If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Hepatocellular carcinoma (HCC) is a highly aggressive malignancy with dreadful clinical outcome. Tyrosine kinase inhibitors and immune checkpoint inhibitors are the only United States Food and Drug Administration-approved therapeutic options for patients with advanced HCC with limited therapeutic success. Ferroptosis is a form of immunogenic and regulated cell death caused by chain reaction of iron-dependent lipid peroxidation. Coenzyme Q10 (CoQ10)/ferroptosis suppressor protein 1 (FSP1) axis was recently identified as a novel protective mechanism against ferroptosis. We would like to explore whether FSP1 could be a potential therapeutic target for HCC.
FSP1 expression in human HCC and paired non-tumorous tissue samples were determined by reverse transcription-quantitative polymerase chain reaction, followed by clinicopathologic correlation and survival studies. Regulatory mechanism for FSP1 was determined using chromatin immunoprecipitation. The hydrodynamic tail vein injection model was used for HCC induction to evaluate the efficacy of FSP1 inhibitor (iFSP1) in vivo. Single-cell RNA sequencing revealed the immunomodulatory effects of iFSP1 treatment.
We showed that HCC cells greatly rely on the CoQ10/FSP1 system to overcome ferroptosis. We found that FSP1 was significantly overexpressed in human HCC and is regulated by kelch-like ECH-associated protein 1/nuclear factor erythroid 2-related factor 2 pathway. FSP1 inhibitor iFSP1 effectively reduced HCC burden and profoundly increased immune infiltrates including dendritic cells, macrophages, and T cells. We also demonstrated that iFSP1 worked synergistically with immunotherapies to suppress HCC progression.
We identified FSP1 as a novel, vulnerable therapeutic target in HCC. The inhibition of FSP1 potently induced ferroptosis, which promoted innate and adaptive anti-tumor immune responses and effectively suppressed HCC tumor growth. FSP1 inhibition therefore represents a new therapeutic strategy for HCC.
Inhibition of ferroptosis suppressor protein 1 suppresses hepatocellular carcinoma tumor growth via the induction of ferroptosis. Ferroptotic cell death also promotes innate and adaptive anti-tumor immunity. This represents a novel and potent therapeutic target for hepatocellular carcinoma with improved treatment outcome.
Hepatocellular carcinoma (HCC) is the most common form of primary liver cancer that originates from hepatocytes. Late symptom presentations result in most patients with HCC diagnosed at advanced stages, and curative treatments like surgical resection are unfeasible.
Effective therapeutic options for patients with HCC are limited, as HCC is highly recurrent and highly resistant to conventional chemo- and radiotherapy. Currently, tyrosine kinase inhibitors (TKIs) (sorafenib, lenvatinib, regorafenib, cabozantinib) and immune checkpoint inhibitors (ICIs) (nivolumab [anti-PD-1] and atezolizumab [anti-PD-L1]) are the only United States Food and Drug Administration-approved drugs for patients with advanced HCC.
Unfortunately, TKIs provide modest survival benefits, whereas ICIs are only effective in a small subset of patients with HCC. Drugs that can target HCC cells to cell death are worth investigation. Further investigation into drugs that can induce HCC cell death will benefit the development of effective HCC treatment, as these agents may be utilized as HCC therapeutics.
Ferroptosis is the most-recently discovered form of regulated cell death (RCD) that aggravates degenerative diseases and paradoxically suppresses cancer cell growth.
Ferroptosis vastly differs from conventional forms of RCDs. Morphologically, ferroptosis displays outer mitochondrial membrane rupture, shrunken mitochondria with condensed membrane densities, and loss of cristae.
Iron reacts with hydrogen peroxides to generate hydroxyl radicals, highly toxic reactive oxygen species (ROS) that further react with polyunsaturated fatty acid (PUFA) tails on the cell membrane lipid bilayer to generate lipid peroxides, a process called lipid peroxidation. Lipid peroxides establish a chain reaction and produce more lipid peroxides. These further cause damage to the cell membrane, which eventually causes ferroptosis. Iron is essential for lipid peroxidation chain reaction propagation and also for catalyzing iron-dependent enzymes like lipoxygenases to generate lipid peroxides.
Cells have 3 major protective mechanisms against ferroptosis: (1) the cysteine/glutathione (GSH)/glutathione peroxidase 4 (GPX4) pathway; (2) the mitochondrial GPX4-dependent coenzyme Q (CoQ)/dihydroorotate dehydrogenase pathway; and (3) the CoQ10/ferroptosis suppressor protein 1 (FSP1) pathway. The cysteine/GSH/GPX4 pathway was the first discovered and most well-known protective pathway against ferroptosis.
In cysteine/GSH/GPX4 pathway, GPX4 removes phospholipid peroxides using reduced GSH. GSH synthesis depends on the xCT cystine-glutamate antiporter (encoded by SLC7A11 and SLC3A2), which pumps cystine into cells in exchange for glutamate.
GPX4 reduces phospholipid hydroperoxides (PL-PUFA-OOH) to counteract ferroptosis. Dihydroorotate dehydrogenase is a mitochondrial enzyme localized on the outer face of the inner membrane. It converts dihydroorotate to orotate and generates FMNH2 to reduce ubiquinone (CoQ) into ubiquinol (CoQH2) to protect cells from ferroptosis at the mitochondrial membrane.
The CoQ10/FSP1 pathway acts independently of GPX4 as FSP1 does not affect the xCT antiporter and cysteine transport, GSH levels, and total PL-PUFA (Figure 1A). Although the GPX4 pathway is an attractive therapeutic target for cancer treatment through ferroptosis induction, the efficacy of current GPX4 inhibitors is limited to in vitro studies, and the responses vary across different cancer cell lines.
Inhibitors, such as erastin and sorafenib, that target xCT antiporter are not highly specific. GPX4 is expressed in T cells, and these inhibitors could therefore inhibit T cells and cause T cell exhaustion in the cancer context.
These limitations prompted us to explore new anti-ferroptotic mechanisms beyond the cysteine/GSH/GPX4 pathway to discover new therapeutic strategies against HCC. The CoQ10/FSP1 pathway is another protective pathway against ferroptosis that works independently of the cysteine/GSH/GPX4 pathway.
Immunogenic cell death (ICD) is characterized by its release or membrane exposure of damage-associated molecular patterns (DAMPs), which recruit and activate innate immune cells, subsequently eliciting adaptive immune responses.
Innate immune cells play a key role in ICD-elicited anti-tumor immune responses. Dendritic cells (DCs) internalize dying cells or cellular fragments for presenting tumor-associated antigens to T cells via major histocompatibility complex molecules.
The immunomodulatory roles of ferroptotic cells in HCC are yet to be explored. Apart from activating innate immune cells through ICD induction, a new generation of immunotherapies that target innate immune cells has emerged as a promising therapeutic approach. Among these is CD40 agonist. CD40 is a co-stimulatory receptor expressed on antigen presenting cells (APCs), such as macrophages and DCs. Ligation of CD40 with CD40L expressed on activated T cells enhances antigen presentation by macrophages and DCs.
Here, we explore whether CD40 agonist can further activate innate immune cells and enhance iFSP1 efficacy in HCC.
