Evidence of pyroptosis and ferroptosis extensively involved in autoimmune diseases at the single-cell transcriptome level

The scRNA-seq datasets of 5 healthy donor skin samples and 8 vitiligo-affected skin samples were acquired from the Genome Sequence Archive (GSA) with accession number PRJCA006797 (https://​ngdc.​cncb.​ac.​cn/​bioproject/​browse/​PRJCA006797) (Table 1). The scRNA-seq datasets of lesional/non-lesional skin biopsies of AD patients were obtained from GEO database (Accession NO. GSE147424). The datasets of skin biopsy tissues from 13 psoriasis patients and 5 healthy volunteers were acquired from GSE151177. Single nuclei RNA-sequencing (snRNA-seq) datasets of brain samples of control and MS patients were obtained from GSE118257. And the droplet-based scRNA-seq datasets of human healthy and SSc-ILD lungs were got from GSE128169. The scRNA-seq datasets of terminal ileum of childhood onset Crohn’s disease and matched healthy controls were downloaded from https://​cellgeni.​cog.​sanger.​ac.​uk/​gutcellatlas/​pediatric_​RAWCOUNTS_​cellxgene.​h5ad. And the scRNA-seq datasets of unilateral decapsulated testes from mouse experimental and control group were obtained from GSM5563668 and GSM5563669.

Cells from psoriasis patients and healthy volunteers were filtered with a gene expression number per cell between 200 and 10,000, and the mitochondrial percentage per cell was below 50. Cells from the vitiligo-affected skin samples and healthy skin samples were filtered with a gene expression number per cell between 200 and 10,000, and the mitochondrial percentage per cell was below 25. And cells from lesional/non-lesional skin biopsies of AD patients were filtered with a gene expression number per cell between 200 and 10,000. Quality control of cells from brain samples was applied to filter out low-quality cells with < 200 or > 10,000 expressed genes. Cells from lung samples were filtered with a gene expression number per cell between 200 and 10,000, and the mitochondrial percentage per cell was below 25. The processed dataset of terminal ileum was converted into h5Seurat data, and used without any additional filtration. And quality control of cells from decapsulated testes was applied to filter out low-quality cells with either (1) < 200 or > 10,000 expressed genes or (2) < 25 mitochondrial percentage.

Datasets of vitiligo skin biopsies, psoriasis skin biopsies, AD skin biopsies, MS brain cells, SSc-ILD lung biopsies, decapsulated testes, and its counterparts, were processed in Seurat V4, integrated by “merge” function, and normalized by “NormalizeData” function. Then “FindVariableFeatures”, “ScaleData”, “RunPCA”, “FindNeighbors”, and “SCTransform” were used to process the datasets. Next, “RunHarmony (object, group.by.vars = ”orig.ident”)” and “RunUMAP(object, reduction = ”harmony” were used, while clusters were calculated using the “FindClusters” function with a resolution of 0.5. The Seurat file of terminal ileum was used without any additional data processing.

Cluster of Melanocytes was characterized by the high expression of DCT, TYRP1, PMEL, and MLANA in vitiligo skin and its counterparts-derived cells. Ketatinocytes were identified by the high expression of KRT14, KRT1, KRT10, and KRT15. Fibroblasts were identified by feature genes including COL1A1, DCN, SFRP2, and TWIST2, while other cell types were identified by below genesets, (1) smooth muscles: TAGLN, ACTA2, and NR2F2; (2) ECs: AQP1, CLEC14A, PECAM1, and ECSCR.1; (3) Langerhans: PCGBP, CD207, CD1A, and S100B; (4) T cells: PTPRC, CD3D, CD3E, TRBC2, TRAC, and TRGC2; (5) myeloid cells: PTPRC, CD14, FCGR3A, and CD1C. Ketatinocytes of psoriasis skin biopsies and its counterparts were further identified according to S.Basale, S.Granulosum&S.Spinosum, and S.Corneum based on the expression of signature genes presenting on Fig. 1B. Clusters of AD skin biopsies and its counterpart-derived cells were characterized by feature genes described in Additional file 2: Fig. S2B, then renamed and merged into several major clusters by the “RenameIdents” function.

