SENP3 sensitizes macrophages to ferroptosis via de-SUMOylation of FSP1

Abstract

Ferroptosis, driven by an imbalance in redox homeostasis, has recently been identified to regulate macrophage function and inflammatory responses. SENP3 is a redox-sensitive de-SUMOylation protease that plays an important role in macrophage function. However, doubt remains on whether SENP3 and SUMOylation regulate macrophage ferroptosis. For the first time, the results of our study suggest that SENP3 sensitizes macrophages to RSL3-induced ferroptosis. We showed that SENP3 promotes the ferroptosis of M2 macrophages to decrease M2 macrophage proportion in vivo. Mechanistically, we identified the ferroptosis repressor FSP1 as a substrate for SUMOylation and confirmed that SUMOylation takes place mainly at its K162 site. We found that SENP3 sensitizes macrophages to ferroptosis by interacting with and de-SUMOylating FSP1 at the K162 site. In summary, our study describes a novel type of posttranslational modification for FSP1 and advances our knowledge of the biological functions of SENP3 and SUMOylation in macrophage ferroptosis.

Keywords: FSP1; Ferroptosis; Macrophage; SENP3.

Graphical abstract

Fig. 1

SENP3 sensitizes macrophages to RSL3-induced ferroptosis. (A–B) RAW 264.7 macrophages were incubated with vehicle, RSL3 (1 μM), and RSL3 (1 μM) + Fer-1 (1 μM) for 5 h. Then, the viability (A) and the Lactate dehydrogenase (LDH) release (B) of RAW 264.7 macrophages were tested. (C) RAW 264.7 macrophages were stimulated with RSL3 (1 μM) for the indicated time, and SENP3 and SUMO2/3 modification were measured by IB. (D–F) RAW 264.7 macrophages stable knockdown of SENP3 or not were constructed by sh-SENP3 and sh-NC. These cells were then incubated with vehicle, RSL3 (1 μM), and RSL3 (10 μM) for 5 h. Next, the viability (D) and Lactate dehydrogenase (LDH) release (E) of RAW 264.7 macrophages were tested. SENP3 expression in sh-NC and sh-SENP3 RAW 264.7 macrophages were validated by IB (F). (G–I) RAW 264.7 macrophages were stably overexpressing SENP3 or empty vector as control. SENP3 overexpressing and control RAW 264.7 macrophages were then incubated with vehicle or RSL3 (1 μM) for 5 h. Next, the viability (G) and LDH release (H) of these cells were tested. Western blots validated SENP3 expression in control and SENP3 RAW 264.7 macrophages (I). (J–M) Naïve bone marrow derived macrophages (BMDM) from SENP3^fl/fl and SENP3^cko (SENP3^Lyz2−/- mice) were treated with vehicle, RSL3 (1 μM), and RSL3 (10 μM) for 5 h. Then, the viability (J) and LDH release (K) of these cells were tested. SYTOX Green staining of SENP3^fl/fl and SENP3^cko BMDMs (L) and statistics analysis of dead/total cells% were shown (M). (N–P) SENP3^fl/fl and SENP3^cko BMDMs were incubated with RSL3 (1 μM) for 1 h. Lipid peroxidation was evaluated by BODIPY 581/591 C11 staining (N) and statistics analysis of mean FITC (O). Western blots validated the expression of SENP3 in SENP3^fl/fl and SENP3^cko BMDMs (P). Scale bars, 100 μm. Data are shown as means ± SD. Data are representative of three independent experiments. ****p < 0.0001, **p < 0.01, *p < 0.05, One-way ANOVA (A, B). ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, Multiple unpaired t-tests (D, E, G, H, J, K, O). **p < 0.01, unpaired t-test (M).

Fig. 2

SENP3 promotes the ferroptosis of M2 macrophages. (A–D) BMDMs from SENP3^fl/fl and SENP3^cko mice were incubated with LPS (100 μg/ml; M1 macrophages) and IL-4 (25 ng/ml; M2 macrophages) for 24 h independently, followed with RSL3 for 5 h. Next, the viability (A) and LDH release (B) of M1 and M2 macrophages from SENP3^fl/fl and SENP3^cko mice were tested. SYTOX Green staining of M1 and M2 macrophages from SENP3^fl/fl and SENP3^cko mice (C) and statistics analysis of dead/total cells% were shown (D). (E–F) Sh-NC and sh-SENP3 RAW 264.7 macrophages were treated by IL-4 (25 ng/ml) for 24 h, followed by RSL3 for 5 h. Next, the viability (E) and LDH release (F) of sh-NC and sh-SENP3 M2 macrophages were tested. (G–J) Peritoneal macrophages in a mouse model of zymosan-peritonitis plus RSL3 were collected for flow cytometry test. Gating strategy of CD45⁺ CD11b + F4/80+ peritoneal macrophages (G). Representative flow cytometry images show CD206+ or CD80⁺ cells gated from CD45⁺ CD11b + F4/80+ peritoneal macrophages (H). Statistics analysis of CD45⁺ CD11b + F4/80 + CD206+ macrophages and CD45⁺ CD11b + F4/80+ CD80⁺ macrophages in each group (I, J). Scale bars, 100 μm. Data represents mean ± SD, n = 3 or 4 biologically independent samples. ****p < 0.0001, **p < 0.01, *p < 0.05, two-way ANOVA (A, B). ****p < 0.0001, *p < 0.05, Multiple unpaired t-tests (E, F). ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, One-way ANOVA (D, I, J).

