All animal procedures were conducted according to the Animal Care Committee of Chongqing Medical University. Male and female C57BL/6 mice weighing 18–22 g and 6–8 weeks old were kept in cages with a 12 h light/dark cycle. In a temperature-controlled room, the mice were provided with unlimited water and normal chow.

After 1 week of acclimation, male mice were randomly divided into the following groups: 1) Ctrl group (n = 25); 2) LPS group (intraperitoneally injected 10 mg/kg lipopolysaccharide, n = 25); 3) LPS + E₂ group (pretreated with 0.5 mg/kg/d 17β-Estradiol subcutaneously for 5 days + at the sixth day only intraperitoneally injected 10 mg/kg lipopolysaccharide, n = 25); 4) E₂ group (treated with 0.5 mg/kg/d 17β-Estradiol subcutaneously for 5 days + at the sixth day intraperitoneally injected the same volume saline instead of lipopolysaccharide, n=25). Female mice were randomly divided into the following groups: 1) Ctrl group (n = 25); 2) LPS group (intraperitoneally injected 10 mg/kg lipopolysaccharide, n = 25).

At the seventh day (at 24h after LPS intraperitoneally injection), lung tissues were collected for H&E staining, the mesenteric arteries were separated for vascular function, and the blood samples were collected from the orbital sinus. Serum was collected by centrifugation at 3,000 rpm for 10 min at ambient temperature after blood coagulation for 30 min. Lungs in the experimental groups were collected after 24 h after injection of LPS for further investigation.

Animal survival was determined over 120 hours. The observation was performed every 12 h on the first day and every 24 h on the following days. All remaining animals were humanely killed at the end of the experiment.

The fresh lung tissues were fixed in 4% paraformaldehyde (PFA). Then, the samples were gradually dehydrated and embedded in paraffin. After that, the samples were cut into 3 μm sections and fixed on a glass slide. The sections were stained with hematoxylin and eosin. Histological analyses were evaluated using a Leica DM 2700 P microscope (× 100; × 200; Leica Microsystems GmbH, Wetzlar, Germany).

The serum levels of aspartate transaminase (AST) and alanine transaminase (ALT) were determined by commercial kits to assess the acute hepatic injury. The serum level of Creatinine (Cr) and blood urea nitrogen (BUN) were determined by commercial kits to assess the acute kidney injury. Serum lactate (LD) and diamine oxidase (DAO) were measured by commercial kits to assess the acute intestines injury.

The mice were sacrificed by cervical dislocation. Then the mesenteric loop was dissected by a laparotomy and immediately placed in the iced Krebs–Henseleit solution. The second-order branch of mesenteric arterioles was separated under a surgical microscope. Krebs–Henseleit solution contained (mM): 118 NaCl, 11.1 D-glucose, 4.7 KCl, 1.6 CaCl₂, 1.2 KH₂PO₄, 1.18 MgSO₄, 25 NaHCO₃ and 11.1 D-glucose.

The relaxation function of the second-order branch of mesenteric arterioles was detected by Mulvany-style wire myograph (Model 520A, DMT, Aarhus, Denmark) and Powerlab analytical system (AD Instruments, Colorado Springs, CO, USA). In the chamber bath with 5 ml Krebs solution, the mesenteric arterioles were fixed to wire myograph by two tungsten wires. In the chamber bath, the Krebs solution was continuously oxygenated with a gas mixture of 5% CO₂ plus 95% O₂ at 37°C with a pH of ~7.4. The fixed arterioles were kept at the tension of zero for 20 min and then normalized.

Cumulative concentration-response curve (CRC) to PPT (10-70 nmol/l), DPN (10-40 nmol/l), G-1 (10 nmol/l -10 μmol/l) and acetylcholine (ACh, 0.01-1000 μM) were performed in NE (5 μM)- or KCl (80 mmol/l)- preconstricted arterioles. Further, to examine the relaxation mechanisms of estrogen receptors, the relaxation function of arterioles was determined by the isometric tension tests after the fixed arterioles were treated with different activators and inhibitors for 20-30 min. Through the isometric tension tests, the relaxation effect of arterioles was unaffected by the respective vehicles (< 0.4% DMSO or H₂O), and the contraction function of arterioles was unaffected by applied activators and inhibitors.

