Asah2 Represses the p53–Hmox1 Axis to Protect Myeloid-Derived Suppressor Cells from Ferroptosis

Tumor-promoting inflammation and avoiding immune destruction are enabling characteristic and an emerging hallmark of cancer, respectively (1). A key component of the inflammation process is generation and activation of myeloid cells, such as neutrophils, monocytes, and macrophages (2, 3), which are the essential functional mediators of the inflammation process. Persistence of such inflammation drives tumorigenesis, particularly colon tumorigenesis in humans, for which a causal relationship to chronic inflammation has been well established (4). The chronic inflammatory tumor microenvironment in turn deregulates myeloid cell lineage differentiation and maturation, resulting in an accumulation of immature immunosuppressive myeloid cells, primarily the myeloid-derived suppressive cells (MDSCs), which not only directly enhance tumor growth and progression through secreting growth-promoting mediators (5, 6) but also suppress T and NK cell functions, thereby assisting tumors in avoiding destruction by the immune system (7, 8). Because of their potent immune suppressive activities and abundance in the tumor microenvironment, MDSCs are a major barrier to immunotherapies and are key targets in human cancer immunotherapy (9).

Myeloid cells are also a major source of programmed death ligand 1 (PD-L1) and therefore are major targets of PD-1 blockade immunotherapy (10, 11). However, MDSCs are heterogeneous in PD-L1 expression. Subsets of tumor-infiltrating MDSCs express PD-L1, whereas other subsets of tumor-induced MDSCs do not express PD-L1 (12). Therefore, current immune checkpoint blockade immunotherapies, such as anti–PD-1 immunotherapy, may only partially block MDSC-mediated immune suppression. This is consistent with the clinical observation that MDSC accumulation is correlated with resistance to nivolumab treatment in human cancer patients (13, 14).

To target MDSCs, various therapeutic approaches are actively being tested in preclinical and clinical studies. These include depletion or inhibition of their suppressive activity, differentiation, and accumulation (9). However, none of these strategies are specific to MDSCs, and some of them, such as chemotherapy, have side effects. The fact that MDSCs massively accumulate in the tumor-bearing host suggests that MDSCs could have dysregulated death pathways (15, 16). Several cellular death pathways have been identified to play important roles in MDSC survival and accumulation (17–22). More importantly, targeting the death receptor TRAILR 2 selectively with an agonistic Ab eliminated MDSCs in tumor-bearing mice and human cancer patients (23), demonstrating the significant clinical promise of targeting MDSC-specific death pathways for MDSC suppression. In this study, we identified ferroptosis, an iron-dependent accumulation of lipid reactive oxygen species (ROS) and resultant regulated cell death, as an MDSC death pathway and determined that MDSCs upregulate neutral ceramidase (N-acylsphingosine amidohydrolase [Asah2]), which catalyzes sphingolipid metabolism, to resist ferroptosis. Importantly, we developed an Asah2-selective small molecule inhibitor, NC06, that activates the p53–Hmox1 pathway to induce MDSC ferroptosis to suppress MDSC accumulation to activate T cell tumor infiltration to suppress tumor growth in vivo.

BALB/c and C57BL/6 mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Six- to twenty-week-old male and female mice were used. All mice were housed and maintained in accordance with an approved protocol by Augusta University Institutional Animal Use and Care Committees (2008-0162). J774M cells were generated and characterized as previously described (12). The mouse colon carcinoma CT26 cell line was obtained from American Type Culture Collection (Manassas, VA) in 2013 and stored in liquid nitrogen as aliquots. The mouse mesothelioma AB1 cell line was obtained from Sigma-Aldrich (St Louis, MO) in 2019 and stored in liquid nitrogen as aliquots. Cells were used within 30 passages. American Type Culture Collection characterized these cells by morphology, immunology, DNA fingerprint, and cytogenetics. Murine colon carcinoma MC38 cells were provided by Dr. J. Schlom (National Cancer Institute, Bethesda, MD) and characterized as previously described (24). Cells were cultured in RPMI 1640 (Invitrogen, Carlsbad, CA) plus 10% FBS(catalog no. SH30396.03; GE Healthcare, Chicago, IL). Cells are tested for mycoplasma every 2 mo, and all cells used in this study were mycoplasma-negative. Human colorectal carcinoma progression tissue microarray specimens were provided by the Cooperative Human Tissue Network (University of Virginia, Charlottesville, VA). Healthy donor blood specimens were collected from consented donors in Shepeard Community Blood Center (Augusta, GA). Blood specimens of colon cancer patients were collected from de-identified colon cancer patients in the Georgia Cancer Center. Blood samples were collected from patients without chemotherapy for at least 7 d. All studies that involved human specimens were approved by Augusta University Institutional Review Board (approval identifier: 1461188-1).

