institutional Animal Care and Use Committee of Sun Yat-sen University Cancer Center (Guangzhou, China, ethics approval number: SYSU-IACUC-2023-B0401).
Mice tumor models and treatments
Establishment of the orthotopic hepatic tumor model: 1×105 Luminescence-Hepa1-6 cells were suspended in 25 µL of 66.7% basement membrane extract (3432-005-01, R&D Systems, USA), and intrahepatically injected into the left lobe of livers of anesthetized 6-week-old C57BL/6 mice. Orthotopic tumor growth was monitored with a Xenogen in vivo imaging system (IVIS, PerkinElmer).
Medium, siNC-containing liposome, or siCD93-containing liposome (C12-200, HY-145405, MCE; cholesterol, C8503, Sigma-Aldrich; DSPC, 850 365P, Avanti; DMG-PEG2000, 880 151P, Avanti) were prepared as previously described.19 They were administered into the mice three times through tail vein—every 3 days from day 7. In some experiments, siNC-containing liposome, or siCD93-containing liposome in combination with control IgG (BE0089, BioXCell, USA), or 25 µg anti-mouse PD-1 antibody (BE0146, BioXcell, USA), were intraperitoneally injected three times—every 3 days from day 7. Tumor-bearing liver weight of different treatment groups was measured at day 16.
RNA-seq data analysis
PBMC-derived CD14+ monocytes were transfected with siNC or siCD93 before being treated with HepG2 TSN for 24 hours. RNA of the cells was extracted by TRIzol method, and RNA-sequencing was processed on a DNBSEQ-T7 platform. The gene matrix files were generated with RNA express app (DNBSEQ) and analyzed using T-test comparison and log2 ratio of classes. Differential expression analysis was performed by DESeq2 and the false recovery rate <0.05 was considered to be significantly differentially expressed. The Bulk RNA-seq data are available from the NCBI’s Gene Expression Omnibus (GEO) database (GEO GSE276062).22
The mRNA expression level of VCAN and the immune infiltration data of LIHC were obtained from the Cancer Genome Atlas (TCGA)23 and TIMER 2.0,24 respectively, then the correlation between the expression level of VCAN and the infiltration level of CD8+ T cells was analyzed.
The scRNA-seq data were downloaded from Tumor Immune Single-cell Hub,25 CD93 and VCAN expression by different cells in HCC were analyzed. Downloaded the scRNA-seq data (BioProject ID PRjCA007744)26 from the Genome Sequence Archive at the National Genomics Data Center (Beijing, China), then performed the pseudobulk differential expression analysis of CD93+ versus CD93− macrophages in HCC.
Statistical analysis
Statistical tests used are indicated in the figure legends. The results are expressed as the mean±SEM. Correlations between parameters were measured by Pearson correlation. Statistical analysis was performed with GraphPad Prism V.8.0 (GraphPad Software, La Jolla, USA). Survival curves were calculated by the Kaplan-Meier method and analyzed by the log-rank test. The Cox proportional hazards model was used to identify prognostic factors through univariate and multivariate analyses. The p values were assessed using two-tailed paired/unpaired Student’s t-test or two-way analysis of variance with the following thresholds for statistical significance: *p<0.05; **p<0.01; ***p<0.001 and ****p<0.0001; ns, no significance.
Further details of materials were provided in online supplemental materials.
Results
CD93+ monocytes accumulate in the peritumor of human HCC and indicate worse patient survival
We set out to analyze the distribution of cells expressing CD93, a key molecule implicated in vasculature normalization, within human HCC tissues. Interestingly, besides intratumoral ECs, a significant number of monocytes in peritumoral regions increased their levels of CD93 expression, compared with those in paired intratumor or non-tumor areas (figure 1A, n=10; CD93+CD68+ cells/CD68+ cells: non-tumor, 5.26%±1.03%; peritumor, 10.91%±2.11%; intratumor, 3.56%±0.86%). Of note, these CD93+CD68+ cells were negative for CD31 staining, excluding the possibility of monocytes conjugating with CD93+ ECs (figure 1A). The increase of CD93 in/on peritumoral monocytes was confirmed through western blotting and flow cytometry analysis, and also on the mRNA level via Q-PCR (figure 1B–D). Consistently, analysis of scRNA-seq data from an online database showed that monocytes/macrophages highly expressed CD93, with levels only lower than those of ECs, in human HCC tumors (figure 1E).