After normalization, the integrity of the arterial endothelium was verified by pre-constricting rings submaximally with 5 μM norepinephrine (NE) followed by relaxation with carbachol (CCh, 100 µM). We considered at least 90% relaxation response to CCh as an endothelium-intact vessel. Human hair was used to rub the intima to remove the endothelium for endothelium-denuded arterioles experiments. Then we verified the successful endothelial denudation by lack of relaxation response to 100 µM CCh (≤10%) and continued the next experiments. Only one experimental curve was proved for each preparation.

In physiological salt solution (PSS), a single HUVEC grown on coverslips was loaded with 5 μM Fluo-4,AM at 37 °C for 60 minutes and then washed for 20 minutes. Afterward, the coverslips were placed in a perfusion chamber on a Nikon microscope stage. Fluo-4 AM fluorescence ratio excited at 488 nm over time and detected by an intensified charge-coupled device camera (ICCD200) and a MetaFluor imaging system. The sampling interval of Fluo-4 AM fluorescence measurements was in the range of 3 s. The PSS included (mM): 140 Na⁺, 5 K⁺, 2 Ca²⁺, 147 Cl⁻, 10 Hepes, and 10 glucose. The osmolality for all solutions was ∼300 mosmol/kg of H₂O.

Human umbilical vein endothelial cells (HUVEC; USA) were incubated in endothelial cell medium (ECM, Sciencell, USA, Cat. #1001) with 10% fetal bovine serum (Gibco, USA) plus 1% penicillin-streptomycin (Beyotime Biotechnology, China). HUVEC were routinely cultured at 37 °C in a 5% CO₂ incubator. Cells passages 10-20 were used for experiments when cell density reached approximately 90%. In subsequent experiments, HUVEC were divided into three groups: (a) Control group: HUVEC were cultured in endothelial cell medium; (b) LPS group: HUVEC were treated with LPS (1 μg/ml); (c) LPS+E₂ group: HUVEC were treated with LPS (1 μg/ml) plus E₂ (10 nmol/l). To examine the mechanisms involved in the protective effects of E₂ in sepsis, HUVEC were incubated with the following compounds: 5 nmol/l PPT, 5 nmol/l DPN, 1 µmol/l G-1.

Serum levels of estrogen were measured by ELISA using mouse estrogen kits (EM1501, FineTest, Hubei, China). Serum levels of the cytokines, including tumor necrosis factor-alpha (TNF-α), interleukin (IL)-1β, and IL-6, were measured by ELISA using mouse-specific kits (R&D Systems, MTA00B, MLB00C, M6000B).

HUVEC were treated with different compounds at a density of 5 × 10⁵cells/well for 24h (seeded in 12-well plates). Then cell culture supernatants were collected and centrifuged at 1000 × g at 4°C for 20 min, and supernatants were then used to determine the levels of IL-6, IL-1β and TNF-a, in accordance with the manufacturer’s instructions (Elabscience, Wuhan, China, E-EL-H6156, E-EL-H0149c, E-EL-H5548c). All samples were assayed in triplicate.

HUVEC were treated with different compounds for 6h. Then total RNA was extracted from HUVEC using RNAeasy kit (beyotime, Shanghai, China) and cDNA was synthesized using the RT Master Mix for qPCR kit (MCE, New Jersey, USA). PCR amplification with a reaction volume of 10 μL, including 0.4 μL forward primer (10 μM), 0.4 μL reverse primer (10 μM), 5 μL SYBR, 3.2 μL RNase-free water and 1 μL cDNA was monitored by BIORAD CFX Connect Real-Time PCR Software, version 1.4.1 (Bio-Rad Laboratories, Hercules, CA, USA). The following primers were used:

