GZMA was decreased in patients with IBD and DSS-induced colitis

To explore the change of GZMA expression in IBD, We firstly utilized ELISA to assess the concentration of GZMA in the serum and colonic tissues of patients with IBD. The results showed that a significant downregulation of GZMA levels was observed in compared to healthy controls in both the serum and colonic tissues of IBD patients. (Fig. 1A-B). In line with this, the GZMA mRNA level in the colon of DSS-treated mice was largely decreased compared to that in normal mice, indicative of a significant increase in IL-6 expression, a marker of intestinal inflammation(Fig. 1C). In addition, immunofluorescence analysis further confirmed GZMA expression in CD8+CD39+ T cells was reduced in the intestinal mucosa of IBD patients (Fig. 1D).

Fig. 1

Decreased GZMA in patients with IBD and DSS-induced colitis. (A-B) The level of GZMA in serum and colonic mucosa were measured by ELISA according to the instruction. Data was displayed as the means ± s.d. of three independent experiments and analyzed by two-sample t test for significance, ***p < 0.001, **p < 0.01. (C) Real-time PCR was employed to assess colonic GZMA and IL-6 mRNA level in indicated group. Data was exhibited as the means ± s.d. of three independent experiments and analyzed by one-sample t-test for significance, ****p < 0.0001. (D) Immunofluorescence assay was performed to detect CD8, CD39, and GZMA expression in indicated group (Scale bar: 100 μm)

GZMA induced intestinal epithelial cell differentiation

Previous studies have implied GZMA might have a role in cell differentiation due to the work showed that GZMA depletion led to a significant osteoclast differentiation [21, 22], which focused us to explore the possible effect of GZMA on intestinal epithelial cell differentiation. As shown in Fig. 2A, CDX2, a mastor of intestinal epithelial cell differentiation, was found to be significantly reduced in the intestinal mucosa, which focused us to explore the possible role of CD8+CD39+ T cells-derived GZMA on CDX2 expression. Next, the 21-day Caco-2 monolayer transwell system [19, 23,24,25] was employed to assess the effect of GZMA on intestinal epithelial cell integrity (Fig. 2B). As expected, a significant downregulation of trans-epithelial flux of FITC-dextran in polarized Caco-2 cells was observed after GZMA treatment (Fig. 2C), which was attributed to enanced intestinal epithelial cell differentiation as evidenced by increased CDX2, ZO-1 and OCLN mRNA and protein level(Fig. 2D-F). In line with this, immunofluorescence analysis also confirmed that GZMA upregulated ZO-1 and OCLN expression in Caco-2 (Fig. 2G).

Fig. 2

GZMA promoted intestinal epithelial cell differentiation. (A) Immunofluorescence assay was performed to detect CDX2 expression of colonic mucosa from clinical sample (Scale bar: 100 μm), quantitation was performed by Image J, and analyzed by two-sample t test, ∗∗∗p < 0:001. (B-C) Caco-2 cells were seeded onto transwell polycarbonate membranes (0.4 μm pores). Upon confluence (21 days after seeding), the cells were treated with GZMA (500 nM) for up to 48 h. The permeability of the monolayer to FITC-dextran (4 kDa) was assessed by measuring the fluorescence intensity in the bottom chamber at Ex/Em = 485/535 nm. Data was displayed as means ± s.d. of three independent experiments and analyzed by two-sample t test for significance, ***p < 0.001.(D) Real-time PCR and (E-F) western blotting as well as (G) immunofluorescence staining were conducted to analyze the indicated gene expression at mRNA and protein level in HT-29 and Caco-2 cells treated with or without GZMA (500 nM) for 48 h (Scale bar: 25 μm), Data was showed as the means ± s.d. of three independent experiments and quantified by one-sample t test for significance, ***p < 0.001, **p < 0.01, *p < 0.05. (H) intestinal crypt isolated from mice was used to explore the effect of GZMA (500 nM) on intestinal organoid generation, microscopic examination of organoids was employed to calculate the proportion of budding organoids among every average 100 organoids. Data was exhibited as means ± s.d. of three independent experiments and analyzed by two sample t test, **p < 0.01 (Scale bar: 50 μm). (I) western blotting was conducted to analyze the indicated proteins in Caco-2 cells after transferred with sh-CDX2 plasmid, followed by stimulation with GZMA (500 nM) for 48 h, with β-actin serving as the internal control. (J-K) CD8+CD39+ T cell subsets isolated from peripheral blood of healthy donors were cultured in medium for 24 h was collected to co-culture with Caco-2 cells for 48 h combined with or without GZMA antibody supplementation. The total lysate was harvested to detect indicated proteins, Data was displayed as mean ± s.d. of three independent experiments and analyzed by one-sample t test for significance, ***p < 0.001, **p < 0.01