Clusters of OPCs were characterized by the high expression of SOX6, OLIG2, PDGFRA, and BCAN in brain cells. Neurons were identified by the high expression of SNAP25 and GABRB2. Oligodendrocytes were identified by the feature genes PLP1 and CNP. And astrocytes were characterized by the high expression of APOE, GJA1, and AQP4, while other cell types were identified by below genesets, (1) immune cells: PTPRC, CD14, and IGHM; (2) EC/VSM: CLEC14A and PECAM1.

Macrophages were identified by the high expression of CD14, FCGR3A, ITGAX, SPP1, APOE, FABP4, and IL1B in lung-derived cells. Club/Goblet/Basal cells were characterized by the high expression of CAPS and TPPP3. Alveolar Type 1/2 cells were identified by the high expression of SCGB3A1, SLPI, SFTPA1, and SFTPC. Other cell types were identified by below genesets, (1) ECs: CD34, FCN3, CLDN5, ACKR1, and VWF; (2) Fibroblasts: DCN and LUM; (3) Smooth Muscles/Pericytes: TAGLN and ACTA2; (4) Mast cells: TPSAB1 and TPSB2.

Epithelial cells featured the high expression of EPCAM and FABP1 in terminal ileum-derived cells. Fibroblasts were identified by the high expression of COL1A1, COL1A2, and DCN. And myeloid cells were identified by these feature genes of PTPRC, CD14, FCGR3A, ITGAM, and CSF1R. Clusters of decapsulated testes were characterized by those feature genes showed in Additional file 7: Fig. S7B.

The AddModuleScore function from Seurat v4.0 was used to score target cells according to the expression of signature gene lists. We grouped the expression values of human ferroptosis driver gene set (Additional file 8: Table S1) [30], constructed a signature that we named the “ferroptosis driver” module, and scored target cell types using the “AddModule Score” function. And we grouped the expression values of human ferroptosis suppressor gene set (Additional file 9: Table S2), constructed a signature that we named the “ferroptosis suppressor” module. What’s more, the mouse gene lists were by Additional file 10: Table S3 and Additional file 11: Table S4. And finally, we grouped the expression values of these pyroptosis-related genes (mouse: Casp1, Casp11, Casp6, Casp8, Gsdma1, Gsdma2, Gsdma3, Gsdmc, Gsdmd, and Gsdme; human: CASP1, CASP4, CASP6, CASP8, GSDMA, GSDMB, GSDMC, GSDMD, and GSDME), constructed a signature that was named the “pyroptosis” module. These geneset scores were showed as violin plots and box-plots.

B16 cell line was treated with mouse IFN-γ (5 ng/mL or 10 ng/mL) (C746, Novoprotein Scientific) and collected at different time points (12 h, 24 h, and 48 h). Total RNA was isolated from Trizol (15596026, Invitrogen) treated cells. cDNA was synthesized using HiScript II Q RT SuperMix for qPCR (+ gDNA wiper) kit (R223-0, Vazyme). qPCR was performed on cDNA using Hieff qPCR SYBR Green Master Mix (Low Rox Plus) (11202ES08, YEASEN). Gene expression was quantified using the following primers: mouse Gapdh forward: CAGTGGCAAAGTGGAGATTGTTG; mouse Gapdh reverse: TCGCTCCTGGAAGATGGTGAT; mouse Gsdmd forward: CACCATGGCCTCAATGTGCT; mouse Gsdmd reverse: GCAAGCCTTCACCTCAGCAT; mouse Gsdme forward: TGCAACTTCTAAGTCTGGTGACC; mouse Gsdme reverse: CTCCACAACCACTGGACTGAG; mouse Casp1 forward: GCCGTGGAGAGAAACAAGGAGTG; mouse Casp1 reverse: TCAATGAAAAGTGAGCCCCTGACAG; mouse Casp8 forward: TCCTGTGCTTGGACTACATCC; mouse Casp8 reverse: TTCCCGCAGCCTCAGAAATAG; mouse Gpx4 forward: ATAAGAACGGCTGCGTGGTGAAG; mouse Gpx4 reverse: TAGAGATAGCACGGCAGGTCCTTC; mouse Slc7a11 forward: TGGCGGTGACCTTCTCTGA; mouse Slc7a11 reverse: ACAAAGATCGGGACTGCTAATGA; mouse Slc3a2 forward: TGCTCAGGCTGACATTGTAGC; mouse Slc3a2 reverse: TCAGCCAAGTACAAGGGTGC. Fold changes in mRNA expression were calculated by the ΔΔCt method using Gapdh as an endogenous control. Results are expressed as fold change by normalizing to the controls. And the expression of Gsdmc, Casp6, Gsdma, Gsdma2, Gsdma3, Casp11, and Ascl4 was too low and didn’t present.