After treatment for 24h, HUVEC were washed with ice-cold phosphate buffered saline (PBS). Briefly, total proteins of cultured cells were extracted with lysis buffer (RIPA, Millipore) with 1 mmol/L phenylmethylsulfonyl fluoride (PMSF), 1 mmol/L phosphatase inhibitor cocktail and 1 mmol/L protease inhibitor. The protein concentration was determined by a bicinchoninic acid (BCA) protein analysis kit (Solarbio, Beijing, China). The total lysate was separated using 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, and the separated proteins were transferred to polyvinylidene difluoride (PVDF) membranes. Membranes were blocked with QuickBlock™ Blocking Buffer for Western Blot (beyotime, Shanghai, China) for 20 min and then incubated with primary antibodies against the human phospholipase C Beta 1 Polyclonal antibody (PLC, 1:1000, Cat No. 26551-1-AP) and human ITPR1-specific Polyclonal antibody (IP₃R,1:1000, Cat No. 19962-1-AP) for 12h. GAPDH was used as an internal standard. Following incubation, PVDF membranes were washed with TBST three times for 10 min. Then PVDF membranes were incubated with peroxidase-conjugated specific secondary antibody for 1 h at room temperature. Bands were visualized by chemiluminescence, representative images were acquired, and Image J software (NIH Image, Bethesda, MD, USA) was used to quantify the density of each band.

All experiments and data collection were proved blindly. We applied technical repetition to strengthen the credibility of the data. All results are represented as means ± SEM, with n representing the number of animals in each group of experiments. In cell Ca²⁺ imaging experiments, n represents the total cells. For animal experiments, the n is at least n=5 for each group. We performed at least three independent repeated experiments for each treatment. The cumulative concentration-response curve (CRC), maximal relaxation (R_max), area under curve (AUC), and the concentration for 50% maximal effect (EC₅₀) were determined with GraphPad Software 7.0 (San Diego, CA).

The statistical significance of differences in the means of experimental groups was determined using Student’s unpaired, two-tailed t-test, 2-way ANOVA or one-way ANOVA followed by Dunnett’s post-test or Post hoc tests were run only if F achieved p<0.05 (GraphPad Prism 7.0, GraphPad Software, Inc.RRID : SCR_002798), and there was no significant variance inhomogeneity. A probability value of P<0.05 was considered statistically significant.

All salts were from Sangon Biotech in Shanghai. Acetylcholine, N^ω-nitro-L-arginine and indomethacin were offered by Sigma-Aldrich (St Louis, USA). NE was purchased from GRANDPHARMA (China) co. LTD. 17β-Estradiol, LPS, Propyl pyrazole triol (PPT), MPP hydrochloride (MPP), G-1, G-15, Diarylpropionitrile (DPN), PHTPP, ouabain, TRAM-34, apamin, U73122, U73343, 2-APB and SKA-31 were supplied by MedChemExpress (MCE; New Jersey, USA), SN-6 was provided by Tocris (Bristol, UK). LiCl was supplied by Macklin (China). The human phospholipase C Beta 1 Polyclonal antibody (Cat No. 26551-1-AP) and human ITPR1-specific Polyclonal antibody (Cat No. 19962-1-AP) were from proteintech. The commercial kits of serum ALT, AST, LD and BUN were from Neobioscience (Jiancheng Bioengineering Institute, Nanjing, China). The commercial kit of serum DAO was from Solarbio. The commercial kit of serum Cr was from Beijing Leagene Biotechnology co. LTD.

First, we applied LPS-induced septic model of mice to study the action of E₂ on sepsis, and our experimental protocol was shown in Figure 1A . The survival rate of LPS-induced sepsis was greater in female mice than in male mice, and E₂ pretreatment (0.5 mg/kg/d subcutaneously for 5 days) significantly improved the survival rate of male septic mice ( Figure 1B ), indicating the anti-septic action of E₂, which accords with an earlier observation. (24) Accordingly, E₂ concentrations in female group were obviously higher than male group ( Figure 1C ). Second, we examined if the development of LPS-induced tissue injury was closely related to the enhanced inflammation. As shown in Figures 1D–F (left panels), serum levels of inflammatory cytokines, including TNF-α, IL-1β and IL-6, were significantly increased in the LPS group, while pretreatment with E₂ effectively reduced the release of inflammatory cytokines. Moreover, LPS-induced inflammatory cytokines were significantly lower in female septic group compared to male septic group (right panels in Figures 1D–F ). Therefore, E₂ can mitigate inflammatory cytokines release to ameliorate septic survival rate.