We hypothesize that targeting FSP1 induces ferroptosis in cancer cells and promotes anti-tumor immune responses. Here, we showed that targeting FSP1 alone is sufficient to severely induce ferroptosis and suppress HCC cells proliferation as well as HCC tumor growth even with intact GPX4. Our work suggested that perturbation of the CoQ10/FSP1 pathway suffices to inhibit HCC development through ferroptosis induction. More excitingly, we unprecedentedly showed that iFSP1 treatment in mice profoundly increased the number of DCs, macrophages, and T cells in the tumors and suggested that iFSP1 works in synergy with multiple immunotherapies to improve the survival of mice with HCC.
FSP1 Overexpression is Significantly Associated With Human HCC
To understand the clinical significance of ferroptosis regulation, especially FSP1 in human HCC, we made use of transcriptome sequencing results from 49 paired human HCC/non-tumorous (NT) liver tissue samples from The Cancer Genome Atlas (TCGA). Significant overexpression of FSP1 was found in human HCC tissues compared with NT tissues (Figure 1B). These transcriptome sequencing results also revealed significant downregulation of pro-ferroptotic gene TF (Figure 1C). Interestingly, genes from the classic cysteine/glutathione/GPX4 anti-ferroptosis pathway, including GPX4, a major anti-ferroptotic gene, were not upregulated in human HCC (Figure 1C). This suggested that HCC was protected from ferroptosis by limiting iron uptake into cancer cells and upregulation of FSP1 pathway but not GPX4 pathway, as both pathways work independently to counteract ferroptosis. An expanded cohort of 70 HCC cases from our institute (HKU-QMH) also demonstrated FSP1 overexpression in HCC tissues over NT liver tissues at the mRNA level determined by real-time quantitative polymerase chain reaction (RT-qPCR). FSP1 overexpression by at least 2-fold was found in 44 of 70 (62.9%) of the HCC tumor samples (Figure 1D). Clinicopathologically, FSP1 overexpression in HCC samples from our institute were significantly associated with patients with HCC with larger tumor size (>5 cm) (Table 1). Moreover, TCGA data indicated patients with HCC with higher FSP1 expression were significantly correlated with poorer overall survival (P = .018) and were associated with poorer disease-free survival (P = .106) (Figure 1E). Interestingly, we found positive correlation between mRNA expression of FSP1 and nuclear factor erythroid 2-related factor 2 (NRF2), transcriptional targets we previously identified, including transketolase (TKT), malic enzyme 1 (ME1), and methylenetetrahydrofolate dehydrogenase 1-like (MTHFD1L), as well as other well-known NRF2 targets NAD(P)H dehydrogenase, quinone 1 (NQO1), aldo-keto reductase family 1 member B10 (AKR1B10), and aldo-keto reductase family 1 member C1 (AKR1C1) in human HCC and NT liver tissues from TCGA (Figure 1F).
Expression analyses from clinical HCC samples on TCGA and our institute converged to identify overexpression of FSP1 in human HCC is significantly correlated with dismal clinicopathological features of HCC.
Table 1Clinicopathological Correlation of FSP1 Overexpression in Human HCC
Next, we aimed to understand the regulatory mechanism that drives the overexpression of FSP1 in human HCC. Because FSP1 is essential for counteracting lipid peroxidation, it was postulated that FSP1 expression is regulated by the kelch-like ECH-associated protein 1 (KEAP1)/NRF2 pathway like other oxidative stress-related genes. KEAP1/NRF2 is the master regulatory pathway for an array of genes known for protection against oxidative stress.
We found positive correlation between FSP1 expression with NRF2 transcriptional targets we previously found, including TKT, ME1, MTHFD1L, and known NRF2 targets NQO1, AKR1B10, and AKR1C1 (Figure 1F), strongly suggesting that FSP1 overexpression in human HCC is regulated by the KEAP1/NRF2 pathway (Figure 2A).
Oxidative stress is a common feature in solid tumors and is believed to promote FSP1 upregulation in human HCC. In silico analysis identified a putative antioxidant response element (ARE) at −22 bp upstream of the transcription start site in FSP1. Furthermore, dramatic enrichment of NRF2 binding at the mapped putative ARE of FSP1 was confirmed by chromatin immunoprecipitation (ChIP) assay using NRF2 and IgG antibodies on multiple HCC cell lines, including MHCC97L, PLC/PRF/5, and Huh7 (Figure 2B). We next sought to confirm the regulatory role of KEAP1/NRF2 and small Maf proteins that dimerize with NRF2 to bind to ARE sequences on FSP1. Stable NRF2, KEAP1, and MAF BZIP transcription factor (MAF) knockdown clones were established in multiple HCC cell lines, including MHCC97L, PLC/PRF/5, and Huh7 using short hairpin RNAs (shRNAs) (shNRF2, shKEAP1, and shMAF). To show that the ARE we identified at FSP1 promoter region is functional and how NRF2 and MAF regulates FSP1, we performed luciferase reporter assay. Wild-type (WT) and mutated (MUT) AREs were first inserted into pGL4 luciferase reporter vector, and we transfected stable NRF2 and MAF knockdown clones of Huh7 with the WT-ARE or MUT-ARE containing luciferase plasmids. We then exposed the cells to treatment of tert-Butyl hydroperoxide (tBHP; Sigma-Aldrich, St. Louis, MO), an oxidant to induce NRF2, and determined the FSP1 promoter luciferase activities. We showed that tBHP treatment induced luciferase activities in cells expressing WT-ARE but not the MUT one (Figure 2C). The induction of promoter activities by tBHP in WT-ARE expressing cells could be abrogated by NRF2 or MAF knockdown (Figure 2C). We previously demonstrated the NRF2 stable knockdown clones down-regulated the expressions of known NRF2-target genes like TXNRD1 and ME1, whereas KEAP1 stable knockdown clones up-regulated the same targets.
These results validate the validity of the NRF2 and KEAP1 inhibitions, and, consistent with our hypothesis, NRF2 inhibition effectively down-regulated FSP1 expressions (Figure 2D) and inhibition of KEAP1, the NRF2 suppressor, effectively up-regulated FSP1 expressions (Figure 2E) at both the mRNA and protein levels. We also confirmed that MAF contributes to FSP1 regulation in stable MAF knockdown clones. The inhibition of MAF effectively down-regulated FSP1 expression at the mRNA level (Figure 3A). Finally, we made use of a chemical NRF2 inducer L-sulforaphane (SFN; Sigma-Aldrich) and treated HCC cells (MHCC97L, PLC/PRF/5, Huh7) at various concentrations. Dose-dependent up-regulation of FSP1 were consistently observed at mRNA and protein levels upon SFN treatment across all 3 HCC cell lines (Figure 2F). We further confirmed our findings using another NRF2 inducer CDDO-Imidazolide (CDDO-Im; Tocris Bioscience, Bristol, UK), as dose-dependent up-regulation of FSP1 was also observed at the mRNA level across different HCC cells upon CDDO-Im treatment (Figure 3B). Taken together, in silico analysis, ChIP assay, luciferase reporter assay, and NRF2 inhibition strongly suggested that FSP1 is transcriptionally regulated by NRF2 in HCC.