To generate spontaneous colon carcinoma tumor, C57BL/6 mice were treated with three cycles of azoxymethane (AOM)/dextran sodium sulfate (DSS) as previously described (6). To establish s.c. tumor models, CT26 cells (2 × 10⁵ cells per mouse) were injected into the right flanks of BALB/c mice. For experimental lung metastasis models, colon carcinoma CT26, mesothelioma AB1 (2 × 10⁵ cells per mouse) and mammary carcinoma 4T1 (2 × 10⁴ cells per mouse) were injected into BALB/c mice. Tumor-bearing mice were treated with vehicle PEG300 (Sigma-Aldrich) and NC06 (dissolved in PEG300), respectively, three to five times every 2 d by i.p. injection. Tumor size was measured in two dimensions with a digital micrometer caliper at the indicated time points. Tumor size was calculated by the following formula: (tumor length × tumor width²)/2. To quantify lung tumor nodules, lungs were inflated with 15% india ink solution and fixed in Fekete’s solution (58% ethanol, 8.7% formaldehyde, and 4.3% glacial acid) as previously described (25).

ASAH2 expression datasets in human primary tumor (n = 380) and normal colon tissues (n = 51) were extracted from The Cancer Genome Atlas (TCGA) Colon and Rectal Cancer (COADREAD) polyA⁺ IlluminaHiSeq pancan normalized RNA-sequencing datasets using University of California Santa Cruz Xena Cancer Genomics Browser (University of California, Santa Cruz, CA). The datasets were analyzed by the GraphPad program (GraphPad Software, San Diego, CA).

Structures in the National Cancer Institute Diversity Set V were supplied in two-dimensional form. Three-dimensional representations were generated with the computer program obprep from the Open Babel suite (26). Docking of these structures was done with the computer program AutoDock Vina 2.0 (27) (Scripps Research, La Jolla, CA) to predict binding modes of a drug candidate against the protein target using a box centered on the ligand binding site. The figure of merit in the scoring output of Autodock Vina is expressed as a binding affinity in kcal/mol, with an SE of 2.85 kcal/mol. Human Asah2 protein structure was used for molecular docking (28).

High-Resolution mass spectrometry was performed by a TSQ MS system (Thermo Fisher Scientific). For the Fourier-transform infrared (FTIR) spectrum, the compound was directly measured using the attenuated total reflection method on a Nicolet iS 5 FTIR Spectrometer (Thermo Fisher Scientific, Waltham, MA). The FTIR spectrum was measured between 4000 and 400 cm⁻¹ at room temperature. For nuclear magnetic resonance (NMR), the ¹H and [¹³C] experiments were recorded using Bruker-500 NMR spectrometer (Bruker, Leipzig, Germany) equipped with a 5-mm broadband observe probe. The NMR sample was prepared by dissolving 5 mg of nC06 in 500 μl DMSO-d6. The [¹H] and [¹³C] chemical shift values were reported on the δ scale in ppm relative to DMSO-d6 (2.50 ppm for [¹H] NMR and 39.52 ppm for [¹³C] NMR, respectively).

The ceramidase activity assay was based on the previously described method (29). Ceramide template RBM14C16 (0.1 mM; catalog no. 860856; Avanti Polar Lipids) were incubated in Asah2 buffer (50 mM HEPES, 150 mM NaCl, 1% sodium cholate [pH 7.4]) or acid ceramidase buffer (25 mM sodium acetate buffer [pH 4.5]) with recombinant mouse Asah2 protein, human ASAH2 protein, and human ASAH1 protein (catalog no. H00000427-P01; Abnova) as described above at 37°C for 3 h. Reactions were stopped by adding 25 μl/well methanol and then 100 μl/well NaIO4 solution (2.5 mg/ml in 100 mM glycine–NaOH [pH 10.6]) and incubated at 37°C for 1 h in the dark. Fluorescence was measured at excitation and emission wavelength of 355 and 460, respectively.