Since sepsis is a life-threatening disease with multiple organ failures including the lung, intestine, liver and kidney, we tested if E₂ has a beneficial role in sepsis-induced tissue damage. As shown in H&E staining, compared to the normal lung ( Figure 2A ), obvious pathological changes in the lung were observed in the sepsis group, including abundant inflammatory cell infiltration and interstitial edema ( Figure 2B ). However, E₂ pretreatment (0.5 mg/kg/d subcutaneously for 5 days) could markedly improve these pathological changes ( Figure 2C ).

It is acknowledged that serum lactate (LD), a product of intestinal bacteria glycolysis, and diamine oxidase (DAO), a high-activity intracellular enzyme in the intestinal villi that can be released into the blood from the damaged intestinal mucosa, are considered sensitive indicators for the intestinal mucosal barrier. Indeed, serum levels of both LD and DAO were increased in male septic mice compared to control mice, but E₂ pretreatment significantly decreased their serum levels (left panels in Figures 2D, E ). In addition, in the LPS-induced septic group, the serum level of DAO in females was lower than in males (right panels in Figure 2D ).

Moreover, the serum levels of aspartate transaminase (AST) and alanine transaminase (ALT), important clinical parameters of hepatic function, were significantly elevated in male septic mice compared to control mice. At the same time, E₂ pretreatment significantly decreased the LPS-induced elevation of these hepatic enzymes (left panels in Figures 2F, G ). In septic mice, the serum levels of ALT and AST were lower in females than in males (right panels in Figures 2F, G ). Finally, the serum levels of creatinine (Cr) and blood urea nitrogen (BUN), important clinical parameters of renal function, were significantly elevated in male septic mice (left panels in Figures 2H, I ), and E₂ pretreatment significantly decreased the LPS-induced elevation of Cr (left panels in Figure 2H ). In septic mice, the serum level of Cr was lower in females than in males (right panels in Figure 2H ). Therefore, E₂ can protect against multiple organ dysfunction in sepsis.

Sepsis is usually accompanied by microvascular dysfunction, finally resulting in hypoperfusion and multiple organ failure. Although E₂ can exert beneficial effects on sepsis-induced multiple organ dysfunction, it is unknown if vascular ER subtypes are involved in these benefits in sepsis. Therefore, we applied PPT, a selective ERα activator, to examine vascular action of ERα and the underlying mechanisms. As shown in Figure 3A , PPT dose-dependently induced vasorelaxation of mouse mesenteric arterioles, which was mostly endothelium-dependent. Moreover, PPT hardly induced vasorelaxation of the arterioles pre-constricted by high K⁺ (80 mM), and there were significant differences in PPT-induced CRC, R_max, AUC, and EC₅₀ between the arterioles pre-constricted by NE and high K⁺, suggesting a possible involvement of cellular K⁺ gradients.

Since VEC mediates vasorelaxation via NO, PGI₂, and EDH, we examined their contributions to PPT-induced vasorelaxation. As shown in Figure 3B , either L-NNA (100 μM), a selective inhibitor of NO, or INDO (10 μM), a selective inhibitor of PGI₂, did not affect PPT-induced CRC, R_max, AUC, and EC₅₀ in comparison to the control. Furthermore, PPT-induced vasorelaxation was barely affected by L-NNA plus INDO ( Figure 3B ), indicating minor roles of NO and PGI₂ in PPT-induced vasorelaxation. However, when L-NNA and INDO eliminated NO and PGI₂, PPT-induced vasorelaxation was significantly inhibited by apamin (3 μM) plus TRAM-34 (30 μM), selective blockers of SK_Ca and IK_Ca; but potentiated by SKA-31 (0.3 μM), a selective activator of SK_Ca and IK_Ca ( Figure 3C ), indicating that EDH plays a predominant role in PPT-induced vasorelaxation.