The THP-1 cells were cultured in RPMI1640 medium supplemented with 10% FBS and 1 × P/S. And THP-1 cells were seeded into 25cm² flasks at the density of 5 × 10⁵ cells/flask and were cultured with 10 ng/mL PMA (abs9107, absin) for 24 h. Then these cells were cultured with medium without stimulatory factor, or supplemented with IFN-γ (50 ng/mL)( abs04123, absin) and LPS (15 ng/mL)( L8880, Solarbio), or only IFN-γ(50 ng/mL) for further 5 days. RNA were extracted from 1 × 10⁶ THP-1 cells or THP-1-derived cells of each sample, and sequenced using RNA-seq platform at Guangzhou Huayin Health Medical Group. Sequencing was run on an Illumina HiSeqTM 2000, and DESeq2 was used for gene expression analysis.

Psoriasis is an autoimmune disease mediated by overactive IL17 secreting type 17 T-cells (such as Th17 and Tc17) and defective negative regulatory networks (such as dysfunctional Tregs). Many researchers have focused on immune-related factors in the development of psoriasis, but haveignored the cell-fate transition of hyperproliferative keratinocytes themselves [31]. As part of this work, scRNA-seq analysis of psoriasis and control skin-derived cells were performed to identify the cell fate transition of nonimmune subsets (Additional file 1: Fig. S1A). Interestingly, compared with the control groups, the psoriatic skin-derived keratinocytes and fibroblasts showed a lower ferroptosis driver gene score (Additional file 8: Table S1 and Additional file 1: Fig. S1B). Some important ferroptosis driver genes, such as RPL8 [32], NCOA4 [33], and ALOXE3 [34], are expressed at low levels in the keratinocyte population in psoriasis. In addition, some critical ferroptosis suppressor genes, such as GPX4 [35], FTL, and FTH1 [36], are expressed at high levels in the keratinocyte population in psoriasis (Fig. 1A). These differences implied that ferroptosis is involved in the development of psoriasis. Keratinocyte cells were subdivided into stratum corneum (S.C, with high expression of KRT23, KRT80, SPRR2G, LCE3D, and CDSN), stratum granulosum, stratum spinosum (S.G & S.S, with high expression of FABP5, KRT10, and KRT1), and stratum basale (S.B, with high expression of MKI67, KRT14, and KRT5) (Fig. 1B) [37]. Psoriasis skin-derived S.B cells expressed GPX4, FTL, and FTH1 at high levels, whereas control skin-derived S.G & S.S cells expressed high levels of GPX4, FTH1, NR4A1, NFE2L2, and MT1G (Fig. 1C). Moreover, control skin-derived S.B and S.G & S.S cells expressed RPL8 at high levels, and control skin-derived S.C cells expressed high levels of NCOA4 and ALOXE3 (Additional file 1: Fig. S1C). These evidences indicated the ferroptosis-related pathways changed in psoriasis skin, and the S.B cells were given a combat advantage of ferroptosis, which is consistent with the phenotype of apoptosis-resistant psoriatic keratinocytes [38, 39]. A recent study showed that the partial index of active ferroptosis-related cell death in bulk psoriasis skin cells was higher than that in bulk normal skin cells[40]. In addition, we distinguished the differences in ferroptosis-related gene expression in different cell types, especially in immune cell types (Additional file 1: Fig. S1B) and S.G&S.S cells (Fig. 1C). It is noteworthy that inconsistency between the transcriptome and protein levels at single-cell level must be taken into consideration [28].