Small interfering RNA (siRNA) was obtained from Santa Cruz Biotechnology. Total RNA was isolated from cells using the GeneJET RNA Purification Kit according to the manufacturer’s instructions (Thermo Fisher Scientific). RNA concentrations were measured using a NanoDrop spectrophotometer (Thermo Fisher Scientific, Mansfield, TX). RNA purity was determined based on an OD260/OD280 ratio, and only samples with ratio >1.90 were used. cDNA was synthesized from total RNA using the ThermoScript RT-PCR Kit (catalog no. 11146-016; Invitrogen) with the substitution of ThermoScript Reverse Transcriptase by M-MLV Reverse Transcriptase (catalog no. M170B; Promega, Madison, WI). PCR was performed in triplicates in the StepOne Plus Real-Time PCR System (Applied Biosystems). Gene expression was determined as ΔΔCT with β-actin as internal normalization control.

Cells were treated with NC06 at 10 μM for 24 h for total RNA preparation. The Mouse Gene 2.0 ST Array (Affymetrix, Santa Clara, CA) was used for the gene expression profiling. Total RNA samples were processed using the Ambion WT Expression Kit (Life Technologies, Carlsbad, CA) and a GeneChip WT Terminal Labeling Kit (Affymetrix). The synthesized sense strand cDNAs were fragmented and biotin labeled using a GeneChip WT Terminal Labeling Kit. The labeled cDNAs were hybridized onto the arrays using Affymetrix GeneChip Fluidics Station 450 systems. The expression data were imported into Partek Genomics Suite version 6.6 using a standard import tool with GeneChip robust multiarray analysis normalization. The differential expressions were calculated using ANOVA of Partek package. The entire dataset is deposited to the National Center for Biotechnology Information Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) database under accession number GSE163399.

Bone marrow (BM) cells were isolated from BALB/c mice and cultured in RPMI 1640 medium with 10% FBS and recombinant mouse GM-CSF (40 ng/ml; BioLegend) for 3–6 d.

Endogenous ceramide was measured by Virginia Commonwealth University Lipidomics/Metabolomics Shared Resource using a high performance liquid chromatography–mass spectrometry approach. The ceramide level was normalized to the total phospholipid contents.

Single cells were prepared from the spleens of tumor-free BALB/c mice and used to purify CD3⁺ T cells with the MojoSort Mouse CD3 T Cell Isolation Kit (BioLegend) according to the manufacturer’s instructions. For the T cell proliferation assay, a 96-well culture plate was coated with anti-mouse CD3 (8 μg/ml) and anti-mouse CD28 mAbs (10 μg/ml) overnight. The purified T cells were labeled with CFSE (Life Technologies) and then seeded in the plate at a density of 1.5 × 10⁵ cells per well. J774M cells were added to the T cell culture at various rations and cultured for 72 h. Cells were analyzed by flow cytometry.

J774M and BM-MDSCs were treated with NC06, Z-DEVD, necrostatin-1 (Nec-1), and ferrostatin-1 (Fer-1) for various time. The cells were then stained with propidium iodide and annexin V in annexin V binding buffer (10 mM HEPES [pH 7.4], 140 mM NaCl, and 2.5 mM CaCl₂) and analyzed in a FACS Caliber (BD Biosciences, San Diego, CA). To analyze cells from tumors and spleens of mice, single cells were prepared from spleens and tumors as previously described (12). Cells were stained with DAPI (BioLegend) and mAbs that are specific for CD45.2, CD8, PD-1, CD11b, and Gr1 (BioLegend). Mouse BM and spleen cells were stained with CD11b- and Gr1-specific mAb. To analyze human MDSCs, PBMCs were prepared from peripheral blood samples of healthy donors and colon cancer patients and centrifuged at 3000 RPM for 10 min to collect the buffy coat. Red cells were lysed. The cells were stained with DAPI, CD11b-, CD33-, and HLA-DR5–specific mAbs (BioLegend), followed by intracellular staining using Alexa Fluor 647–conjugated anti-human neutral ceramidase Ab (Santa Cruz Biotechnolgy) and the Cytofix/Cytoperm Plus Intracellular Staining Kit (BD Biosciences). The stained cells were analyzed in an LSR Fortessa cytometer (BD Biosciences). Data were analyzed with the FlowJo v10 program (FlowJo, Ashland, OR).