Since EDH is regulated not only by endothelial IK_Ca and SK_Ca, but also by Na⁺-K⁺ ATPase (NKA) and Na⁺/Ca²⁺ exchanger (NCX) on VSMCs, we further examined if NKA and NCX are involved in PPT-induced vasorelaxation. Indeed, in the presence of L-NNA and INDO, PPT-induced vasorelaxation was significantly attenuated by ouabain (100 μM), a selective inhibitor of NKA, and SN-6 (10 μM), a selective inhibitor of NCX ( Supplementary Figure 1A ). Finally, to confirm the role of ERα in PPT-induced vasorelaxation, we applied MPP (1 μM), a selective inhibitor of ERα. As shown in Supplementary Figure 1A , PPT-induced vasorelaxation was significantly inhibited by MPP. Therefore, ERα subtype mediates vasorelaxation of mesenteric arterioles via EDH mechanism.

Due to the critical role of PLC/IP₃ pathway in the process of membrane receptor activation, we further examined its possible involvement in PPT-induced vasorelaxation. First, in the presence of L-NNA and INDO, PPT-induced vasorelaxation was significantly attenuated by U73122 (10 μM), a selective PLC inhibitor but was unaffected by U73343 (10 μM), an inactive analog of U-73122 ( Figure 3D ). Second, PPT-induced vasorelaxation was inhibited by LiCl (30 mM), an inhibitor of IP₃ production, and 2-APB (40 μM), an inhibitor of IP₃ receptor (IP₃R) ( Figure 3D ). 2-APB can not only inhibit IP₃R in vascular endothelial cells, leading to a decrease in Ca²⁺ signaling to attenuate endothelium-dependent vasorelaxation, but it can also activate some TRP channels to cause an increase in Ca²⁺ signaling and to promote vasorelaxation. (25–27) Therefore, in the present study, PPT-induced vasorelaxation was significantly inhibited by 2-APB, indicating it inhibits IP₃R rather than activates TRP channels. Taken together, ERα subtype regulates EDH-mediated vasorelaxation via PLC/IP₃ pathway.

We applied either DPN, a selective ERβ activator, or G-1, a selective GPER activator, to examine vascular action of ERβ and EPER and the underlying mechanisms. As shown in Figures 4A and 5A , either DPN or G-1 induced vasorelaxation in dose-dependent and endothelium-dependent manners. There were significant differences in DPN- and GPER-induced CRC, R_max, AUC, and EC₅₀ between the arterioles pre-constricted by NE and high K⁺ ( Figures 4A , 5A ).

As shown in Figures 4B and 5B , L-NNA and INDO have no effects on DPN- and G-1-induced CRC, R_max, AUC, and EC₅₀ compared to their controls. However, in the presence of L-NNA plus INDO, DPN- and G-1-induced vasorelaxation was inhibited by apamin plus TRAM-34; but potentiated by SKA-31 ( Figures 4C , 5C ), further suggesting that DPN- and GPER-mediated vasorelaxation via EDH predominantly. As shown in Supplementary Figures 1B, C , in the presence of L-NNA and INDO, DPN- and G-1-induced vasorelaxation was significantly attenuated by ouabain and SN-6 respectively, indicating the involvements of NKA and NCX. Finally, we applied either PHTPP, a selective inhibitor of ERβ, or G-15, a selective inhibitor of GPER, to examine the involvements of ERβ and GPER. As shown in Supplementary Figures 1B, C , in the presence of PHTPP (3 μM) and G-15 (1 μM), DPN- and G-1-induced vasorelaxation was significantly attenuated, indicating ERβ and GPER subtypes mediate vasorelaxation of mesenteric arterioles via EDH mechanism.

We also examined the possible involvements of PLC/IP₃ pathway in ERβ- and GPER-induced vasorelaxation. As shown in Figures 4D , 5D , in the presence of L-NNA and INDO, DPN- and G-1-induced vasorelaxation was significantly attenuated by U73122, LiCl and 2-APB, but was unaffected by U73343. Taken together, ERβ- and GPER-mediated EDH-induced vasorelaxation via PLC/IP₃ pathway.