Fig. 3

SENP3 is highly expressed in macrophages of diabetic wounds. C57BL/6 wild-type mice or SD rats were intraperitoneally injected with streptozotocin (STZ) to induce diabetics (DM), or PBS as control (NDM). Full-thickness wounds were created on these NDM or DM mice or rats and collected at the indicated time. (A–F) wounds from NDM or DM mice were collected at 5 days and 7 days post-surgery. The skin wounds were subjected to immunohistochemistry of SENP3 (A); and statistics analysis of SENP3 positive area/wound area% (B). The level of SENP3 in mice skin wounds was evaluated by IB (C); and statistics analysis of SENP3/actin (D). Wounds collected at 5 days post-surgery were subjected to immunofluorescence staining of SENP3 or F4/80 (E) and statistics analysis of F4/80(+) SENP3(+)/F4/80(+) cells% (F). (G–J) wounds from NDM or DM rats were collected at 5 days and 7 days post-surgery. The skin wounds were subjected to immunohistochemistry of SENP3(G) and statistics analysis of SENP3 positive area/wound area% (H). Immunofluorescence staining of SENP3 or CD68 of wounds collected at 5 days post-surgery (I) and statistics analysis of CD68(+) SENP3(+)/CD68(+) cells% (J). Scale bars, as indicated in the figures. Data represents mean ± SD, n = 3 biologically independent samples. ****p < 0.0001, ***p < 0.0001, *p < 0.05, two-way ANOVA (B, D, H). **p < 0.01, ****p < 0.0001, unpaired t-test (F, J).

Fig. 4

SENP3^cko mice exhibited an increased proportion of M2 macrophages and promoted diabetic wound healing. (A) SENP3^fl/fl (n = 19) and SENP3^cko (n = 19) mice were injected with STZ. After 6–8 weeks, two full-thickness wounds were created on each mouse. (B–C) Photographs of full-thickness wounds of diabetic SENP3^fl/fl and SENP3^cko mice were taken at the indicated time and representative photographs were shown (A); Statistical results of relative wound area (fold of the wound on Day 0) (n = 5) (B). (D–E) Skin wounds were collected at 15 days post-surgery and subjected to hematoxylin and eosin (HE) staining and Masson staining (D); statistical results of wound diameter measured in pictures from HE staining (n = 5) (E). (F–I) Skin wounds were collected at 7 days post-surgery and subjected to HE staining (F) and statistical results of migrating tongue on day 7 (n = 5) (G). Immunohistochemistry of CD31 (H) and statistical results of CD31 positive area/total area% (n = 6) (I). (J–M) Skin wounds were collected on days 5, 7, and 9. The skin wounds slices were subjected to immunofluorescence staining of F4/80 and iNOS (J) and statistics analysis of F4/80(+) iNOS (+)/F4/80(+) cells% in day 5 diabetic wounds from SENP3^fl/fl and SENP3^cko mice (K). Immunofluorescence staining of F4/80 and ARG1 (L) and statistical results of F4/80 (+) ARG1 cells/F4/80 (+) cells% of day 7 and day 9 diabetic wounds (M). Scale bars, as indicated in the figures. Data represents mean ± SD. The above results were acquired from four independent experiments. ***p < 0.001, **p < 0.01, *p < 0.05, unpaired t-test (C, E, G, I, K). ****p < 0.0001, *p < 0.05, two-way ANOVA (M).

Fig. 5

SENP3-driven ferroptosis is dependent on FSP1. (A–B) BMDMs from SENP3^fl/fl and SENP3^cko mice were treated with RSL3 (1 μM) for 4 h. FerroOrang staining of cells was completed (A). The statistics of mean fluorescence intensity were shown (B). (C) SENP3^fl/fl and SENP3^cko BMDMs were incubated with RSL3 for 2 h. GPX activity was evaluated. (D–G) SENP3^fl/fl and SENP3^cko BMDMs were pretreated with or without viFSP1 (1 μM) for 24 h followed by RSL3 (1 μM) incubation for 5 h. Next, the viability (D) and LDH release (E) of these cells were tested. In addition, SYTOX Green staining of these cells and statistics analysis of dead/total cells% were shown (F–G). (H–L) HT1080 cells were stable over expressed SENP3 or empty vector as control, with 3 × flag-FSP1 or 3 × flag-vector. These cells were subjected to RSL3 (1 μM) treatment for 5 h. Then, the viability (H) and LDH release (I) of vector, SENP3, flag-FSP1 and flag-FSP1+ SENP3 transfected HT1080 cells were tested. In addition, SYTOX Green staining of these cells (J) and statistics analysis of dead/total cells% (K) were shown. The expression of SENP3 or FSP1 of the four group cells was validated by IB (L). Scale bars, 100 μm or as indicated in the figures. Data are shown as means ± sd. Data are representative of two (A, B) or three independent experiments (D–L). ****p < 0.0001, ***p < 0.001, ns p > 0.05, multiple unpaired t-tests (B–E). ns p > 0.05, unpaired t-test (G). ****p < 0.0001, ***p < 0.001, two-way ANOVA (H, I, K).