Knockdown of FSP1 Resulted in Ferroptosis-induced HCC Cell Death
We next studied the functional roles of FSP1 in human HCC, and we performed loss-of-function studies. First, stable FSP1 knockdown clones were established in MHCC97L cells with 2 independent shRNA sequences (shFSP1-23, -24). Significant inhibition of FSP1 expressions compared with non-targeting control (NTC) were evident in mRNA and protein levels (Figure 4A). FSP1 is an oxidoreductase, converting oxidized CoQ10 into reduced CoQ10H2. Liquid chromatography-mass spectrometry (LC-MS) confirmed that knockdown of FSP1 significantly reduced intracellular CoQ10H2/CoQ10 ratio (Figure 4B). Through the reduction of CoQ10, FSP1 is important for lipid peroxidation prevention, and we detected significant increase in intracellular lipid peroxidation in MHCC97L cells upon FSP1 knockdown (shFSP1-23, -24) compared with NTC control cells using BODIPY 581/591 C11 staining (Invitrogen, Carlsbad, CA) (Figure 4C). FSP1 knockdown obstructed HCC cell growth in vitro as cell proliferation rates of both knockdown clones (shFSP1-23, -24) were significantly lower than NTC cells. The reduced cell proliferation rates of the FSP1 knockdown clones were rescued upon treatment of ferroptosis inhibitor ferrostatin-1 (Fer-1; Sigma-Aldrich) (Figure 4D). This suggested that FSP1 knockdown adversely affects HCC growth through lipid peroxidation-mediated ferroptosis. Luciferase-labeled MHCC97L stable FSP1 knockdown clones and NTC were orthotopically injected into nude mice livers. Tumor volume and bioluminescent measurements consistently revealed that smaller HCC tumors were derived from FSP1 knockdown clones. Moreover, lung tissue imaging ex vivo revealed FSP1 knockdown suppressed the growth of lung metastases (Figure 4E). To confirm that FSP1 is essential for preventing ferroptosis-induced HCC cell death, we confirmed that FSP1 inhibition resulted in the accumulation of intracellular lipid peroxidation. A significant increase in intracellular levels of lipid peroxidation by-products 4-hydroxynonenal (4-HNE) and malondialdehyde (MDA) were detected from FSP1 knockdown stable clones relative to NTC (Figure 4F). We then examined whether cytotoxic effects were elicited because of lipid peroxidation. Quantification of lactate dehydrogenase (LDH) released from damaged cells, an indicator of cell death, revealed that FSP1 knockdown cells released significantly higher levels of LDH compared with NTC (Figure 4G). Furthermore, SYTOX Green dead cell staining (Invitrogen) also demonstrated similar results as LDH release. Fer-1 treatment to FSP1 knockdown clones (shFSP1-23, -24) significantly lowered the relative SYTOX Green staining compared with untreated cells (Figure 4H). Fer-1 treatment was also found to significantly increase relative bromodeoxyuridine (BrdU) intensities in MHCC97L-shFSP1-23 and -24 cells. Treatment of inhibitors of other forms of RCDs including necroptosis (GSK’872), apoptosis (Z-VAD-FMK), and autophagy (MRT68921), however, did not rescue the suppressed cell proliferation rates in FSP1 stable knockdown HCC cells (Figure 4I). The observations altogether demonstrated that only ferroptosis inhibitor (eg, Fer-1) was able to rescue FSP1 knockdown-induced cell death in HCC cells. This indicated the important role of FSP1 for the protection of HCC cells from ferroptosis and cell death.
iFSP1 Effectively Induced HCC Cell Death Via Immune Cell Recruitment
We demonstrated that FSP1 knockdown led to accumulation of intracellular lipid peroxidation, which ultimately resulted in ferroptosis-induced cell death in human HCC. As we now look to translate this concept for clinical application, we made use of a specific chemical FSP1 inhibitor, iFSP1.
As a novel and sole FSP1 inhibitor, we tested the effects of iFSP1 on MHCC97L cells at multiple concentrations. Cell proliferation rates were significantly suppressed dose-dependently upon iFSP1 treatment (Figure 5A). Additionally, iFSP1 treatment, like erastin, which targets the xCT antiporter of the GPX4 system, also induced a greater extracellular release of LDH compared with vehicle control, further indicating that iFSP1 elicited cytotoxic effects on HCC cells (Figure 5B). Treatment of iFSP1 with the same concentrations as Figure 5A also displayed dose-dependent increase of intracellular lipid peroxidation with BODIPY 581/591 C11 staining, and cell death indicated by increased SYTOX Green dead cell staining (Figure 5C‒D). These results together suggested that pharmacological inhibition of FSP1 using iFSP1 effectively induces ferroptosis, which inhibits HCC growth. We then performed in vitro assays to detect the intracellular levels of lipid peroxidation by-products 4-HNE and MDA in HCC cells upon iFSP1 treatment. Both 4-HNE and MDA levels were significantly increased upon iFSP1 treatment relative to vehicle control cells (Figure 5E). Upon iFSP1 treatment in MHCC97L cells, we observed that co-treatment of Fer-1 reversed the effects as the relative BrdU intensities significantly increased. Co-treatment of necroptosis (GSK’872), apoptosis (Z-VAD-FMK), and autophagy (MRT68921) inhibitors had minimal rescue effect to the suppressed cell proliferation induced by iFSP1 in the BrdU assay (Figure 5F). Dying cells recruit APCs and phagocytes that mediate their clearance. We then asked whether iFSP1 treatment promotes HCC cells to be phagocytosed by macrophages. We first pre-treated mouse HCC cells Hepa1-6 with iFSP1 or vehicle control and labeled them with CellTrace Violet (CTV) dye (Invitrogen). Then, we co-cultured CTV-labeled Hepa1-6 cells with murine bone marrow-derived macrophages (BMMs). Phagocytic events were measured by CTV+ cells in CD11b+ F4/80+ macrophage population using flow cytometry (Figure 5Gi). We showed that iFSP1 treatment significantly enhanced phagocytosis of Hepa1-6 cells by BMMs (Figure 5Gii). Co-treatment with iFSP1 and Fer-1 abrogated the increase of phagocytosis of Hepa1-6 cells by BMMs (Figure 5Giii). We have demonstrated that ferroptotic HCC cells recruit macrophages and enhance the engulfment of ferroptotic cells, suggesting the potential immunomodulatory effect of iFSP1. To further study the therapeutic potential of iFSP1 and whether it affects the intra-tumoral immune composition, we induced HCC tumor with Keap1KO/c-MycOE genetic background by hydrodynamic tail vein injections (HDTVi) in immunocompetent C57BL/6N mice fed with high-fat diet (HFD) (60 kcal% fat, 0.1% methionine, no choline) (Research Diets, Inc, New Brunswick, NJ). In this model, HFD promoted lipid accumulation in liver, whereas Keap1KO resulted in Nrf2 activation, generating a highly aggressive murine HCC model (Figure 11A). To characterize this model, we performed hematoxylin and eosin staining on the tumors, and they were confirmed to be HCC by a pathologist (Figure 6A). We showed that FSP1 expression was particularly induced in Keap1KO tumors (Figure 6B). Furthermore, conventional HCC marker alpha fetoprotein (Afp) was overexpressed in tumor tissues (Figure 6C). We showed that iFSP1 treatment significantly suppressed HCC tumor growth with no adverse effects on body weight (Figure 5Hi). Furthermore, flow cytometry analysis illustrated that iFSP1 greatly increased the number of macrophages, DCs, and CD4+ and CD8+ T cells in HCC tumor compared with vehicle control (Figure 5Hii). We observed no change in the number of neutrophils and a slight increase in monocytes (Figure 6D). Further characterizations showed that iFSP1 promoted tumor infiltration of M1 macrophages and effector T cells but no change in M2 macrophages (Figure 6D‒E). Tumor infiltration of CD4+ and CD8+ T cells was confirmed by immunohistochemical (IHC) staining (Figure 5I).