Interestingly, GZMA treatment could increase the budding of intestinal organoids (Fig. 2H), and depletion of CDX2 expression in Caco-2 could reverse the effect of GZMA on ZO-1 and OCLN expression, indicating that GZMA modulated intestinal epithelial integrity through enhancing CDX2 expression (Fig. 2I). most importantly, addition of anti-GZMA in co-cultured system could rescue the influence of CD8+CD39+ T cells isolated from PBMCs from healthy donors’ peripheral blood on CDX2, ZO-1 and OCLN expression (Fig.S1F, Fig. 2J-K). Taken together, these data suggested that GZMA is critical mediator in CD8+CD39+ T cells-maintained intestinal epithelial barrier function through CDX2.

GZMA promoted CDX2 expression through ferroptosis inhibition

Next, we sought to ask how GZMA regulated CDX2 expression. Previous studies have found that ferroptosis is prominently triggered in IECs of both UC patients and DSS-induced colitis [10], which attracted us to study the relationship between CDX2 and ferroptosis. A time course of Caco-2 cells differentiation showed that GPX4 expression was gradually enhanced during intestinal epithelial cells differentiation characterized by increased CDX2 expression, despite no significant difference in SLC7A11 expression was observed (Fig. 3A). Further results revealed that CD8+CD39+ T cells-derived GZMA or GZMA alone could inhibit ferroptosis characterized by enhanced GPX4 and SLC7A11 mRNA and protein levels as well as reduced ROS level (Fig. 3B-E). Moreover, immunoblotting, intestinal organiods, the trans-epithelial assessment and luciferase assay showed that activation of ferroptosis by RSL3 could reverse the promotion of GZMA on CDX2-dependent ZO-1 and OCLN expression (Fig. 3F-H, Fig.S2A-2C). what’s more, the similar phenomenon was observed in IECs treated with GZMA after specific GPX4 slienced by shRNA plasmids (Fig.S2D).

Fig. 3

GZMA promoted intestinal epithelial cell differentiation through inhibition of ferroptosis. (A) Lysate was extracted from Caco-2 at day 1/3/5/7 post-confluence, and the expression of target proteins were analyzed by western blotting. Data was represented the mean ± s.d. of three independent experiments and determined by one-way ANOVA for significance, ****p < 0.0001, ***p < 0.001. HT-29 and Caco-2 cells were treated with GZMA (500 nM) for 48 h, WB (B) and qPCR (C) were employed to determine indicated gene at mRNA and protein level. The control was normalized as 1. The band intensity and relative mRNA expression was represented as mean ± s.d. of three independent experiments and analyzed by one-sample t test for significance, ***p < 0.001, **p < 0.01, *p < 0.05. (D) After serum starvation for 24 h, Caco-2 cells were treated with GZMA (500 nM) for 48 h and the ROS level was detected. (E) immunoblotiting was performed to detect indicated protein in coculture model established by CD39+CD8+ T cell and Caco-2 cells combined with or without anti-GZMA addition, Data was displayed as the mean ± s.d. of three independent experiments and analyzed using one sample t test for significance, ****p < 0.0001, ***p < 0.001. The control was normalized to 1. (F-G) HT-29 and Caco-2 cells were treated with GZMA (500 nM) combined with or without RSL3 (5 µM) for 48 h, WB was conducted to analyze the indicated protein. Data was presented as the means ± s.d. of three independent experiments and analyzed by one-way ANOVA and Dunnett’s multiple comparison test, ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05. (H) After co-transfection with CDX2-luc plasmid and a control renilla luciferase vector for 12 h, 293T cells were treated with GZMA (500 nM) or GZMA (500 nM) combined with RSL3 (5 µM) for 48 h, the relative luciferase unit (RLU) was measured. Data was presented as the means ± s.d. of three independent experiments and analyzed by one-way ANOVA, *p < 0.05,**p < 0.01. (I) Image of colon length and (J) body weight change in indicated group were exhibited. the percentage of initial body weight at the start of the experiments as 100%. Statistical difference was determined by one way ANOVA analysis, ****p < 0.0001, **p < 0.01. (K) HE staining of representative colon mucosa and scored to determine difference using one way ANOVA in indicated groups, *p < 0.05,**p < 0.01,***p < 0.001. (L) Immunofluorescence was performed to detect CDX2, ZO-1, and OCLN expression in indicated group