To explore the potential role of CD93 expression by monocytes in disease progression, we divided HCC patients who had received curative resection with follow-up data into two groups according to the median value of their CD68+ or CD93+CD68+ cell density in peritumoral tissues. As shown in figure 1F,G, both high levels of CD68+ and CD93+CD68+ cell infiltration in peritumoral regions indicated worse patient survival, with CD93+CD68+ exhibiting relatively better prognostic values than the CD68+ groups (CD68+: n=99, p=0.0425 for OS, p=0.0488 for recurrence (TR); CD93+CD68+: n=99, p=0.0022 for OS, p=0.0155 for TR). Moreover, the density of CD93+CD68+ cells at peritumoral tissues could serve as an independent prognostic factor for both the OS (p=0.010) and TR (p=0.026) of HCC patients (online supplemental table S5). The above results suggested that the selective and significant upregulation of CD93 on peritumoral monocytes might facilitate the disease progression of human HCC.
Glycolytic switch induces CD93 upregulation on monocytes via the Erk signaling pathway
Our previous study found that peritumoral monocytes significantly upregulated glycolytic activity to support their protumor functions,20 27 indicating a possible link between the metabolic switch and CD93 expression of these cells. To evaluate this hypothesis, we first established an in vitro experimental model using CD14+ cells purified from the peripheral blood of healthy donors which were treated with HepG2 hepatoma culture supernatants (TSN). TSN effectively induced the upregulation of both mRNA and protein levels of CD93 expression in CD14+ cells in comparison to medium control (figure 2A–C), but the increase of CD93 was transient, with a peak on protein level at about 24 hours after TSN treatment, a kinetic that was consistent with and possibly explained the preferential spatial distribution of CD93+ monocytes in the peritumoral tissues, through which substantial amounts of fresh monocytes migrated into the intratumoral regions.
The mechanism regulating CD93 upregulation was then explored via the in vitro TSN-treated monocytes model. As shown in figure 2D,E, 2DG and 3PO (an inhibitor for a key glycolytic enzyme PFKFB3) could both effectively abrogate the TSN-induced CD93 increase by CD14+ monocytes, on the mRNA as well as protein levels. Our previous study found that the NF-κB, mitogen-activated protein kinase, and JAK-STAT signaling pathways were preferentially activated in TSN-exposed monocytes.20 By targeting these pathways with their respective inhibitors, we found that only Erk inhibitor could attenuate the TSN-induced CD93 upregulation in monocytes (figure 2F), and the effects of Erk inhibition on CD93 expression were further confirmed through flow cytometry analysis (figure 2G). Notably, while TSN treatment increased p-Erk levels in relation to total Erk compared with control, 2DG substantially attenuated such an increase (figure 2H). Moreover, levels of PFKFB3 and CD93 were positively correlated in monocytes purified from fresh HCC tumors (online supplemental figure S2). These data suggested that glycolysis-mediated Erk signaling activation induced the expression of CD93 by tumor-associated monocytes.
CD93 mediates tumor-induced PD-L1 expression on monocytes
Since peritumoral monocytes in human HCC usually exhibit both immune activation and suppression phenotype simultaneously, we transfected tumor-exposed monocytes with siNC or siCD93 (figure 3A) to evaluate the possible functions of CD93 on these cells. As shown in figure 3B,C and online supplemental figure S3, compared with siNC, siCD93 did not affect the TSN-induced upregulation of cytokine production (IL-6, IL-1β, TNF-α) and surface marker expression (HLA-DR, CD86, CD204) by/on CD14+ monocytes from healthy PBMCs. However, while TSN increased monocytes expression of PD-L1, as measured by flow cytometry and western blotting, siCD93 could significantly attenuate such increase in comparison with control groups (figure 3D,E). Consistently, CD93+ monocytes in the peritumor of HCC were found to be also positively stained with anti-human PD-L1 antibodies (figure 3F), and levels of CD93 and PD-L1 expression were positively correlated in monocytes purified from the peritumoral tissues of HCC (figure 3G, n=9; p=0.0011; r=0.8024). Interestingly, notwithstanding changes on protein levels, siCD93 did not impact the TSN-induced PD-L1 mRNA upregulation in monocytes compared with control (online supplemental figure S4), indicating that CD93 might regulate PD-L1 accumulation in monocytes via post-transcriptional mechanisms.