Human colon carcinoma specimens were stained with anti-Asah2 (Santa Cruz Biotechnology). The AOM–DSS–induced mouse colon carcinoma tissues were stained with anti-Asah2. The tissues were then probed using a mouse-specific immunohistochemistry polymer detection kit (catalog no. ab209101; Abcam) according to the manufacturer’s instructions. The stained tissues were counterstained with hematoxylin (Richard-Allan Scientific, Kalamazoo, MI).

All statistical analysis was performed using SAS 9.4 (SAS Institute, Cary, NC), and statistical significance was assessed using an α level of 0.05. A repeated-measures mixed model with fixed effects of treatment groups and day, as well as all two- and three-factor interactions between these effects, was used to examine changes in tumor size over time. A one- or two-factor ANOVA was used to examine the fixed effects of F5446 and anti–PD-1 groups and the two-factor.

ASAH2 expression datasets were extracted from the TCGA database. Comparison of datasets from human colorectal carcinoma to datasets from normal colon tissues revealed that ASAH2 mRNA levels are significantly higher in tumor tissues than in the normal colon (Fig. 1A). Immunohistochemical analysis of human and the AOM–DSS–induced mouse colon carcinoma tissues validated that the Asah2 protein level is highly expressed in tumor cells (Fig. 1B). However, ASAH2 protein is also detected in tumor-infiltrating immune cells in both human and mouse colon carcinoma (Fig. 1B). MDSCs are a major subset of immune cells in the tumor microenvironment. To determine ASAH2 expression level in MDSCs, PBMCs were prepared from healthy donors and colon cancer patients. CD11b⁺CD33⁺HLA-DR5⁻ MDSCs were analyzed for ASAH2 protein level by intracellular staining (Fig. 1C). Quantification of ASAH2 protein level in these MDSCs indicates that the ASAH2 protein level is significantly higher in MDSCs from colon cancer patients than cells with the same phenotype from healthy donors (Fig. 1D).

One of the functions of Asah2 is to regulate survival of colon tumor cells (30). Our above finding that Asah2 is overexpressed in tumor-infiltrating MDSCs indicates that Asah2 is potentially a molecular target for MDSC suppression. To target Asah2, we capitalized on the recently determined human ASAH2 protein structure (28). We used the molecular docking program AutoDock Vina (27) to screen the National Cancer Institute Small Molecule Repository for ASAH2 inhibitors. The top 23 hits were selected and then tested for inhibition of human ASAH2 protein enzymatic activity in vitro. The compound with the highest inhibitory activity was characterized for its chemical structure. A high-resolution mass spectrum with low mass error was acquired. The high-resolution mass spectrum of the compound showed that the molecular ion constitutes the base peak ([M+H]⁺) at m/z = 324.9778 (Supplemental Fig. 1A). Using FTIR, the presence of major functional groups in the compound was also confirmed (Supplemental Fig. 1B). A broad absorption peak at 3075 cm⁻¹ represented C–H bond vibration. In addition, a strong and broad characteristic absorption corresponding to the C = O stretch was exhibited at 1725 cm⁻¹. An absorption peak was also observed at 3279 cm⁻¹, which might be assigned to the stretching vibration of N–H. Supplemental Fig. 1C, 1D shows the [¹H] and [¹³C] spectra of this compound, respectively. Thirteen distinct [¹³C] peaks were observed in the [¹³C] NMR spectrum. Combining the analysis results of FTIR, [¹H] and [¹³C] NMR, and high-resolution mass spectrometry, the chemical structure of the synthetic compound (97-chloro-2-(3-chloroanilino) pyrano[3,4-e][1,3]oxazine-4,5-dione) is confirmed and named as NC06 (Fig. 2A).

