Artificial cells delivering itaconic acid induce anti-inflammatory memory-like macrophages to reverse acute liver failure and prevent reinjury

Summary

Hepatic macrophages represent a key cellular component of the liver and are essential for the progression of acute liver failure (ALF). We construct artificial apoptotic cells loaded with itaconic acid (AI-Cells), wherein the compositions of the synthetic plasma membrane and surface topology are rationally engineered. AI-Cells are predominantly localized to the liver and further transport to hepatic macrophages. Intravenous administration of AI-Cells modulates macrophage inflammation to protect the liver from acetaminophen-induced ALF. Mechanistically, AI-Cells act on caspase-1 to suppress NLRP3 inflammasome-mediated cleavage of pro-IL-1β into its active form in macrophages. Notably, AI-Cells specifically induce anti-inflammatory memory-like hepatic macrophages in ALF mice, which prevent constitutive overproduction of IL-1β when liver reinjury occurs. In light of AI-Cells’ precise delivery and training of memory-like hepatic macrophages, they offer promising therapeutic potential in reversing ALF by finely controlling inflammatory responses and orchestrating liver homeostasis, which potentially affect the treatment of various types of liver failure.

Graphical abstract

Highlights

• •AI-Cells deliver itaconic acid to target and train macrophages for liver repair
• •Itaconic acid modulates the crosstalk between macrophages and damaged hepatocytes
• •IL-1β regulated by itaconic acid plays a vital role in reversing acute liver failure
• •AI-Cells prevent liver reinjury by inducing memory-like hepatic macrophages

Macrophages are key regulators of tissue homeostasis. Yin et al. construct artificial apoptotic cells loaded with itaconic acid (AI-Cells) to accurately capture and train hepatic macrophages for liver repair. AI-Cells offer promising therapeutic potential in the rescue of liver failure through finely modulating hepatic macrophages to orchestrate liver homeostasis.

Introduction

Acute liver failure (ALF) is a lethal condition characterized by loss of hepatocyte function and subsequent multi-organ failure.^1,2 The most common reasons for ALF are hepatitis virus infection and drug-induced injury.³ The recent increase in unexplained acute pediatric hepatitis has appeared globally, which leads to ALF.^4,5 Triggering liver inflammation and an overreaction of the immune system are thought to cause unexplained acute hepatitis.^6,7 Moreover, the occurrence of primary liver failure increases susceptibility toward liver reinjury.^8,9 Therapeutic management of ALF is constrained mainly to N-acetyl cysteine (NAC) therapy. However, once NAC cannot reverse liver failure, the only treatment is orthotopic liver transplantation.^10,11 Donor organ shortages, recipients with more advanced disease at transplantation, and adverse effects associated with long-term immunosuppression greatly limit the applicability of liver transplantation.¹² Therefore, alternative potent therapeutic approaches to reduce or reverse ALF are urgently needed.

Regardless of etiology, the release of damage-associated molecular patterns (DAMPs) from dying hepatocytes exists in response to toxic injury, which activates the innate immune system and triggers inflammation.¹³ A modest immune response facilitates the clearance of necrotic cell debris and tissue-restoring responses, but an overreaction of immune system promotes immune-mediated cytokine storms and drives fulminant liver failure.¹⁴ Intriguingly, it has been shown that trained immunity nonspecifically protects from liver reinjury regardless of virus or ischemia/reperfusion injury, indicating innate immune cells play a non-negligible role in tissue damage.^15,16 Trained immunity as a de facto immune memory of the innate immune system can protect the host against a secondary challenge by robustly reactivating innate immune cells after a first challenge and returning to a non-activated stage.¹⁷ Therefore, rapid restoration of immune homeostasis and effective induction of immune memory hold the key to the broad-spectrum therapy of various types of liver failure.

Macrophages are one of the major innate immune cells that are most often implicated in the pathogenesis of cytokine storms.^18,19 Hepatic macrophages comprise two developmentally distinct populations, liver resident Kupffer cells and monocyte-derived macrophages, which can promptly infiltrate the liver following the initial sensing of liver damage by Kupffer cells.¹⁸ Infiltrating monocytes can have both anti-inflammatory and pro-inflammatory functions.¹⁸ Once infiltrating monocytes become overactivated, they will secret excessive amounts of cytokines to trigger cytokine storms, leading to severe tissue damage. Interestingly, macrophages can be trained to develop adaptive immune features that mount a protective response during future encounters.^20,21 Thereby, targeting macrophages to calm cytokine storms and inducing immune memory of hepatic macrophages might reduce cell necrosis after a toxic injury.