CD93 increases levels of PD-L1 through the AKT-GSK3β signaling pathway
It has been reported that CD93 can activate the FAK or p38-AKT signaling pathways in non-immune cells.28 29 Our results showed that while FAK and p-FAK exhibited no significant difference between TSN-treated and TSN-untreated CD14+ monocytes, levels of both p-p38 and p-AKT expression, in relation to total p38 and AKT respectively, were substantially upregulated in TSN-exposed monocytes compared with controls (figure 4A and online supplemental figure S5A). The upregulation of p-p38 and p-AKT could be attenuated by the treatment of cells with siCD93 (figure 4A and online supplemental figure S5A). Moreover, GSK3β, a reported downstream target of p-AKT and an enzyme which had been implicated in accelerating the degradation of PD-L1 by phosphorylating this molecule,30 31 has markedly increased its phosphorylation in relation to total protein levels—indicating a decrease of enzymatic activity—in TSN-treated monocytes, a change that could be reversed by treating these cells with siCD93 (figure 4A and online supplemental figure S5A).
To evaluate roles of p-AKT and GSK3β in regulating CD93-induced PD-L1 expression, we treated CD14+ monocytes with TSN in the presence or absence of a p-AKT inhibitor or a GSK3β inhibitor (LiCl, a substance which could increase the phosphorylation of GSK3β and reduce the enzymatic activity of GSK3β). As shown in figure 4B,C and online supplemental figure S5B, p-AKT inhibitor could significantly abrogate the TSN-induced p-GSK3β and PD-L1 upregulation in monocytes, while conversely, LiCl could enhance the phosphorylation of GSK3β in TSN-treated monocytes, and further increase the expression of PD-L1 by these cells compared with those treated with TSN alone (figure 4D,E and online supplemental figure S5C). Supporting these data, levels of PD-L1 expression were found positively correlated with those of both p-AKT and p-GSK3β in monocytes purified from HCC tissues (figure 4F,G; PD-L1 and p-AKT: n=9; p=0.0399; r=0.6893; PD-L1 and p-GSK3β: n=9; p=0.0406; r=0.6877).
The above results suggested that CD93 might trigger the p38-AKT pathway to phosphorylate and deactivate GSK3β, thus possibly reducing the phosphorylation of PD-L1 to stabilize and increase its accumulation in monocytes.
CD93 enhances the production of versican by monocytes
To explore other possible new targets of CD93 activation and explain the preferential accumulation of CD93+ monocytes in the peritumoral tissues of human HCC, we performed RNA-seq for healthy PBMC-purified CD14+ cells which were treated with hepatoma TSN and transfected with siNC or siCD93. As shown in figure 5A, of the most downregulated genes in siCD93-treated versus siNC-treated monocytes, the first was VCAN, whose product versican was an important ECM (Extra Cellular Matrix) component. Using online HCC scRNA-seq data, we identified 145 genes upregulated in CD93+ versus CD93− macrophages. When these genes were compared with the downregulated genes in siCD93-treated versus siNC-treated monocytes from our RNA-seq data, VCAN was among the only two overlapped genes (figure 5B). In consistent, Q-PCR and western blotting showed that levels of VCAN expression were higher in peritumor-purified monocytes than those in paired non-tumor and intratumor-derived monocytes (figure 5C,D), and levels of VCAN and CD93 expression were positively correlated in peritumoral monocytes (figure 5E, n=10; p=0.045; r=0.643). Moreover, similar to the kinetics of CD93 expression, PBMC-purified CD14+ cells transiently upregulated the mRNA levels of VCAN after the treatment of TSN, with a peak at about 48 hours, compared with control (figure 5F), and siCD93 could significantly attenuate the TSN-induced versican upregulation, both on mRNA and on protein levels, in monocytes in vitro (figure 5G,H).