Deng et al. identified that Streptococcus cysteine protease SpeB virulence factor can trigger keratinocyte pyroptosis by cleaving GSDMA [41]. Therefore, it is proposed that pyroptosis also participates in the process of autoimmune skin diseases. Results showed that psoriasis skin-derived keratinocytes, fibroblasts, and endothelial cells (ECs) had a lower pyroptosis geneset score than control skin cells, with the exception of macrophages (Additional file 1: Fig. S1D and Fig. 1D). GSDMA and GSDMC were highly expressed in S.C cells, whereas GSDMB was highly expressed in S.B cells (Fig. 1E). Interestingly, compared with psoriasis skin-derived S.C cells, control skin cells highly expressed pyroptosis-related genes (CASP4, GSDMA, and GSDMC), suggesting that psoriasis skin-derived S.C cells were not sensitive to pyroptosis. In contrast, psoriasis skin-derived S.B cells expressed high levels of pyroptosis-related genes, such as CASP1, CAPS8, GSDMA, GSDMB, and GSDMD. Further, CASP3 was downregulated in S.C and S.G&S.S cells of psoriatic skin. Together, these data show that pyroptosis participates in the imbalance between the sensitivity of pyroptosis and apoptosis in different keratinocyte lineages (Fig. 1F). However, this biological phenotype requires further validation in future studies.

AD, a common inflammatory skin disease, is primarily characterized by a type-II immune response that leads to skin barrier damage [42]. To investigate potential cell death modalities, AD-related scRNA-seq datasets obtained from GSE147424 was analyzed. No major differences were observed in the ferroptosis suppressor gene score, ferroptosis driver gene score, and pyroptosis gene score in the fibroblast population, keratinocyte population, and EC population from healthy skin biopsy, lesional skin biopsy, and non-lesional biopsy (Additional file 2: Fig. S2A). However, GSDMC and GSDMD were highly expressed in lesional skin-derived keratinocytes (Fig. 2A). It is noteworthy that nuclear programmed death ligand 1 (nPD-L1) can upregulate GSDMC, whereas TNF-α-activated caspase-8 can cleave GSDMC, which will switch apoptosis to pyroptosis in cancer cells [43]. TNF-α has been shown to play a role in the pathogenesis of allergic inflammation in AD [44]. This evidence demonstrated that TNFα-Caspase8-GSDMC mediated pyroptosis may trigger keratinocyte death and inflammation. Keratinocytes were segregated into nine subtypes (Additional file 2: Fig. S2B, C). In addition, based on the expression patterns of GSDMC and GSDMD, keratinocytes were divided into three subpopulations: GSDMC^hiGSDMD^low, GSDMC^hiGSDMD^hi, and GSDMC^lowGSDMD^hi (Fig. 2B). Further analysis revealed that lesional skin-derived keratinocytes were located in the GSDMC^hiGSDMD^hi subset (Fig. 2C). Accordingly, it is speculated that this unique transition of the pyroptosis-sensitive keratinocyte subpopulation may play a role in pyroptosisin AD.

Vitiligo is caused by the loss of epidermal melanocytes due to the activity of autoreactive CD8⁺ T cells. To clarify the potential mechanism leading to the loss of epidermal melanocytes, we analyzed scRNA-seq datasets of skin-derived cells from healthy donors and patients with vitiligo (Additional file 3: Fig. S3A). The melanocytes were defined as signature genes (DCT, TYRP1, PMEL, and MLANA) (Additional file 3: Fig. S3B). Further analysis of melanocytes revealed strong pyroptosis responses existed in patients with vitiligo (Fig. 3A).Results demonstrated that CASP1, CASP4, CASP6, CASP8, and GSDMD expression was significantly upregulated in melanocytes of patients with vitiligo (Fig. 3B). As previously reported, CASP4, CASP8, and CASP1 are involved in GSDMD-dependent pyroptosis [45‐47]. Caspase-6 promotes Z-DNA binding protein 1 (ZBP1)-mediated apoptosis, necroptosis and pyroptosis [48]. These results indicate that pyroptosis signaling is an important pathway in the development of vitiligo.