To provide insights into the effect of GZMA and RSL3 in vivo, DSS-induced colitis model was employed to confirm the in vitro results combined with intraperitoneal injection of 20 ug GZMA or RSL3 at a dose of 10 mg/kg on days 0, 2, 4, and 6. As shown in Fig. 2I-L, GZMA-treated mice were shown to be protected from experimental colitis, as evidenced by enhanced mean bodyweight and colon length as well as alleviation of intestinal inflammation, while RSL3 administration displayed an impaired effect against GZMA, including CDX2, ZO-1 and OCLN expression detected by IF analysis. In summary, these work suggested GZMA induced CDX2-mediated IECs differentiation to improve epithelial integrity through inhibition of GPX4-mediated ferroptosis.

GZMA-mediated ferroptosis required CREB

It has been demonstrated that CREB plays a role in the transactivation of GPX4 [26]. Therefore, we intended to investigate whether CREB is required in the regulation of GZMA-mediated GPX4 expression in IECs. Subcellular fraction isolation and immunofluorescence were performed to detect the nuclear localization of CREB in response to GZMA stimulation. As shown in Fig. 4A-B, nuclear CREB level and nuclear translocation was drastically increased in HT-29 and Caco-2 cells after GZMA treatment. What’s more, inhibition of CREB largely reversed the effect of GZMA on GPX4, CDX2, and downstream proteins (Fig. 4C, Fig.S3A). Furthermore, we found that the GPX4 transactivation induced by GZMA was reversed after CREB depleted in HEK293 cells (Fig. 4D). Altogether, these results suggested GZMA promoted GPX4 expression through CREB.

Fig. 4

GZMA modulated CREB nuclear translocation to promote intestinal epithelial integirty (A) HT-29 and Caco-2 cells were serum-starved for 24 h, followed by stimulation with GZMA (500 nM) for indicated time. Cell cytoplasmic and nuclear proteins were extracted to detect CREB. Data represent the mean ± s.d. of three independent experiments and were analyzed by one-way ANOVA with multiple comparisons, followed by Dunnett post hoc test for significance versus Control, ****p < 0.0001. the control was normalized as 1. (B) Immunofluorescence of CREB localization in HT-29 and Caco-2 cells treated with or without GZMA for 1 h after serum starved for 24 h. (C) western blotting was conducted to analyze the expression of the specific proteins in Caco-2 cells after transferd with sh-CREB plasmid, followed by stimulation with GZMA (500 nM) for 48 h. Data represent the mean ± s.d. of three independent experiments and were analyzed by one-way ANOVA with multiple comparisons, followed by Dunnett post hoc test for significance versus Control, **p < 0.01, ***p < 0.001, ****p < 0.0001. The control was normalized to 1. (D) After co-transfected with indicated plasmids combined with GPX4-Luc plasmid, sh-CREB plasmids and a control Renilla luciferase expression vector for 48 h, 293T cells were treated with or without GZMA(500 nM) for 48 h, the relative luciferase unit (RLU) was presented as the fold activation relative to Renilla luciferase activity. Data represent the mean ± s.d. of three independent experiments and were analyzed by two-way ANOVA, followed by Dunnett post hoc test for significance versus Control. *p < 0.05, ***p < 0.01