Monocytes-derived versican suppresses the migration of CD8+ T cells
To put the above results into perspective, we analyzed VCAN gene expression by different cells in human HCC using online scRNA-seq data. As shown in figure 6A, monocytes/macrophages and fibroblasts constituted two major types of VCAN-expressing cells in HCC. Through IHC staining of HCC tissue sections with anti-human versican, CD68, α-SMA, and CD8 antibodies, we found that compared with intratumoral tissues, levels of versican expression were preferentially higher in the peritumoral regions of HCC, where CD68+ monocytes/macrophages and α-SMA+ fibroblasts were highly accumulated (figure 6B). Interestingly, the microlocations of CD68+ and α-SMA+ cells within the peritumoral regions are different, if not totally complementary to each other, implying that these two types of cells might non-redundantly contribute to the local accumulation of versican (figure 6B). At the same time, many CD8+ T lymphocytes were also found located in the same regions as CD68+ cells, α-SMA+ cells, and versican (figure 6B). By correlation analysis, we found that levels of versican expression were negatively associated with the intratumoral infiltration levels of CD8+ T cells of human HCC (figure 6C, online database; p=0.047; r=−0.1032), and notably, levels of CD93+ monocytes were positively correlated with the density of peritumoral CD8+ T cells while negatively correlated with that of intratumoral CD8+ T cells in HCC (figure 6D; left: p=0.0442, n=13; right: p=0.0251, n=13), indicating that CD93+ monocytes might retain CD8+ T cells within the peritumor, thus inhibiting their infiltration into the intratumoral tissues.
To confirm this hypothesis, we established an in vitro culture model in which the membrane of a transwell was coated with collagen I plus hyaluronan, or collagen I plus hyaluronan plus versican, and CD8+ T lymphocytes from the PBMC of healthy donors were added to the upper chamber to observe their transmigration into the lower chamber through the coated membrane. As shown in figure 6E, compared with collagen I and hyaluronan alone, a combination of these two ECM components with versican could significantly reduce the number of transmembrane migrating CD8+ T cells. Similar to the effects of versican, supernatants from TSN-pretreated monocytes, in combination with ECM components, could inhibit CD8+ T cells transmigration in comparison to the supernatants from control monocytes together with two ECM components, and such reduction in cell migration could be substantially reversed when monocytes were treated with siVCAN (figure 6F and online supplemental figure S6), suggesting that tumor-induced versican production by monocytes might interact with other ECM components, possibly condensing the ECM structure, thus reducing the penetration and migration of CD8+ T cells.
Targeting CD93+ monocytes inhibits tumor progression in mice in vivo
Given the observed effects of monocyte-expressing CD93 on the production of versican, and versican on the CD8+ T cell migration, we hypothesized that CD93 might represent an exploitable target for the treatment of HCC. To confirm this hypothesis, we established a Hepa1-6 orthotopic hepatic tumor model using wild-type C57BL/6 mice (figure 7A). These tumor-bearing mice were treated with siNC-containing liposome, or siCD93-containing liposome, which preferentially targeted CD93 expression on monocytes/macrophages instead of ECs (online supplemental figure S7). As shown in figure 7B,C and online supplemental figure S8A, compared with control or siNC, siCD93 liposome significantly reduced tumor growth within mice. Accordingly, decreased levels of tumor-infiltrating CD93+ (online supplemental figure S8B), PD-L1+ (online supplemental figure S8C,D), and versican+ (figure 7D and online supplemental figure S8E) monocytes/macrophages were observed in siCD93 liposome-treated mice, and both the infiltration and levels of IFN-γ/TNF-α expression of/by CD8+ T cells were markedly increased in siCD93 liposome-treated tumors compared with the untreated or siNC-treated controls (figure 7E–G).
Since monocytes/macrophages were not the only cells expressing PD-L1 within tumor microenvironments, siCD93 might only partially reverse the PD-L1/PD-1 signal-mediated CD8+ T cell anergy. Therefore, we went on to evaluate the possible therapeutic effects of combining siCD93 and anti-PD-1 antibodies in treating HCC (figure 7H). While both siCD93-containing liposome and anti-PD-1 antibodies could inhibit tumor growth (figure 7I,J and online supplemental figure S9) and increase levels of IFN-γ/TNF-α expression by CD8+ T cells (figure 7L,M), only the siCD93 treatment could significantly upregulate tumor infiltration of CD8+ T cells (figure 7K). Moreover, a combination of siCD93 and anti-PD-1 antibodies exhibited more significant effects in prohibiting tumor growth and enhancing CD8+ T cell activation than either treatment alone (figure 7I–M and online supplemental figure S9), confirming a synergistic effect between these two agents. Therefore, CD93 on monocytes might constitute a plausible therapeutic target for the treatment of HCC.