We further examined the effects of ferroptosis induction in human HCC using erastin (Selleck Chemicals, Houston, TX), another well-established ferroptosis inducer targeting the xCT antiporter and inhibiting the cysteine/GPX4 pathway.
We found that erastin significantly induced ferroptosis, suppressed HCC cell proliferation, and promoted their uptake by BMMs in vitro (Figure 7A‒D). Furthermore, erastin effectively inhibited HCC tumor growth in vivo, but in stark contrast to iFSP1, the number of intratumoral CD4+ and CD8+ T cells was also reduced (Figure 7E). The discrepancy in T cell number upon iFSP1 (FSP1 inhibitor) and erastin (GPX4 inhibitor) treatments in HCC tumors may be due to T cells’ reliance on the GPX4 pathway to counteract ferroptosis.
Our findings illustrated that ferroptosis induction via FSP1 inhibition using iFSP1 suppressed HCC tumor growth not only through mediating HCC cell death, but also by altering the immune landscape of HCC tumors. Our findings highlight FSP1 inhibition as a unique anti-tumor strategy in comparison to other ferroptosis inducers.
Single-cell RNA Sequencing Revealed Immune Landscape Alteration Upon iFSP1 Treatment
To further explore immune landscape changes in HCC tumors upon iFSP1 treatment, we performed single-cell RNA sequencing (scRNA-seq) on single cells isolated from murine HDTVi-induced HCC tumors treated with iFSP1 or vehicle control (Figure 8A). Using unsupervised clustering analysis, we identified major immune cell populations by their marker genes. The representative marker genes were visualized by t-distributed stochastic neighbor embedding (tSNE) plots. These cell populations included B cells (Cd19), DCs (Flt3), macrophages (Cd68), monocytes (Lyz2), neutrophils (Csf3r), NK cells (Ncr1), pDCs (Siglech), T cells (Cd3e), and mast cells (Cpa3) (Figure 9A‒B). Consistent with our observations from flow cytometry analysis, iFSP1 treatment resulted in increase of DCs, macrophages and T cells (Figure 9C).
We first investigated the myeloid compartment. Unsupervised clustering analysis further identified 4 DC subsets, 2 macrophage subsets, and 3 monocyte subsets (Figures 9D, 8B, 8C). All DC subsets increased upon iFSP1 treatment (Figure 9E). pDC represented plasmacytoid DCs expressing Siglech, Irf8, and Bst2. cDC1 expressed Xcr1, Clec9a, and Batf3, whereas cDC2 expressed Cd209a, Itgax, and Clec10a, corresponding to typical marker genes of conventional DC1 and conventional DC2, respectively (Figure 8B). cDC3 expressed Fscn1, Ccl22, and Ccr7 (Figure 8B). It resembled a mature DC subset that expressed high level of activation and regulatory markers and shown to be derived from both cDC1 and cDC2.
The 3 cDC subsets expressed a high level of major histocompatibility complex II molecules for antigen presentation (Figure 9F). For the 2 macrophage subsets, iFSP1 reduced Mac-1 and increased Mac-2 (Figure 9E). We further studied the differentially expressed genes among these 2 macrophage subsets. Interestingly, we found that Mac-2 was enriched in phagocytosis-related genes, such as C1qa, C1qb, C1qc, and Mertk (Figure 9G). Mac-2 also expressed a panel of receptors mediating phagocytosis (Figure 9H). These results illustrated that iFSP1 treatment recruit DCs with antigen presentation ability and phagocytic macrophages to tumor site.
We next studied subsets in T cells. We identified 2 CD4+ and 4 CD8+ T cell clusters, and they were annotated based on their marker gene expression (Figures 10A, 8D). Proliferating and exhausted CD8+ T cells were particularly increased upon iFSP1 treatment (Figure 10B). Proliferating CD8+ T cells are characterized by their expression of Mki67. Exhausted T cells expressed exhaustion markers including Tox, Pdcd1, Lag3, Tigit, and Havcr2 (Figures 10C, 8D). Importantly, both T cell populations expressed multiple cytotoxic markers, including Nkg7, Gzmk, Gzmb, Gzma, and Prf1 (Figure 10D). This suggested that iFSP1 treatment promoted tumor infiltration of CD8+ T cells with cytotoxic functions.
To understand how these immune cells were recruited to tumor site, we then explored the expression of chemokines and observed that Cxcl9 and Cx3cl1 were induced in multiple DC and macrophage populations upon iFSP1 treatment (Figure 10E). CXCL9 is a well-known chemoattractant for CXCR3+ T cells, whereas CX3CL1 recruit DCs and macrophages expressing receptor CX3CR1. Moreover, we observed induction of costimulatory receptor, Cd40 expression on DCs and macrophages and Pdcd1, encoding PD-1, on T cells (Figure 10F), pointing to potential targets for combination treatment.
As a systemic treatment, it is possible that iFSP1 also influences immune cell populations; therefore, we studied the expression of anti-ferroptotic genes in major immune cell populations. Interestingly, we found that immune cells expressed a high level of Gpx4 but a low level of Fsp1 (Figure 10G). Moreover, components of xCT antiporter were generally expressed in major immune cell populations (Figure 8E), suggesting that immune cells greatly rely on the GPX4 pathway to protect themselves from ferroptosis. This could also explain the reduction in T cells we observed in erastin-treated tumors.
iFSP1 in Combination With Immunotherapies Effectively Prolonged HCC Survival
The upregulation of PD-1 on T cells and CD40 expression on DCs and macrophages prompted us to explore the synergistic effect of iFSP1 and immunotherapies. HCC tumors were induced by HDTVi in mice fed with HFD, and mice were treated with iFSP1 in combination with either ICIs (anti-PD-1 and anti-PD-L1; BioXcell, Lebanon, NH), or a new generation of immunotherapy, CD40 agonist (BioXcell). Single treatment of iFSP1 or anti-PD-1 improved survival of tumor-bearing mice similarly, whereas co-administration of them effectively prolonged survival of mice (Figure 11B). Interestingly, iFSP1 showed better efficacy in prolonging survival than anti-PD-L1 as a single treatment in our HCC model. Combining iFSP1 and anti-PD-L1 can further extend the survival of HCC-bearing mice (Figure 11C). Apart from the 2 ICIs currently used in patients with HCC, we also explored the effect of CD40 agonist in combination with iFSP1. CD40 agonist worked with iFSP1 to prolong the survival of tumor-bearing mice (Figure 11D). Here, we demonstrated that iFSP1 treatment can improve or potentiate the effect of multiple immunotherapies without adverse effects on body weight (Figure 12). Our results suggested that iFSP1 treatment recruited APCs and promoted antigen presentation by APCs to T cells. This is further enhanced by immunotherapies that either reinvigorate T cells (anti-PD-1 and anti-PD-L1) or activate APCs (CD40 agonist) (Figure 11E).