Molecular docking of NC06 revealed that NC06 has a high binding affinity to the catalytic site of human ASAH2 protein. In vitro enzymatic activity inhibition assays with recombinant human proteins determined that NC06 has an IC₅₀ of 25.91 μM for human ASAH2 and 30.20 μM for mouse Asah2, depending on the ceramide substrates (Fig. 2B). Acid ceramidase (ASAH1) is another ceramide that catalyzes ceramide metabolism (31, 32) However, NC06 exhibited no significant inhibitory activity against human acid ceramidase ASAH1 protein (Fig. 3), demonstrating selectivity for ASAH2. As expected, inhibition of Asah2 with NC06 increased ceramide accumulation in MDSCs (Supplemental Fig. 2).

To determine the function of Asah2 in MDSC survival, we made use of the mouse MDSC-like cell line J774M (12). J774M cells has a CD11b⁺Gr1⁺ phenotype (Fig. 4A) and exhibit potent suppressive activity against T cell activation and proliferation (Fig. 4B, 4C). NC06 treatment induced J774M cell death in a dose-dependent manner (Fig. 4D). A complimentary approach was then used to validate Asah2 function in MDSC survival. MDSCs were induced from BM cells using GM-CSF. The BM-MDSCs were then treated with NC06. Similar to the results observed in J774M cells, NC06 induced BM-MDSC cell death in a dose-dependent manner (Fig. 4E).

MDSC turnover is regulated by multiple cellular death pathways. Thus, we next sought to determine which cell death pathway(s) are regulated by Asah2. J774M cells were treated with NC06 in combination with inhibitors for apoptosis, necroptosis, and ferroptosis. Only inhibition of ferroptosis significantly decreased NC06-induced J774M cell death (Fig. 5A). A similar phenomenon was observed in BM-MDSCs (Fig. 5B). Ferroptosis is a recently identified metabolic stress-induced iron- and lipid peroxidation-dependent regulated cell death that is caused by overproduction of the toxic lipid ROS (33, 34). Glutathione (GSH) is a key substrate for GSH peroxidase 4–mediated detoxification of lipid hydroperoxide and production of ROS to protect cells from ferroptosis (33, 35). Consistently, treatment of J774M and BM-MDSCs with NC06 significantly decreased GSH level (Fig. 5C) and increased lipid ROS levels as measured by BODIPY [¹¹C] staining (Fig. 5D) in these MDSCs. Strikingly, although GSH levels were significantly decreased in NC06-treated cells, uptake of cystine, the precursor for GSH, was also decreased by NC06 treatment (Supplemental Fig. 3), suggesting that NC06 targets GSH synthesis downstream of cysteine. Taken together, our data indicate that Asah2 regulates ferroptosis in MDSCs.

We next determined whether the above observation that inhibiting Asah2, thereby inducing MDSC death, can be extended to MDSC suppression in vivo. Colon carcinoma cells (CT26) were transplanted to mice to establish colon tumors. We first validated MDSC accumulation in CT26 tumor-bearing mice. Analysis of spleen cells (Fig. 6A) and BM cells (Fig. 6B) indicates that the level of CD11b⁺Gr1⁺ MDSCs is significantly higher in the spleen and BM of the tumor-bearing mice than in the tumor-free mice (Fig. 6A, 6B). Because there is a correlation between the MDSC accumulation level and tumor sizes (36–38), we allowed mice to develop an extensive tumor burden before NC06 therapy. In this way, NC06 is not expected to change tumor size significantly, but it will enable us to unmask the direct effect of NC06 on MDSCs. NC06 therapy did not significantly decrease the size of the established tumors (Fig. 6C). However, NC06 therapy significantly increased MDSC cell death in the tumor-bearing mice in vivo (Fig. 6D). Consistent with increased MDSC turnover, NC06 therapy significantly decreased MDSC accumulation in tumor-bearing mice (Fig. 6E, 6F). Decreased MDSC accumulation is correlated with increased CD8⁺ CTL tumor infiltration (Fig. 6G). The majority of these tumor-infiltrating CTLs are also PD-1⁺ (Fig. 6H). We therefore conclude that targeting Asah2 is an effective strategy to suppress MDSC accumulation to increase CTL tumor infiltration and activation in tumor-bearing mice.