GZMA triggered PDE4/PKA/CREB cascade signaling

The above results indicated that CREB is crucial for GZMA-mediated ferroptosis in IECs, highlighting its significance. PKA phosphorylated CREB to enhance CREB nuclear translocation to initiate target genes expression [27]. Therefore, we tried to investigate the impact of the cAMP/PKA/CREB cascade signaling in response to GZMA due to the previous work from our lab demonstrated that this pathway was critical for GPX4 expression [12]. As shown in Fig. 5A, cAMP level was increased in colon tissue from DSS group after GZMA administration in vivo, and in vitro also confirmed GZMA could induce cAMP gerentiation. What’s more, addition of Rp-cAMPS, a PKA inhibitor, largely blocked the effect of GZMA on ferroptosis and CDX2 as well as downstream targets expression in Caco-2 cells(Fig. 5B). Of note, further work showed that GZMA could suppress PDE4 phosphorylation, leading to increase cAMP generation in vivo and in vitro, which further triggered PKA/CREB activation, while no significant changes of AC6 was observed (Data not shown) (Fig. 5C). Most importantly, GZMA-induced endogenous PKA/CREB complex co-localization was enhanced after PDE4 inhibitor dipyridamole (DIP) (Fig. 5D-E). These results suggested that GZMA inhibited ferroptosis through activation of PDE4/PKA/CREB signaling pathway.

Fig. 5

GZMA modulated PDE4/PKA/CREB cascade signaling. (A) The level of cAMP was measured by Elisa according to the instruction. Data presented as the means ± s.d. of three independent experiments and were analyzed by one-way ANOVA (left panel) and two sample t test (right panel), *p < 0.05. (B) HT-29 and Caco-2 cells were serum-starved for 24 h, followed by stimulation with GZMA (500 nM) for 1 h followed by addition of with Rp-cAMPS (10 µM) for 48 h, WB was conducted to analyze the indicated protein. The band was quantified and analyzed by one-sample t test for significance, the control was normalized as 1, data represent the mean ± s.d. ***p < 0.001, **p < 0.01, ****p < 0.0001. (C) after starvation overnight, HT-29 cells were treated with GZMA for 1 h, and the total protein was collected to detect indicated protein, Data was presented as the mean ± s.d. of three independent experiments and were analyzed by one-way ANOVA (left panel) and two sample t test (right panel), **p < 0.01, ***p < 0.001. (D) Caco-2 cells were serum starved for 24 h after 80% confluence, then stimulated as indicated for 1 h. Immunoprecipitated (IP) was employed to analyze the interaction between PKA and CREB. (E) Immunofluorescence of co-localization between PKA and CREB in Caco-2 cells treated with or without GZMA for 1 h after serum starved for 24 h. scale bar = 50 μm

Ferroptosis inhibition caused by GZMA administration ameliorated DSS-induced colitis

In vitro work has demonstrated GZMA alleviated PDE4-mediated ferroptosis to promote CDX2 expression providing insights into the effect of GZMA on ferroptosis in vivo, DSS-induced colitis model combined with intraperitoneal injection of 20 μg GZMA on days 0, 2, 4, and 6 was employed to confirm the above results. As shown in Fig. 6A-B, GZMA-treated mice was exhibited to be protected from experimental colitis as evidenced by mean bodyweight and colon length. Further work showed that decreased ferroptosis, characterized by enhanced GPX4 and xCT in IECs labeled with EpcAM was observed after GZMA administration in DSS group, which was attributed to PDE4 phosphorylation inhibition confirmed by immunofluorence (Fig. 6C). Taken together, these results suggested that GZMA is critical for improvement of intestinal barrier function through modulating PDE4-mediated ferroptosis.

Fig. 6

GZMA improved DSS-induced colitis in vivo (A) Representative colon length and (B) body weight changes in indicated group were measured and analyzed by one way ANOVA, The body weight changes were expressed as the percentage of initial body weight at the start of the experiments as 100%, *** p < 0.001, ** p < 0.01. (C) Immunofluorescence was performed to detect GPX4, xCT and phosphorylation of PDE4 expression in indicated group. (D) Schematic model of GZMA regulating IEC differentiation through ferroptosis