Fig. 6

SENP3 De-SUMOylates FSP1 at the K162 site. (A) prediction of SUMO modification site of Q9BRQ8.FSP1_HUMAN using GPS-SUMO, SUMOplot, and CLPM. (B–C) HEK293T cells were co-transfected with co-transfected with flag–SENP3 and HA-FSP1 plasmids or GFP–SENP3 and 3 × flag-FSP1 plasmids with vectors as control for 36 h. Co-IP was performed using FLAG-M2 beads for immunoprecipitation and using anti-HA (B) or anti-GFP antibodies (C) for IB. (D) HEK293T cells were transfected with 3 × flag-FSP1, RH–SUMO3, UBC9, HA-SENP3, or HA-SENP3 C532A mutant with indicated vectors for 48 h. RH-SUMO3 was pulled down using Ni-NTA beads and then analyzed by IB as indicated. Close brace indicated SUMO3-conjugated FSP1. (E) HEK293T cells were transfected with 3 × flag-FSP1, RH–SUMO3, and UBC9 with indicated vectors for 48 h followed by stimulation with RSL3 (10 μM) for the indicated time. FSP1 that was conjugated with RH–SUMO3 was detected by Ni-NTA pull-down assay. (F) HEK293T cells were transfected with RH–SUMO3, UBC9, and 3 × flag-FSP1 or 3 × flag-FSP1 K43R, K162R, K225R mutants with indicated vectors for 48 h. Cells were lysed and RH–SUMO3 was pulled down using Ni-NTA beads and then analyzed by IB as indicated. Close brace indicated SUMO3-conjugated FSP1. (G) HT1080 cells were stably overexpressing 3 × flag-FSP1 or 3 × flag-FSP1 K162R. Co-IP was performed using FLAG-M2 beads for immunoprecipitation and using anti-SENP3 antibody for IB. (H) Structure of human FSP1 protein from AlphaFold Protein Structure Database. The K162 site in FSP1 is conserved. The sequences around FSP1 K162 from different species were aligned. Conserved lysine residues corresponding to human FSP1 are marked in red. WB data represents at least three independent experiments.

Fig. 7

SENP3 de-SUMOylated FSP1 at the K162 Site and reversed its effect of inhibiting ferroptosis. (A–D) HT1080 cells were stably overexpressing 3 × flag-vector, 3 × flag-FSP1 or 3 × flag-FSP1 K162R, with SENP3 overexpression or empty vector, followed by RSL3 (1 μM, 5 h) incubation. Next, the viability (A) and LDH release (B) of cells were tested. SYTOX Green staining of cells (C) and statistics analysis of dead/total cells% were shown (D). (E–G) HT1080 cells were stably overexpressing 3 × flag-vector, 3 × flag-FSP1 or 3 × flag-FSP1 K162R, with SENP3 overexpression or empty vector, followed by RSL3 (1 μM, 1 h) incubation. Lipid peroxidation was evaluated by BODIPY 581/591 C11 staining(E). Statistics analysis of mean FITC intensity was shown (F). Western blots validated the expression of FSP1 and SENP3 expression in HT1080 cells which were transfected with 3 × flag-vector, 3 × flag-FSP1, or 3 × flag-FSP1 K162R, with SENP3 overexpression or empty vector (G). (H–K) RAW264.7 macrophages were stably overexpressing vector as control, 3 × flag-FSP1 or 3 × flag-FSP1 K162R followed by RSL3 (1 μM, 5 h) incubation. Next, the viability (H) and LDH release (I) of cells were tested. SYTOX Green staining of cells (J) and statistics analysis of dead/total cells% were shown (K). (L–N) RAW264.7 macrophages were stably overexpressing empty vector, 3 × flag-FSP1 or 3 × flag-FSP1 K162R followed by RSL3 (1 μM, 1 h) incubation. Lipid peroxidation was evaluated by BODIPY 581/591 C11 staining (L). Statistics analysis of mean FITC intensity was shown (M). Western blots validated the expression of FSP1 expression in RAW 264.7 cells which were transfected with vector, mouse flag-FSP1, or mouse flag-FSP1 K162R (N). Scale bars, 100 μm. Represents results of three independent experiments. ****p < 0.0001, ***p < 0.001, *p < 0.05, two-way ANOVA (A, B, F, H, I, M). ****p < 0.0001, ns p > 0.05, one-way ANOVA (D, K).