Ferroptosis, as a distinct form of cell death, was first reported to suppress cancer in 2003; however, the implications of ferroptosis in cancers were only revealed in recent years.
A recent study demonstrated that ferroptosis inhibits metastasis. Melanoma cells in the blood were shown to encounter a higher level of ferroptosis than melanoma cells in the lymph, resulting in more melanoma metastases formed from lymph node metastasis than vascular metastasis.
Interestingly, triple-negative breast cancer cells express higher levels of pro-ferroptotic genes (TfR and ACSL4) and lower levels of anti-ferroptotic genes (ferritin heavy chain 1, ferroportin, GPX4, GSH synthetase) and are therefore more susceptible to ferroptosis.
The roles of ferroptosis in HCC, especially FSP1-driven mechanisms, have not been fully explored. Our study unprecedentedly reported that HCC is reliant on FSP1 for protection against ferroptosis rather than GPX4, as we showed that FSP1 is upregulated in human HCC but not GPX4. FSP1 deficiency led to decreased ubiquinol-to-ubiquinone ratio, diminishing the ability of cancer cells to counteract lipid peroxidation. This induced ferroptosis in cancer cells and suppressed HCC tumor progression.
Multiple pathways are involved in the regulation of ferroptosis. P53 transcriptionally represses SLC7A11 to inhibit cysteine intake, which leads to ROS-induced ferroptosis in osteosarcoma, NSCLC, and breast cancer cells.
In HCC cells treated with ferroptosis inducers such as erastin and sorafenib, P62 releases NRF2 from KEAP1 binding, thereby stabilizing NRF2 so that NRF2 mediates the transcription of NQO1, heme oxygenase-1, and FTH1 to counteract ferroptosis.
In this study, we showed for the first time that FSP1 is transcriptionally regulated by NRF2. NRF2 inhibitors such as clobetasol propionate, which we previously showed to have remarkable anti-tumor effects in HCC, are likely to induce ferroptosis through inhibiting GPX4 and FSP1 pathways.
However, systemic administration of NRF2 inhibitors might also elicit ferroptosis of immune cells, affecting the combined effects with ICIs and CD40 agonist.
Ferroptosis is a form of ICD. Multiple studies illustrated that ferroptotic cancer cells released DAMPs such as ATP, HMGB1, and oxidized phospholipids, which promoted their uptake by DCs and macrophages.
However, whether and how ferroptosis induction in cancer promotes anti-tumor immune responses are yet to be investigated. Here, we showed that induction of ferroptosis by FSP1 inhibition suppressed HCC tumor growth. This anti-tumor effect is not only due to the direct induction of cell death but also the activation of anti-tumor immune responses. We demonstrated that iFSP1 treatment suppressed HCC progression and promoted infiltration of antigen presenting DCs, phagocytic macrophages, and cytotoxic T cells into tumor. These immune cells could be recruited by chemokines that were upregulated upon iFSP1 treatment. CXCL9 is a well-studied chemokine that is important for CXCR3+ T cell trafficking and recruitment to tumor site.
This study mainly focuses on how T cells induce ferroptosis of cancer cells, whereas our study revealed that induction of ferroptosis in cancer cells could reciprocally promote T cell tumor infiltration. Together, these suggest a potential feedforward loop to suppress cancer progression.
Drugs that induce ICD are being hotly pursued because of their potential as combination treatments with ICIs in multiple cancer types.
Treatments inducing ICD lead to the release of tumor antigens and DAMPs by cancer cells and thus recruiting APCs to engulf the tumor antigens. This enhances antigen presentation by APCs to T cells, initiating anti-tumor immune response.
Upon iFSP1 treatment, we observed an upregulation of CD40 on DCs and macrophages while PD-1 on T cells. Excitingly, we showed that combination treatment of iFSP1 and anti-PD-1, anti-PD-L1, or CD40 agonist can all prolong survival of tumor-bearing mice. Although anti-PD-1 and anti-PD-L1 aim at blocking PD-1 and PD-L1 ligation and thus reinvigorating T cells, CD40 agonist enhances antigen presentation ability of APCs and promotes the secretion of a plethora of pro-inflammatory cytokines and has been demonstrated to effectively suppress cholangiocarcinoma.
CD40 agonist demonstrated clinical activity when used in combination with chemotherapies and nivolumab in treating metastatic pancreatic adenocarcinoma, whereas its effects in HCC remain to be investigated.
CD40 agonistic monoclonal antibody APX005M (sotigalimab) and chemotherapy, with or without nivolumab, for the treatment of metastatic pancreatic adenocarcinoma: an open-label, multicentre, phase 1b study.
Recently, it was reported that TYRO3 is also an anti-ferroptotic gene through inducing NRF2-mediated transcription to suppress lipid ROS accumulation. Cancers with TYRO3 overexpressed are resistant to anti-PD-1 treatment, indirectly suggesting that ferroptosis improves the efficiency of ICIs.
Together with our findings, these show that ferroptosis induction could work synergistically with immunotherapies and shed light on a novel therapeutic approach for HCC.
The effects of ferroptosis inducers on immune cells require closer investigation. Interestingly, our flow cytometry data showed that total CD4 T cells were enriched in HCC from iFSP1-treated mice, whereas scRNA-seq data showed that the percentage of CD4 central memory T cells decreased. Although the detailed mechanisms explaining why CD4 central memory T cells are reduced remain unknown, studies showed that anti-PD-1/PD-L1 treatment was expected to promote the memory of T cells, further providing mechanistic explanation to combine iFSP1 with ICIs.
Another study performed a high throughput pharmacologic screening to identify specify metabolic vulnerabilities of CD8+ T cells and cancer cells and showed that CD8+ T cells are more susceptible to RSL3, a GPX4 inhibitor, than cancer cells.
These indicated T cells highly rely on GPX4 pathway as a protective mechanism against ferroptosis. Our data showed that the dose of iFSP1 we used for in vivo treatment did not influence the T cell population. But we did observe a reduction in T cell population in GPX4 inhibitor (erastin)-treated tumors despite the suppression of tumor growth. This discrepancy could be due to the reliance of immune cells on the GPX4 pathway to protect themselves from ferroptosis because our scRNA-seq analysis demonstrated that immune cells in general only express GPX4 but not FSP1. Despite the tumor-suppressing effect of inhibitors targeting GPX4 pathway, immune cells also rely on GPX4 pathway to protect them from ferroptosis. Therefore, GPX4 inhibition may not fully demonstrate the potential of ferroptosis inducers as cancer treatment because it could dampen the immune responses. This highlights the importance of further investigation to identify a molecular target that can selectively induce ferroptosis in cancer cells but not immune cells. We showed that FSP1 is overexpressed in human HCC but not GPX4, whereas tumor-infiltrating immune cells express a high level of GPX4 but not FSP1. These suggest that cancer cells and immune cells rely on different mechanisms to protect themselves from ferroptosis in HCC. Therefore, FSP1 could be a superior candidate because its inhibition selectively induces ferroptosis in HCC cells and unleashes anti-tumor immune responses.