The above finding that Asah2 regulates ferroptosis is not entirely surprising because ceramide is known to regulate various cell death pathways (39–45). In addition, ferroptosis is caused by lipid peroxidation (46), and Asah2 regulates sphingolipid ceramide metabolism (47). To identify the molecular link between Asah2 and ferroptosis, we then performed genome-wide gene expression profiling in MDSCs. J774 M cells were treated with NC06 and the cells were analyzed by DNA microarray. Gene Set Enrichment Analysis (GSEA) of the differentially expressed genes between control and NC06-treated cells revealed that the most dominant enrichment is in genes of the p53 pathway based on gene ontology terms (Fig. 7A, 7B; the entire dataset is deposited to Gene Expression Omnibus database). Strikingly, the p53 mRNA level in J774M cells was not significantly altered by NC06 treatment (Fig. 7A, 7C). However, NC06 treatment increased the p53 protein level in a dose-dependent manner (Fig. 7D). Quantitative PCR (qPCR) analysis validated that NC06 regulates the expression of p53 pathway genes such as Slc7a11, Notch, and Hmox1 (Fig. 7E). Notably, two of the p53 pathway genes, Slc7a11 and Hmox1, are known to regulate ferroptosis (48, 49) and are regulated by Asah2 in ferroptosis in MDSCs (Fig. 7E). Consistent with the finding that inhibition of Asah2 with NC06 decreased GSH level (Fig. 5C), silencing Asah2 decreased GSH level in J774M cells (Fig. 7F and Supplemental Fig. 4). Furthermore, silencing p53 significantly increased GSH levels in J774M cells (Fig. 7F). Consistent with the GSH level changes, silencing Asah2 increased lipid ROS, and silencing p53 decreased lipid ROS levels in MDSCs (Fig. 7G).

Hmox1 degrades heme and releases free iron to generate oxidized lipids in the mitochondria membrane to induce ferroptosis (50, 51). The expression of Hmox1 is regulated by p53 (49, 52). Consistently, NC06 treatment increased Hmox1 expression in a dose-dependent manner in J774M cells (Fig. 7E). Silencing Hmox1 increased the GSH level (Fig. 7F) and decreased the lipid ROS level in MDSCs (Fig. 7G). Our data thus indicate the Asah2 represses Hmox1 expression to reduce lipid ROS to suppress ferroptosis.

Our above findings determined that Asah2 suppresses ferroptosis to promote MDSC survival and accumulation. MDSCs are not only a potent immune suppressor but also a promoter for tumor growth and progression (6, 53). Therefore, NC06 therapy may be expected to suppress tumor development. To test this hypothesis, 4T1 mouse mammary carcinoma cells were injected i.v. into syngeneic BALB/c mice, and the tumor-bearing mice were treated with NC06 in the early stage of tumor development. NC06 significantly decreased 4T1 tumor growth in the mouse lungs (Fig. 8A). To determine whether this efficacy can be extended to other types of tumors, mouse colon tumor CT26 and mesothelioma AB1 cells were injected into syngeneic BALB/c mice. The tumor-bearing mice were then treated with NC06. As observed in the 4T1 tumor model, NC06 also significantly decreased colon and mesothelioma tumor growth in the lungs in the tumor-bearing mice (Fig. 8B, 8C). In summary, NC06 is an effective therapeutic agent in suppression of tumor growth in preclinical mouse models.

MDSCs are a major population of immune cells in the tumor microenvironment. In this study, we observe that colon tumor-infiltrating MDSCs overexpress Asah2. Asah2 is a key ceramidase that mediates sphingolipid metabolism (47) and regulates colon tumor cell apoptosis and autophagy (30). However, we observed that Asah2 protects MDSCs from ferroptosis. Our findings thus indicate that Asah2 function may be cell type-dependent and reveal a previously uncharacterized function of Asah2 in MDSC survival in the tumor microenvironment. Triggering ferroptosis is considered one of the best strategies to circumvent drug-resistant cancer (54) and to enhance the efficacy of T cell–based immunotherapy (55). In this study, we extend the ferroptosis-based anticancer target from tumor cells to MDSCs. MDSCs use dysregulation of ferroptosis as one mechanism of survival that allow them to accumulate in the tumor microenvironment. Accordingly, we show that induction of MDSC ferroptosis with a pharmacological agent effectively suppresses MDSC accumulation to increase tumor-infiltrating CTL activation, resulting in tumor suppression.