Discussion
Dense ECM is a key barrier to the tumor penetration of CTLs (Cytotoxic T Lymphocytes), which greatly compromises T cell-dependent immunotherapy of solid tumors.7 32–36 Peritumor is a special region between the non-tumor and intratumoral tissues of HCC, with high levels of immune infiltrates.21 37 Whether the physical structure or ECM in peritumor contributes to the retainment of immune cells in this region remains elusive. The current study provided evidence that glycolysis-mediated CD93 upregulation on monocytes in peritumoral tissue could increase the expression of PD-L1 and production of ECM component versican by these cells, thus compromising CD8+ T cell activation and probably enhancing the barrier for T cell penetration into tumor tissues via the interaction of versican with other ECM components.
CD93 (also known as C1qR1 or C1qRp) is a transmembrane glycoprotein that is highly expressed by ECs and regulates vascular angiogenesis.16–18 Interestingly, our results showed that besides intratumoral ECs, monocytes in the peritumoral regions of human HCC samples significantly enhanced their expression of CD93. While CD93 has been reported to be able to function in its soluble form, we found that peritumoral monocytes mainly upregulated their membrane-bound CD93. Documented ligands for membrane CD93 include insulin-like growth factor binding protein 7 (IGFBP7), Multimerin-2 (MMRN2), and IL-17D.16 17 29 Although we did not identify which one, or ones, or whether some factors other than the above-mentioned were responsible for the tumor-induced functions of CD93 on monocytes, our data implied that the ligands might be ubiquitous in tumor supernatants, since modifying CD93 itself could lead to the observed functional changes of monocytes.
Our previous study found that glycolysis switch in peritumoral monocytes of human HCC could upregulate their expression of PD-L1 via the NF-κB-TNF-α pathway.20 The current study complemented the previous findings by showing that Erk-CD93 also participated in the process of glycolysis-induced PD-L1 expression on monocytes. However, these two pathways seemed to be independent given that (1) siCD93 did not impact cytokines production by monocytes; (2) while NF-κB-TNF-α pathway enhanced the transcription and thus expression of PD-L1 on cells, siCD93 did not influence the mRNA levels of PD-L1. We further identified AKT/GSK3β as a key signaling pathway mediating the CD93-induced PD-L1 upregulation on monocytes through three sets of evidences: (1) siCD93 substantially abrogated TSN-induced increase of p-AKT and p-GSK3β expression in monocytes compared with control groups; (2) Inhibitor for p-AKT, and LiCl (which increased the phosphorylation of GSK3β and inhibited the activity of GSK3β), could downregulate and upregulate levels of PD-L1 expression, respectively, in TSN-treated monocytes; (3) Levels of PD-L1 were found positively correlated with those of both p-AKT and p-GSK3β in monocytes purified from human tumor tissues. We hypothesized that p-AKT phosphorylated and thus inhibited the activity of GSK3β, a kinase that might otherwise phosphorylate PD-L1 and facilitate its degradation. Therefore, CD93 may increase the stability rather than the transcription of PD-L1 in monocytes. Further experiments are needed to validate this hypothesis in the future.
The present data also found that CD93 could induce the production of versican by peritumoral monocytes. Versican is an evolutionary conserved ECM proteoglycan that can interact with other ECM components such as hyaluronan and collagens to regulate cell sorting, proliferation, migration, and survival.38–40 Although versican is ubiquitous within solid tissues,41 its distribution might vary among different spaces. For example, our results showed that levels of versican were markedly higher in the peritumoral regions of HCC compared with other areas. Interestingly, both fibroblasts (another reported source of versican) and monocytes were highly accumulated in the peritumoral regions, but with different microdistributions, suggesting that these two types of cells might non-redundantly contribute to the local production of versican, which subsequently facilitated the retainment of lymphocytes within such tissue regions. These results agreed with the current acknowledgment that the CD8+ T cell “problem” in immune-excluded tumors related more to features of the peritumoral microarchitecture that favor their retention in the stromal compartment rather than a strictly intrinsic and irreversible rewiring of CD8+ T cell physiology.42 As to the question about whether the versican-composed peritumoral ECM had some selectivity to the immune cell types that might be retained in the regions, we only observed its effects on CD8+ T cells and did not have a definitive answer for other cell types. We hypothesized that different cells might respond differently to the versican-composed ECM, given that levels of myeloid cells were much higher than those of T lymphocytes within intratumoral tissues of HCC.