Materials and Methods
The Cancer Genome Atlas
Transcriptome sequencing results on 49 pairs of human HCC and corresponding NT liver tissues were retrieved from TCGA through cBioportal (http://www.cbioportal.org/). The mRNA expression levels of FSP1, TF, and GPX4 were examined. Correlation between expression levels of FSP1 and antioxidant genes TKT, ME1, MTHFD1L, NQO1, AKR1B10, and AKR1C1 in human HCC and NT liver tissues were also retrieved from TCGA.
Paired clinical HCC tumor and NT liver tissues were used in this study. Tissue samples were collected from patients with HCC admitted to Pamela Youde Nethersole Eastern Hospital, Queen Elizabeth Hospital, and/or Queen Mary Hospital. Resected tissue samples were snap-frozen with liquid nitrogen and stored at −80 °C for future use. Approval for the use of clinical HCC tumor samples and paired NT liver tissues was previously obtained from the Institutional Review Board of The University of Hong Kong and the Hospital Authority of Hong Kong (HKU/HA HKW IRB; ref. no. UW 09-158). Prior to the collection and use of resected clinical tissues, signed consents were obtained as acknowledgements from patients.
Clinical human and mouse HCC/NT tissues fixed in 4% formalin and 75% ethanol were embedded in paraffin. Before IHC staining, paraffin sections were dewaxed with xylene and rinsed with ethanol. Antigen retrieval was performed by boiling in EDTA buffer, and the liver tissue sections were stained with primary antibody (Table 2) at 4 °C overnight, followed by horseradish peroxidase-conjugated secondary antibody (Dako, Glostrup, Denmark) staining at room temperature for 30 minutes. Sections were developed using 3,3’-diaminobenzidine (Sigma-Aldrich) and counterstained with hematoxylin.
HCC cells were fixed in 3.7% formaldehyde for DNA crosslinking followed by sonication. Antibodies against NRF2 and IgG control (Table 2) were added to sheared DNA fragments. DNA-protein-antibody complexes were incubated with Protein A/Salmon Sperm DNA Agarose Beads (Merck Millipore, Burlington, MA), sequentially washed using gradients of salt buffer, and eluted in 1% SDS/NaHCO3. ChIP DNA samples were analyzed by RT-qPCR with primers flanking the putative ARE (Table 3).
Oligonucleotides containing FSP1 WT or MUT AREs were inserted into the pGL4 firefly luciferase promoter reporter vector. Insert sequences were included in Table 4. These plasmids containing WT or MUT AREs were transfected into Huh7 cells (-NTC, -shNRF2 and -shMAF) together with pRL-CMV and treated with 10μM tBHP for 24 hours. Luciferase activities were determined using the Dual-Luciferase Reporter System (Promega) according to the manufacturer’s instructions.
Table 4shRNA Target Sequences and Insert Sequences for Luciferase Reporter Assay
FSP1, ferroptosis suppressor protein 1; KEAP1, kelch-like ECH-associated protein 1; MAF, MAF BZIP transcription factor; MUT, mutated; NRF2, nuclear factor erythroid 2-related factor 2; shRNA, short hairpin RNA; WT, wild-type.
Briefly, shRNAs targeting NRF2 (shNRF2), KEAP1 (shKEAP1), FSP1 (shFSP1), MAF (shMAF), or NTC contained within pLKO.1-puro vectors were stably transfected into HCC cells. Puromycin was used to select HCC cells that stably express the shRNA-containing vectors. Sequences of all shRNAs used are provided in Table 4.
HCC Cell Culture
Multiple human HCC cell lines including MHCC97L, PLC/PRF/5, Huh7, and a mouse HCC cell line (Hepa 1-6) were used in this study. MHCC97L and Huh7 were kind gifts from Dr Z.Y. Yang of Fudan University (Shanghai, China) and Professor H. Nakabayashi of Hokkaido University School of Medicine (Sapporo, Japan), respectively. PLC/PRF/5 and Hepa 1-6 were obtained from American Type Culture Collect (Manassas, VA). All cell lines were cultured in Dulbecco’s Modified Eagle Medium-high glucose media with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (P/S) supplementations (all reagents from Gibco by Life Technologies, Grand Island, NY). Cells were cultured in a 37 °C humidified incubator supplied with 5% CO2. Authentication of all cell lines within this study was performed using the AuthentiFiler PCR Amplification Kit (Applied Biosystems, Foster City, CA). All authenticated cell stocks were thawed and used within 4 passages.
RNA Extraction, Reverse Transcription, and RT-qPCR
Total RNA was extracted from paired clinical HCC/NT liver tissues, which were snap-frozen with liquid nitrogen and stored at −80 °C, and HCC cell lines with TRIzol reagent (Ambion by Life Technologies, Austin, TX). Purified RNA samples were synthesized into cDNA by reverse transcription using GeneAmp RNA PCR Core Kit (Applied Biosystems). From paired clinical HCC/NT liver tissue samples, RT-qPCR amplifications of FSP1 and internal control 18S were performed using TaqMan Gene Expression Assay (Applied Biosystems). RT-qPCR amplifications of NRF2, KEAP1, FSP1, internal control 18S from HCC cell lines, and Afp and Gapdh from mouse HCC/NT liver tissue samples were performed using SYBR Green PCR Master Mix (Applied Biosystems) with the respective RT-qPCR primers listed in Table 3. All RT-qPCR amplifications were performed using the StepOne Real-Time PCR System (Applied Biosystems).
Protein Extraction and Western Blotting
Total protein lysates from human HCC cell lines were extracted using RIPA lysis buffer with cOmplete protease and PhosSTOP phosphatase inhibitor supplementations (both from Roche Diagnostics, Mannheim, Germany). Equal volumes of protein lysates were separated by SDS-PAGE and transferred onto PVDF membranes (GE Healthcare, Little Chalfont, UK). Protein expression of targeted genes were determined with their respective antibodies listed in Table 2.
Liquid Chromatography-mass Spectrometry
Ubiquinol/ubiquinone were extracted from HCC cells plated on 10-cm culture dishes at 70% to 80% confluence using cold pure methanol. Supernatants from all pelleted samples were transferred to new tubes, dried using a vacuum concentrator, and reconstituted with 100 μL methanol. LC-MS analyses were performed on a ExionLC AC System coupled with X500R QTOF (Sciex Applied Biosystems, Framingham, MA) mass spectrometer. Samples were separated with isocratic elution of methanol containing 2 mM ammonium acetate at a flow rate of 0.4 mL/minute using an ACQUITY UPLC BEH C18 reverse-phase column (2.1 mm × 50 mm, 1.7 μm).
The column temperature was kept at 40 °C. The mass spectrometer was operated in positive ionization mode. Quantification of CoQ10H2 and CoQ10 were performed using TOF-MS/MS as [M + NH4]+ with the following transitions from precursor to product ions: CoQ10H2 : m/z 883 >197, CoQ10: m/z 881 >197. Data acquisition and analyses were performed with SCIEX OS software (Sciex Applied Biosystems).