A ferroptosis-related analysis of melanocytes was conducted, in which only a small number of samples (1/10) showed a higher ferroptosis driver gene score than that of healthy donors (Additional file 3: Fig. S3C). Nevertheless, several important ferroptosis driver genes, such as ASCL4 and TFRC, were upregulated in the patient samples (Additional file 3: Fig. S3D). Acyl-CoA synthetase long-chain family member 4 (ACSL4), a critical component of ferroptosis, is enriched in cellular membranes with long polyunsaturated ω6 fatty acids and increases the sensitivity of target cells to ferroptosis [49]. TFRC overexpression increases the intracellular iron pool and enhances lipid peroxidation. Notably, we found that melanocytes of patients (7/10) had a lower ferroptosis suppressor gene score than that of healthy donors, indicating that these melanocytes were sensitive to ferroptosis (Fig. 3C). Healthy donor-derived melanocytes stably expressed key anti-ferroptosis genes at high levels, such as GPX4, NR4A1, FTH1, FTL, MT1G, and SLC40A1 (Fig. 4D). Our results confirmed that both pyroptosis and ferroptosis are involved in the loss of melanocytes during vitiligo development.

The IFN-γ signaling pathway is activated in melanocytes of patients with vitiligo [50]. Therefore, B16 cell line was used as a cell model to explore the effects of IFN-γ signaling-mediated ferroptosis and pyroptosis (Fig. 3F). IFN-γ-mediated pyroptosis depends on the IFN-γ concentration (Fig. 3G). In general, IFN-γ induces pyroptosis susceptibility based on upregulation of GSDMD, GSDME, CASP1, and CASP8. Interestingly, IFN-γ exposure did not significantly reduce the expression of GPX4, SLC7A11, or SLC3A2, implying that other factors are involved in the increasing ferroptosis-sensitive state of melanocytes in patients with vitiligo. Taken together, IFN-γ is a critical factor that drives the pyroptosis-sensitive state, but not ferroptosis.

To investigate further the involvement of ferroptosis and pyroptosis in the pathology of autoimmune diseases, we compared the single-nucleus transcriptional profile of brain cells derived from MS patients and control individuals [51]. Although the two groups didn not show obvious differences in ferroptosis driver/suppressor gene scores and pyroptosis gene score (Additional file 4: Fig. S4), neurons had a lower ferroptosis suppressor gene score than other cell types (oligodendrocytes, astrocytes, immune cells, EC/VSM, and OPCs) (Fig. 4A). In addition, the neurons of MS patients expressed some important anti-ferroptosis genes (such as GPX4, FTH1, MT1G, and MTF1) at low levels (Fig. 4B). The same phenomenon was observed in some cell types, such as oligodendrocytes (GPX4, NFE2L2, FTH1, MT1G, SLC16A1, MTF1 etc.) (Fig. 4C), astrocytes (GPX4, NFE2L2, FTH1, MT1G, SLC40A1, MTF1 etc.) (Fig. 4D), OPCs (GPX4, SQSTM1, and FTH1) (Fig. 4E), and EC/VSM (GPX4, FTH1, MT1G, SQSTM1, SLC40A1, SLC16A1, and MTF1) (Fig. 4F). It has been observed that GPX4, as the most robust anti-ferrotposis functional gene, was commonly downregulated in MS patient-derived cell types, which convincingly demonstrated that ferroptosis is involved in the progression of MS. The downregulation of ferroptosis-resistant genes in MS brain cells leads to a tendency towards ferroptosis.