The tumor suppressor p53 is a master regulator of ferroptosis, and p53 can function as either a ferroptosis suppressor or activator, depending on the level of oxidative stress in tumor cells (34, 51). Under cellular stress, p53 represses the expression of SLC7A11 to suppress cystine uptake and decrease extracellular cysteine, the rate-limiting precursor for GSH that is essential for GPX4 function in suppression of lipid peroxidation and lipid ROS generation (48, 56). In contrast, p53 can activate p21 transcription to delay the onset of ferroptosis in response to subsequent cystine deprivation in tumor cells (34). P53 also can directly regulate Hmox1 expression (49), and Hmox1 can induce Fe²⁺ overproduction and resultant lipid ROS (50, 51, 54). In this study, we observe that inhibiting Asah2 induces ferroptosis in MDSCs. However, inhibiting Asah2 did not alter the p53 expression level in MDSCs. Instead, genome-wide gene expression profiling reveals that inhibiting Asah2 activates the p53 pathway, suggesting that Asah2 may regulate p53 stabilization to suppress the p53 pathway in MDSCs (34), which requires further study. Although inhibiting Asah2 significantly decreases the GSH level and increases lipid ROS levels in MDSCs, cystine uptake actually increases after inhibition of Asah2, suggesting that Asah2 acts on a target between extracellular cysteine synthesis and GSH synthesis. It is known that Hmox1 regulates GSH synthesis (57). It is therefore likely that the Asah2–p53–Hmox1 axis regulates GSH synthesis downstream of cysteine and that Hmox1 represses GSH synthesis. The precise mechanism underlying Asah2 repression of GSH synthesis while enhancing cysteine uptake requires further study.

We propose that the canonical cystine–Slc5a11–GSH–lipid ROS pathway (48) protects MDSCs from ferroptosis to accumulate under pathological conditions such as cancer. MDSCs overexpress Asah2 to degrade sphingolipid ceramide (46) to decrease membrane lipid peroxidation potential. Targeting Asah2 stabilizes p53 to activate the p53 pathway, including upregulation of Hmox1, a gene that causes overproduction of Fe²⁺ (50, 57). Overproduction of Fe²⁺ may not only repress GSH synthesis to decrease GSH (N. L. Jenkins, S. A. James, A. Salim, F. Sumardy, T. P. Speed, M. Conrad, D. R. Richardson, A. I. Bush, and G. McColl, manuscript posted on bioRxiv, DOI: 10.1101/594408) to increase lipid ROS (58, 59), but may also enhance membrane ceramide lipid peroxidation and the resultant lipid ROS (60). These Homx1-mediated two lipid ROS generation pathways lead to ferroptosis of MDSCs.

A limitation of this study is the use of the J774M cell line as an MDSC model. Although we determined that J774M cells phenotypically and functionally resemble MDSCs in in vitro assays, it cannot completely recapitulate the biological aspects of tumor-induced MDSCs in vivo. More studies are needed to determine the function of Asah2 in ferroptosis and MDSC accumulation in the tumor microenvironment under physiopathological conditions. Nevertheless, our findings that MDSCs from human colon cancer patients overexpress Asah2 and that Asah2 protects MDSCs from ferroptosis indicate that Asah2 is a molecular target for suppression MDSCs. Accordingly, the Asah2-selective small molecule inhibitor NC06 effectively suppresses MDSC accumulation, thereby facilitating an increase in CTL tumor infiltration in colon tumor-bearing mice. Furthermore, NC06 therapy suppresses colon carcinoma, mammary carcinoma, and mesothelioma in preclinical mouse models. NC06 is a novel MDSC-targeting agent, and the encouraging results reported in this study further support its development and evaluation in cancer immunotherapy.

We thank Dr. Kimya Jones at Georgia Esoteric and Molecular Labaratory (Augusta, GA) for assistance in immunohistochemical staining and Dr. Thomas Albers for assistance in molecular docking. We thank Dr. Jeremy Allgood at the Lipidomics/Metabolomics Shared Resource, Virginia Commonwealth University for assistance in ceramide analysis.

The authors have no financial conflicts of interest.