BODIPY 581/591 C11 Lipid Peroxidation Detection
Intracellular lipid peroxidation was detected in HCC cells plated on 6-well culture dishes at equal densities. Cells were incubated with 5 μmol/L lipid peroxidation sensor BODIPY 581/591 C11 (Invitrogen) at 37 °C for 30 minutes. Cells were washed twice with phosphate buffered saline (PBS) and collected by trypsinization. Pelleted cells were washed with PBS and re-suspended with 2% FBS in PBS. Relative cell staining was determined by the BD LSRFortessa Flow Cytometer (BD Biosciences, San Jose, CA), and results were analyzed using FlowJo Software (BD, Ashland, OR).
Cell Proliferation Assays
HCC cell proliferation rates were determined using 2 different methods: cell counting and Cell Proliferation enzyme-linked immunosorbent assay, BrdU (colorimetric) assay (Roche Diagnostics). For cell counting, HCC cells were plated onto 12-well culture dishes at the original density of 2 × 104 cells per well. Culture media were changed 24 hours after plating cells for pharmacological treatments. Cell counting began 24 hours after treatment. Cells were trypsinized and counted using the TC20 Automated Cell Counter (Bio-Rad Laboratories, Hercules, CA) daily for 4 consecutive days. Uncounted cells were replenished with fresh culture media with appropriate treatments every 24 hours. BrdU colorimetric assay was performed according to the manufacturer’s instructions. Briefly, HCC cells of equal density were seeded onto 96-well culture dishes. Pharmacological treatments and BrdU labeling were added 24 hours after cells were plated. Addition of conjugated anti-BrdU antibodies and substrate reactions were performed the next day, and absorbance for each sample was measured on a microplate reader to quantify the cell proliferation.
For orthotopic implantation, 1.5 × 106 luciferase-labeled MHCC97L cells were re-suspended in 100% Matrigel (BD Biosciences) and implanted into left lobes of livers of 6- to 8-week-old, male BALB/cAnN-nu (nude) mice. After 6 weeks (42 days) post implantation, tumor-bearing mice were subjected for bioluminescent imaging. The mice were injected intraperitoneally with 100 mg/kg D-Luciferin (Gold Biotechnology, St Louis, MO) then imaged using the PE IVIS Spectrum In Vivo Imaging System (PerkinElmer, Inc, Waltham, MA). Livers and lungs were also extracted for ex vivo imaging.
For HDTVi, plasmid DNA mixed with sterile saline (0.9% NaCl) at a total volume equivalent to 10% of body weights were injected into the lateral tail vein of 8- to 10-week-old, male C57BL/6N mice within 5 to 7 seconds. A total of 20 μg of CRISPR-Cas9 vector system carrying single guide RNA targeting Keap1 with transposon system carrying c-Myc vector were injected. For HDTVi model, fatty livers were induced by replacing standard diets with HFD 24 hours prior to HDTVi and for the duration of the experiment. FSP1 was inhibited in vivo by daily intraperitoneal injections of 10 mg/kg iFSP1 (Cayman Chemical). Administrations of anti-PD-1 mAb (clone RMP1.14, BioXcell) and anti-PD-L1 mAb (clone 10F.9G2, BioXCell) both began 1-week post HDTVi for 3 weeks, twice weekly by intraperitoneal injection at 10 mg/kg. CD40 agonist (clone FGK45, BioXCell) was administered 2 weeks post HDTVi by intraperitoneal injections at 2.5 mg/kg once per week. To study the effect of iFSP1 on tumor immune infiltrates, HDTVi-induced tumors were harvested 4 weeks post HDTVi for dissociation and histological analysis. To study the therapeutic benefits of combination treatment of iFSP1 and different immunotherapies, mice were sacrificed when they have reached humane endpoint.
All animal procedures performed throughout this study were approved by the Committee on the Use of Live Animals in Teaching and Research of The University of Hong Kong and adhered to the Animals (Control of Experiments) Ordinance of Hong Kong. All animal procedures were performed under the UK Co-ordinating Committee on Cancer Research (UKCCCR) PMID: 9459138 Guidelines for the Welfare of Animals in Experimental Neoplasia to minimize suffering on the animals throughout the experiments.
Quantitation of relative 4-HNE in HCC cells were detected with the Lipid Peroxidation (4-HNE) Assay Kit (Abcam, Cambridge, UK) according to the manufacturer’s protocol. Briefly, samples and standards were incubated with anti-4-HNE antibody followed by secondary antibody-HRP conjugate on a 4-HNE conjugate pre-coated plate. After incubation with substrate, enzymatic reactions were stopped with a stop solution, and the absorbance for each sample mixture was measured on a microplate reader at OD 450 nm.
Malondialdehyde Colorimetric Assay
Relative productions of MDA in HCC cells were detected using the Lipid Peroxidation (MDA) Assay kit (Abcam) following the manufacturer’s instructions. Briefly, HCC cells were harvested and lysed by homogenization in lysis buffer. Homogenized samples and standards were incubated with thiobarbitoric acid. Absorbance measurement of each sample mixture was performed on a microplate reader at OD 532 nm.
Lactate Dehydrogenase Release Measurement
LDH released from damaged HCC cells were measured with Cytotoxicity Detection KitPLUS (LDH) (Roche Diagnostics) according to the manufacturer’s instructions. Briefly, equal densities of HCC cells were plated onto 6-well culture dishes for 24 hours followed by pharmacological treatments where applicable. Conditioned medium from HCC cells were collected, mixed with reaction mixture, and incubated at room temperature in dark for 30 minutes. Reactions in each sample mixture were stopped, and absorbance was measured with microplate reader at OD 490 nm.
Cell Death Analysis
The percentage of dead cells was determined in equal densities of HCC cells plated on 6-well culture dishes for 24 hours followed by pharmacological treatments where applicable. Conditioned medium and HCC cells were collected. The pelleted cells were washed with PBS and re-suspended in SYTOX Green dead cell stain (Invitrogen) at room temperature in the dark for 40 minutes. Dead cell percentages were determined by flow cytometry using the BD LSRFortessa Flow Cytometer (BD Biosciences).
Preparation of Bone Marrow-derived Macrophages
Femur and tibia were obtained from 5- to 7-week-old C57/BL6N male mice. Bones were cut at both ends, and bone marrow was flushed out with medium. Bone marrow cells were isolated using Ficoll Paque PLUS (GE Healthcare) according to manufacturer’s instruction. Cells were then cultured for 6 days in RPMI 1640 medium supplemented with 25 mM HEPES, 50 μM 2-mercaptoethanol (Sigma-Aldrich), 10% FBS, and 1% P/S in the presence of 20 ng/mL M-CSF (R&D Systems, Minneapolis, MN). More than 90% of the adherent cells were macrophages as identified by CD11b+ CD11c- F4/80+.