We further verified if ER subtypes induce Ca²⁺ signaling in HUVEC via PLC/IP₃/IP₃R pathway. As shown in Figures 6A, B , PPT-induced obvious [Ca²⁺]_cyt signaling, which was attenuated by MPP, U73122, LiCl and 2-APB, indicating that PPT-induced endothelial [Ca²⁺]_cyt signaling via ERα/PLC/IP₃/IP₃R pathway. Similarly, DPN-induced [Ca²⁺]_cyt signaling was attenuated by PHTPP, U73122, LiCl and 2-APB, indicating that DPN-induced endothelial [Ca²⁺]_cyt signaling via ERβ/PLC/IP₃/IP₃R pathway ( Figures 6C, D ). Finally, G-1-induced [Ca²⁺]_cyt signaling was attenuated by G-15, U73122, LiCl and 2-APB, indicating that G-1 induced endothelial [Ca²⁺]_cyt signaling via GPER/PLC/IP₃/IP₃R pathway ( Figures 6E, F ). Therefore, these data further support that endothelial PLC/IP₃/IP₃R pathway plays an important role in ER subtype-induced vasorelaxation.

As described earlier, we found the survival rate of sepsis was greater in female mice than in male mice due to the protective effect of E₂ via ER activation; however, nothing is known if ER subtype-mediated vasorelaxation contributes to the beneficial effect of E₂. To this end, first, we examined the relaxation response to vagus neurotransmitter acetylcholine (ACh), an acknowledged vasodilator, in LPS-induced septic mice. As shown in Figure 7A , there were no differences in ACh-induced vasorelaxation between healthy male and female mice. However, in LPS-induced septic mice, ACh-induced vasorelaxation was more impaired in males than in females. Furthermore, in the presence of L-NNA and INDO, there were no differences in ACh-induced vasorelaxation between male and female mice. However, in LPS-induced septic mice, ACh-induced vasorelaxation was more impaired in males than in females ( Figure 7B ). Finally, in LPS-induced male septic mice, ACh-induced vasorelaxation was significantly impaired, but E₂ (0.5 mg/kg/d subcutaneously for 5 days) pretreatment rescued the impaired ACh-induced EDH-mediated vasorelaxation ( Figures 7C, D ).

Second, we examined if ER subtype-mediated vasorelaxation plays a protective role against sepsis. As shown in Figures 8A, C, E , in the absence or the presence of L-NNA and INDO, PPT-, DPN- and G-1-induced vasorelaxation was more impaired in septic male mice than female mice. Taken together, septic female mice are more resistant to the dysfunction in EDH-mediated vasorelaxation because ER subtypes play protective roles against sepsis.

Although we found that PLC/IP₃ pathway is involved in ER-induced vasorelaxation under physiological situations, nothing is known about its contribution to the gender differences in EDH-mediated vasorelaxation in sepsis. As shown in Figures 8B, D, F , in the presence of L-NNA and INDO, the EDH-mediated vasorelaxation induced by PPT, DPN, and G-1 was inhibited by U73122 or LiCl in septic female mice, leading to the abolishment of the gender differences in PPT-, DPN- and G-1-induced EDH-mediated vasorelaxation between septic female and male mice. Taken together, PLC/IP₃ pathway contributes to the gender differences in EDH-mediated vasorelaxation in sepsis.

Since sepsis is usually accompanied by VEC dysfunction, and proinflammatory cytokines IL-1β, IL-6 and TNF-α are important in the occurrence and development of sepsis, we first examined if E₂ can affect release of inflammatory cytokines from HUVEC. As shown in Supplementary Figures 2A, B , compared to ctrl group, both mRNA expression and release of IL-1β, IL-6 and TNF-α were significantly enhanced in LPS-pretreated group, which were reduced by pretreatment with E₂ (10 nM).

We further examined the contributions of ER subtypes to E₂-mediated inflammatory responses, As shown in Supplementary Figures 3A-C , the release of proinflammatory cytokines IL-1β, IL-6 and TNF-α from HUVEC were increased by LPS, which were attenuated by PPT, DPN, and G-1, respectively. Taken together, E₂ can activate ER subtypes to exert protective effects by inhibiting release of inflammatory cytokines from VEC in sepsis. Finally, we examined if endothelial PLC/IP₃R pathway is involved in the protective effect of E₂ against sepsis. As shown in Supplementary Figures 4A–C , compared to ctrl group, both mRNA and protein expressions of PLC and IP₃R in HUVEC were significantly increased in LPS-pretreated group, which could be rescued by E₂. These data further support that E₂ exerts protective effect against sepsis via endothelial PLC/IP₃R pathway.