Systemic sclerosis (SSc) is an autoimmune disorder that can lead to interstitial lung disease (ILD). To explore the role of pyroptosis and ferroptosis in SSc-ILD, we analyzed the scRNA-seq dataset from GSE128169, which included eight SSc-ILD samples and five control samples. Analysis of pyroptosis-related driver genes in several major cell types revealed many significant changes between normal lungs and SSc-ILD samples. CASP1, CASP6, CASP8, GSDMB, GSDMC, and GSDMA were markedly upregulated in alveolar type 1&2 cells derived from SSc-ILD samples (Fig. 5A). Notably, group A Streptococcus (GAS) cysteine protease SpeB virulence factor can cleave GSDMA after Gln246, which triggers pyroptosis [41]. Upregulated GSDMA levels in patient with SSc may lead to pneumonia caused by Streptococcus pyogenes or other toxins. In addition, the club/gobelet/basal cells tended to upregulate the expression of CASP8, CASP1, GSDMB, and GSDMC in the SSc-ILD samples (Fig. 5B). Smooth muscle cells and pericytes in SSc-ILD lungs showed increased expression of GSDMD, DHX9, CASP4, CASP1, CASP6, and CASP8 compared to healthy cells (Fig. 5C). SSc-ILD lung-derived ECs exhibited high expression of DHX9, GSDMB, and GSDMC (Fig. 5D). Thus, increased GSDMB expression in alveolar type 1&2 cells, club/gobelet/basal cells, and ECs derived from SSc-ILD might increase the risk of pyroptosis mediated by the GZMA of autoactive CTLs.

To evaluate the role of ferroptosis in SSc-ILD, the ferroptosis suppressor/driver gene set score was identified and found no significant difference between healthy lungs and SSc-ILD lungs in several cell types (fibrobasts, alveolar type 1&2 cells, smooth muscle/pericytes, club/gobelet/basal cells, and ECs) (Additional file 5: Fig. S5A). Several studies have shown that fibroblasts play a key role in fibrotic ILD owing to abnormalities in aberrant extracellular matrix remodeling [52]. Accordingly, we analyzed the cell trajectory of fibroblasts, and identified two different branches with various effector gene expression patterns. Branch 1 expressed IL6 at high level, and branch 2 highly expressed profibrotic features, such as CXCL12 [53], VEGFA [54], IGF1 [55], and TGFB1 [56] (Additional file 5: Fig. S5B). This profibrotic branch mostly exists in SSc-ILD lungs, but not in healthy lungs, and expresses ferroptosis-related genes (ferroptosis-resistant genes: GPX4 and NR4A1 [57]; pro-ferroptotic genes: NCOA4 and SAT1 [58]); and pyroptosis drivers (CASP4 and GSDMD) (Fig. 5E). These results revealed that SSc-ILD profibrotic fibroblasts had high activity of ferroptosis and pyroptosis. Interestingly, SSc-ILD mast cells substantially expressed CASP4, CASP6, CASP8, DHX9, and GSDMD (Additional file 5: Fig. S5C), suggesting relationship between pyroptosis and mast cell degranulation.

Crohn’s disease (CD), a common form of inflammatory bowel disease (IBD), is characterized by irreversible aberrant immune responses [59]. Public scRNA-seq dataset of the terminal ileum of patients with CD and control donors was obtained from https://​www.​gutcellatlas.​org/​ (Additional file 6: Fig. S6A). First, we investigated the ferroptosis state of fibroblasts (Additional file 6: Fig. S6B), and confirmed that some proferroptosis genes, such as ACSL4, RPL8, SAT1, and CS [60], were upregulated (Fig. 6A). In addition, CD fibroblasts highly expressed some ferroptosis-resistant genes (such as FTH1, GPX4, NR4A1, NFE2L2, and FTL), but not others (such as MT1G and SLC40A1), indicating that pathological changes in fibroblast heterogeneity existed to some extent (Additional file 6: Fig. S6C). Consequently, the analysis of the cell trajectory of fibroblasts (Additional file 6: Fig. S6D) and revealed a unique differentiated cell state with ASCL4, CTSB, SAT1, and NFE2L2 expression at high levels (Fig. 6B). This differentiated cell state was consistent with SSc-ILD profibrotic fibroblast branch 2. Interestingly, this branch expressed SLC40A1 at low levels. These results highlighted that ferrotposis affected the balance of fibroblast differentiation. However there were no obvious differences in the pyroptosis-related genes between CD patient-derived fibroblasts and normal cells (Additional file 6: Fig. S6E, F).