BMMs were seeded at 1 × 105 per well in 12-well plate. Medium was changed the next day. Hepa1-6 cells were treated with iFSP1 at indicated concentration with or without Fer-1 for 24 hours. Hepa1-6 cells with or without treatment were labeled with 2.5μM CTV (Invitrogen) for 20 minutes at 37 °C with mixing at regular intervals. Labeled Hepa1-6 cells were co-cultured with macrophages in 5:1 ratio for 1 hour at 37 °C. Macrophages were harvested and stained for macrophage markers CD11b and F4/80. Phagocytosis events were analyzed by flow cytometry.
Preparation of Single-cell Suspension From Tumor Tissues
Single-cell suspensions from mouse tumors were prepared prior to flow cytometry analysis. Tumors harvested from HCC-bearing mice were cut into fine pieces and transferred to gentleMACS C tube (Milteyni Biotech, Bergisch Gladbach, Germany) containing 10 mL of DMEM/F12 medium (Gibco). Thirty μL of Liberase (2.5 mg/mL; Roche Diagnostics) and 60 μL of DNase I (10 mg/mL; Roche Diagnostics) were added. Samples were further dissociated with GentleMACS Dissociator (Miltenyi Biotech) using the installed programs “h_tumor_01” thrice, “h_tumor_02” twice, and “h_tumor_03” once. Samples were incubated at 37 °C for 15 minutes with regular mixing in between 2 programs. The cell suspensions were then passed through MACS SmartStrainer (Miltenyi Biotech) to remove undigested tissues and cell clumps, followed by centrifugation. The filtered cell suspensions were resuspended with 10 mL ACK lysing buffer (155 mM ammonium chloride, 10 mM potassium bicarbonate, and 0.1 mM EDTA in water) and incubated at room temperature for 5 minutes to lyse the red blood cells. The reaction was stopped by adding 2 volumes (20 mL) of cell staining buffer (0.5% w/v BSA and 2 mM EDTA in PBS). The cell suspensions were filtered again with MACS SmartStrainer and centrifuged. The cell pellets were then resuspended in proper amount of cell staining buffer and were ready for cell counting. All centrifugations were carried at 300 × g for 5 minutes.
Flow Cytometry Analysis
Cells were blocked with anti-mouse CD16/32 Antibody (Biolegend, San Diego, CA) at room temperature for 10 minutes and incubated with fluorochrome-conjugated primary antibody at 4 °C for 30 minutes. Antibodies were used in 1:100 dilution. The cells were washed with 500 μL cell staining buffer once. They were then resuspended in 300 μL cell staining buffer and ready for flow cytometry analysis with the analyzer BD LSRFortessa (BD Biosciences). Primary antibodies used were listed in Table 2. Software FlowJo was used for data analysis.
Single-cell RNA Sequencing
After obtaining single cell suspension from mouse tumors as described above, cells were purified with Dead Cell Removal Kit (Miltenyi Biotech). Purified cells were loaded on the 10× Chromium system and were encapsulated with Single Cell 5’ Kit (10× Genomics, Pleasanton, CA). Single-cell gene expression libraries were generated according to manufacturer’s protocol.
ScRNA-seq data processing and batch effect correction
The generated scRNA-seq FASTQ files was aligned to mm10 reference by Cell Ranger 3.0.2. First, the raw data files of barcodes, features, and matrix were imported and created as a Seurat object. Next, we counted the mitochondrial percentage of each cells, and each data of samples was normalized by the sctransform function
They were subsequently merged as a Seurat object. Then, the batch effect was corrected by Fast Mutual Nearest Neighbors (FastMNN) function. To confirm the correction and visualize the data, the non-linear dimension reduction method tSNE was applied using Seurat function RunTSNE with maximum 30 dimensions.
Major cell type annotation and cluster-specific marker gene identification
To annotate each cell with cell type, we first clustered the cells using Seurat function FindNeighbors and FindClusters with ‘resolution’ parameter from 0.1 to 1. We applied R-package Clustree to establish tree for visualizing the clustering results with different ‘resolution’ parameter. ‘Resolution’ of 0.8 was used in subsequent analysis. Next, we used another R-package SingleR to annotate the cell types based on the transcriptome of each cell with reference to ImmGen database. Then, we used the Seurat FindMarkers function to identify the marker genes of each cluster. We defined the marker genes by: (1) 0.25 increasing logfc and (2) P-value less than .05 based on the Wilcoxon rank sum test. After that, the ribosomal and mitochondrial genes were filtered out. We identified 9 major immune cell types (B cells, DCs, macrophages, monocytes, neutrophils, NK cells, pDCs, T cells, and mast cells). The dimensionality of the clustering result was reduced and visualized by tSNE.
Identification of T cell subsets and DC/monocyte/macrophage subsets
To identify the subsets of T cells and DC/monocyte/macrophage, raw data from T cells or monocytes, macrophages, DCs, and pDCs were extracted respectively. A second round of clustering was performed using the same procedures and parameters as the identification of major cell types described above. For T cell subsets, we identified 2 CD4+ T cell populations (CD4-Th and CD4-Tcm) and 4 CD8+ T cell subsets (CD8-Trm, CD8-Proliferating, CD8-Tex and CD8-Tcm). For DC/monocyte/macrophage subsets, we identified 2 macrophage subsets (Mac-1, Mac-2), 3 monocyte subsets (Mono-1, Mono-2, Mono-3), and 4 DC subsets (cDC1, cDC2, cDC3, and pDC). The subset clustering results were mapped to the tSNE plot of major cell types for visualization.
Comparison of cell type proportion in iFSP1 and vehicle-treated tumor
The percentage of cell types in iFSP1 and vehicle-treated tumor was calculated, respectively, based on the total cell numbers of each sample. The percentage of T cell subsets and DC/monocyte/macrophage subsets were normalized to the total number of cells extracted for the sub-clustering. The 95% confidence intervals were calculated by the freqCI function of R-package REdaS.
Software used in the scRNA-seq analysis are listed in Table 5.
All statistical analyses were performed using the GraphPad Prism 8.0 or above software (GraphPad Software, Inc, La Jolla, CA). Data were analyzed by the 2-tailed Student t test, the log-rank test, or as indicated otherwise. All functional assays are representations of at least 3 independent experiments and are expressed as mean ± standard deviation (SD). P values < .05 were considered statistically significant.
The authors thank the Centre for Comparative Medicine Research (CCMR) of The University of Hong Kong in support of our animal experiments, and the School of Biological Sciences, The University of Hong Kong for their technical support in our mass spectrometry experiment. We also thank the Genomics and Bioinformatics Cores, and the Imaging and Flow Cytometry Core of the Centre for PanorOmic Sciences (CPOS) from the Li Ka Shing Faculty of Medicine, The University of Hong Kong for their technical support in our single-cell RNA sequencing and in vivo imaging experiments, respectively.
CD40 agonistic monoclonal antibody APX005M (sotigalimab) and chemotherapy, with or without nivolumab, for the treatment of metastatic pancreatic adenocarcinoma: an open-label, multicentre, phase 1b study.
Conflicts of interest The authors disclose no conflicts.
Funding This study was supported by the Hong Kong Research Grant Council (RGC) General Research Fund (17112521) and RGC Theme-Based Research Fund (T12-704/16-R). Carmen Chak-Lui Wong is a recipient of the Croucher Innovation Award and The University of Hong Kong Outstanding Young Researcher Award.