A growing body of evidence indicates that itaconic acid (ITA) plays a crucial role in tricarboxylic acid (TCA) cycle remodeling and cytokine regulation in activated macrophages. Itaconate metabolism is the regulatory node between trained immunity and tolerance induction,²² implying the importance of itaconate for inflammation modulation and trained immunity. ITA can be produced through the decarboxylation of cis-aconitate, a TCA cycle intermediate, in response to immunoresponsive gene 1 (IRG-1).^23,24 ITA, also known as 2-methylenesuccinic acid, has high solubility in water and suffer from poor membrane permeability. Esterified derivatives of ITA, such as dimethyl itaconate (DI), 4-octyl itaconate (4-OI), and 4-monoethyl ITA (4-EI), have been widely used as substitutes for ITA for macrophage-based immune studies. 4-OI has been proposed to act through nuclear factor erythroid 2-related factor 2 (Nrf2) to limit inflammation and modulate type I interferons.²⁵ DI could ameliorate skin pathology in a mouse model of psoriasis by targeting the DI-IκBζ regulatory axis.²⁶ Furthermore, DI exerted anti-inflammatory effects by inhibiting succinate dehydrogenase (SDH), regulating succinate levels, and controlling mitochondrial respiration changes in lipopolysaccharide (LPS)-activated macrophages.^27,28 From the above, this clearly indicates the importance of ITA in active macrophages during immunomodulation and inflammation. Nevertheless, ITA and its esterified derivatives are significantly different in terms of metabolic, electrophilic, and immunologic profiles.²⁹ Therefore, direct application of unmodified ITA will help better understand the immunomodulatory effects as well as the mechanism of action.

Here, we constructed artificial apoptotic cells (A-Cells) loaded with unmodified ITA (AI-Cells) (Figure S1). Because of morphological and membrane composition similarity to apoptotic cells, AI-Cells are localized predominantly to the liver and subsequently delivered precisely to hepatic macrophages in acetaminophen (APAP)-induced ALF mouse model. Furthermore, AI-Cells ameliorated APAP-induced liver injury through macrophage regulation. Notably, by training anti-inflammatory memory-like hepatic macrophages, AI-Cells significantly protects mice from liver reinjury. Mechanistically, AI-Cells suppressed macrophage caspase-1 activation and NLRP3 inflammasome-mediated cleavage of pro-IL-1β into its active form. We developed a potential AI-Cells therapy for liver failure through finely modulating hepatic macrophages to orchestrate liver homeostasis.

Results

A-Cells are predominantly localized in the injured liver and precisely delivered to macrophages

During apoptosis, phosphatidylserine (PS) acting as a key “eat-me” signal is exposed on the surface of apoptotic cells, which promotes the specific recognition by the macrophages and subsequent internalization of the corpse.^30,31 Thus, taking advantage of the high affinity of macrophages for PS, A-Cells were constructed for targeted drug delivery to macrophages (Figure 1A). Positively charged stearylamine (SA) liposomes (lipos) with a particle size of about 2 μm and negatively charged PS lipos with a particle size of about 100 nm were prepared by the thin film hydration method. The small-sized negatively charged PS lipos were adsorbed around the large-sized positively charged SA lipos to construct biomimetic artificial apoptotic cells via electrostatic interactions. To investigate whether the uptake of A-Cells by macrophages could mimic the apoptotic cells, the PS contents of A-Cells were optimized. We found a positive correlation between the PS contents and the uptake efficiency of A-Cells by macrophages (Figures S2A and S2B). As enhanced pro-uptake capacity of A-Cells was favorable to their application, 15% PS-containing A-Cells thereby were selected for the following experiments. Using transmission electron microscopy (TEM), we verified that A-Cells had a morphology similar to apoptotic cells. Specifically, small-sized lipos in the outer layer like apoptotic blebs surrounded the giant lipos, which provided topological context for the enhanced binding of macrophages (Figure 1B). The confocal images further confirmed that the green fluorescent giant SA lipos were encircled by the red fluorescent small-sized PS lipos (Figure 1C). Next, using Förster resonance energy transfer (FRET), we observed FRET from the donor DiI-labeled SA lipos to acceptor DiD-labeled PS lipos in A-Cells, indicating the electrostatic interactions between these two lipos (Figure 1D). These results suggested that A-Cells were successfully constructed by mimicking apoptotic cells with the exposure of PS on the outer layer, micro-sized phospholipid vesicles, and apoptotic bleb-like topological surface.