Peritumor is a region between the non-tumor and intratumoral tissues, often with accumulated immune infiltrates.21 37 Previous study found that peritumoral monocytes exhibited mixed phenotypes—upregulating the expression of HLA-DR and pro-inflammatory cytokines while simultaneously increased PD-L1 expression.15 Their ultimate functions were determined by the interaction of monocytes with their adjacent cells. Our current study unveiled that CD93+ monocytes not only impacted the activation of CD8+ T cells via the PD-L1-PD-1 pathway but also contributed to the remodeling of peritumoral physical compositions, possibly thickening the ECMs and baring the infiltration of T cells into intratumoral tissues. Consistently, levels of CD93+ monocytes were positively correlated with those of CD8+ T cells in peritumoral tissues, but negatively associated with levels of intratumoral CD8+ T cells. In line with these in vitro findings, in vivo mice experiments confirmed that targeting CD93+ monocytes could increase both the infiltration and activation of CD8+ T cells within HCC tumor tissues.
Our previous study found that HCC tumor cell-derived hyaluronan (HA) fragments could induce glycolytic switch in monocytes.20 But blockade of HA only partially attenuated the effects of TSN on monocytes, indicating that HA might combinate with other factors to regulate the activation of these cells. Whether tumor cell-derived HA induced the expression of CD93 through glycolysis also requires further verification.
There are several limitations of the current study. For example, we were not able to determine the amount of versican in supernatants of either the CD93+ monocytes, or the peritumoral tissues of HCC; the mechanism mediating CD93-induced versican production has not been clarified; it has not been explored whether versican also impacts the functions CD8+ T cells, and the possible Fc-mediated functions of anti-PD-1 antibodies used in the mice experiments could not be excluded. Nonetheless, our results established a previously unexplored role of CD93 on monocytes, linking these cells to the inhibition and possible physical retainment of CD8+ T cells within the peritumoral tissues of HCC, therefore, providing a reasonable target for the future immunotherapy of solid tumors.
Supplemental material
Data availability statement
Data are available in a public, open access repository. All data relevant to the study are included in the article or uploaded as online supplemental information. The Bulk RNA-seq data from the TCGA LIHC dataset were obtained from the Cancer Genome Atlas (TCGA) (https://www.cancer.gov/ccg/research/genome-sequencing/tcga). The immune infiltration data of TCGA LIHC were obtained from TIMER 2.0 (http://timer.cistrome.org/). Some scRNA-seq data were downloaded from Tumor Immune Single-cell Hub (http://tisch.comp-genomics.org/), other scRNA-seq data (BioProject ID PRjCA007744) were downloaded from the Genome Sequence Archive at the National Genomics Data Center (https://ngdc.cncb.ac.cn/gsa-human/browse/HRA001749). The Bulk RNA-seq data of siNC and siCD93 monocyte with accession number GSE276062 have been deposited at the NCBI’s Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/).The other data generated in this study are included in the article or uploaded as online supplemental materials.
Ethics statements
Patient consent for publication
Ethics approval
All samples were anonymously coded in accordance with local ethical guidelines (as stipulated by the Declaration of Helsinki). Written informed consent was obtained from each patient, and the study protocol was approved by the Review Board of Sun Yat-sen University Cancer Center (G2023-226-01).
Acknowledgments
The authors thank Sun Yat-sen University Cancer Center and Sun Yat-sen Memorial Hospital for their assistance in obtaining clinical samples and related clinical data. The authors thank Ling-Yan Zhu for help with flow cytometry analysis and cell sorting (School of Life Sciences, Sun Yat-sen University).
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
DJ and AH contributed equally.
Contributors DJ designed the experiments, processed tissues, performed flow cytometry, immunoblotting, mouse experiments, collected data, and participated in the paper writing. AH performed immunohistochemical and immunofluorescence staining, mouse experiments, and analyzed the data. B-XZ and JG performed ELISA and transwell assay. Y-HR and X-CL helped with and performed RNA-seq analysis. LZ and YW planned and supported the project, analyzed data, and wrote the paper. YW was responsible for the overall content as the guarantor.
Funding The work was supported by project grants from the National Key R&D Program of China (2023YFA0915701), the National Natural Science Foundation of China (82372877, 82071743, 32230034), the Fundamental Research Funds for the Central Universities (23yxqntd001), Guangdong Science and Technology Department (2023B1212060028), Guangdong Basic and Applied Basic Research Foundation (2022A1515110248, 2023A1515012540), and the Key-Area Research and Development Program of Guangdong Province (2023B1111020005).
Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographic or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.