The expression of multiple ferroptosis suppressors and driver genes in the epithelial cells (Fig. 6C, D) showed that epithelial cells derived from patients with CD expressed anti-ferroptosis genes (ZFP36 [61], HSPA5 [62], AKR1C3 [63], PLIN2 [64], FTL, FTH1, GPX4, and MT1G [65]) at lower levels than those of control cells, while pro-ferroptosis gene levels (RPL8 and MTDH [66]) were higher than those control cells. However, no clear evidence of upregulating pyroptosis drivers was identified (Additional file 6: Fig. S6G and Fig. 6E).

Next, it was necessary to investigatewhether pyroptosis and ferroptosis are involved in pathological myeloid differentiation. Myeloid cells in CD patients highly express anti-ferroptosis genes, such as GPX4, NFE2L2, FTH1, FTL, P4HB [67], and PRDX1 [68]. Moreover, compared to healthy myeloid cells, these myeloid cells expressed NCOA4 at low levels. Interestingly, SLC40A1 was highly expressed in healthy donor-derived myeloid cells (Fig. 6F). Cell trajectory analysis showed that macrophage M1 and M2 polarizations both existed in myeloid cells (Fig. 6G and Additional file 6: Fig. S6J). In addition, myeloid cells of patients with CD trended to polarize into the ferroptosis-resistant M1 state (Fig. 6H). SLC40A1 may also play a critical role in the maintaining of M2 state. SLC40A1 is highly expressed in tumor-associated macrophages, which suppresses the production of IL-1β [69], and is consistent with low IL-1β expression in control myeloid cells.

To identify the ferroptosis/pyroptosis patterns of macrophages under inflammatory conditions, as described CD inflammatory microenvironment, PMA, IFN-γ, and LPS were added to induce the differentiation of active macrophages and M1 macrophages (Fig. 6I(i)). Upon the PMA stimulation, THP-1 cells were induced into an activated state by secreting both M1 cytokines (IL18, IL1B, and TNF) and M2 cytokines (IL10 and TGFB1) (Fig. 6I(ii)). The cells were then differentiated into strong M1 states under IFN-γ and/or LPS culture conditions. Interestingly, with terminal M1 differentiation, THP-1-derived cells were more susceptible to pyroptosis owing to the upregulated expression of GSDMD, CASP4, CASP9, GSDMA, CASP5, GSDMC, and CASP8, but not GSDMB (Fig. 6I(iii)). The ferroptosis-related gene expression patterns of THP-1-derived M1-like cells were consistent with those of the above patterns of CD-related macrophages at scRNA-seq levels, such as SLC40A1 (Fig. 6I(iv)). THP-1-derived M1-like cells showed increased anti-ferroptotic activity (GPX4, NR4A1, FTH1, MT1G, SLC7A11, SLC3A2, and FTL) and decreased expression of NCOA4 and ALOXE3. Together, these results suggest that inflammatory conditions could drive the reversal of the sensitivity to ferroptosis and pyroptosis, and these properties could be targeted for treatment.

We elucidated the roles of ferroptosis and pyroptosis in six human autoimmune diseases. Next scRNA-seq dataset analysis of experimental autoimmune orchitis (EAO) (Additional file 7: Fig. S7A, B), which is a widely used as the mouse model of testicular inflammation [70], was conducted. Compared with other cell types, spermatids had a low ferroptosis suppressor gene expression score, which indicated that spermatids were sensitive to ferrotposis (Additional file 7: Fig. S7C). Except for spermatogonia/sertoli cell cluster, control sample-derived spermatids, spermatocytes, and Leyding cells/immune cells showed a high ferroptosis suppressor geneset score (Additional file 7: Fig. S7D). There was no significant difference in the ferroptosis driver gene score and pyroptosis gene score between EAO and control samples or different clusters (Additional file 7: Fig. S7E–H). In addition, GPX4 was downregulated in EAO cells (Additional file 7: Fig. S7I). Specifically, spermatids, Leydig cells/immune cells, spermatocytes, and spermatigonia/sortoli cells in EAO trended to have reduced Gpx4 expression (Additional file 7: Fig. S7J). These results confirmed that testicular cells were sensitive to ferroptosis under EAO conditions.