Figure 1.

A-Cells robustly promoted liver accumulation and hepatic macrophage targeting by mimicking apoptotic cells

(A) Schematic illustration of enhanced macrophage uptake of A-Cells by recognizing the “eat me” signal and surface topology.

(B) Transmission electron microscope images of A-Cells. Scale bars: 500 μm (inner) and 200 μm (outer).

(C) The CLSM image of A-Cells. SA lipos were stained with DiI, and PS lipos were stained with DiD (green, SA lipos; red, PS lipos). Scale bars: 5 μm.

(D) Analysis of FRET between DiI-labeled SA lipos and DiD-labeled PS lipos in A-Cells. FRET was observed in the emission spectra at excitation wavelength of 525 nm (b). (a) SA lipos stained with DiI and (b) A-Cells stained with DiI and DiD.

(E) Schematic illustration of macrophage targeting mechanism of A-Cells in a hepatic sinusoidal capillary. A-Cells were not able to penetrate the sinusoidal endothelial fenestrae and interact with hepatocytes, thus achieving selective macrophage targeting.

(F) Fluorescence images of cellular uptake of DiI-labeled conventional lipos and A-Cells (red) after incubation with Raw-264.7 cells in the upper chamber and AML-12 cells in the lower chamber of the Transwell system for 1 h. Cell nuclei were stained with Hoechst 33342 (blue). Scale bars: 200 nm.

(G and H) Representative fluorescence intensities and mean fluorescence intensities with (G) Raw-264.7 cells in the upper chamber and (H) AML-12 cells in the lower chamber after incubation of DiI-labeled lipos and A-Cells by flow cytometry (n = 3 biological replicates).

(I–K) The uptake kinetics of lipos and A-Cells by macrophages. DiI-labeled lipos or DiI-labeled A-Cells were incubated with macrophages for the indicated times. Representative fluorescence intensities and mean fluorescence intensities at different time points were determined using flow cytometry (n = 3 biological replicates).

(L) Cell viability of Raw-264.7 cells treated with different phospholipid concentrations of lipos and A-Cells by using MTT assay (n = 6 biological replicates).

(M and N) In vivo biodistribution of lipos and A-Cells in an APAP-induced ALF mouse model at 24 h. Lipos and A-Cells were labeled with DiD. Ex vivo fluorescence imaging of major organs harvested at 24 h.

(O) Fluorescence images of liver sections harvested from APAP-induced ALF mice with post-injection of DiD-labeled A-Cells. Colocalization of DiD-labeled A-Cells (green) with Cy3-F4/80 antibody-labeled hepatic macrophages (red) was observed. Nuclei were stained with DAPI (blue). Scale bars: 100 or 50 μm. Colocalization analysis of DiD-labeled A-Cells (green curves) and Cy3-F4/80 antibody-labeled hepatic macrophages (red curves) by ImageJ software.

All data are expressed as mean ± SD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, and ns (not significant) is p > 0.05 by Student’s t test or two-way ANOVA.

The diameter of sinusoidal endothelial fenestrae (SEF) in liver sinusoidal endothelial cells is 100–150 nm and can be widened to a maximum of 320 nm during liver failure.^32,33,34 SEF allows the passage of small-sized nanoparticles from the sinusoidal lumen to the perisinusoidal space, where they are taken up by hepatocytes, thereby reducing macrophage targeting (Figure 1E).³² We used a Transwell system to mimic the sinusoidal capillary microenvironment, and investigated the targeting efficiency of A-Cells on macrophages. The results demonstrated that A-Cells with a particle size of about 2 μm hardly crossed the membrane pores of the Transwell system, resulting in strong cellular uptake by macrophages in the upper chamber, but almost not by hepatocytes in the lower chamber. However, conventional lipos with a particle size of about 100 nm did not show selective cellular uptake (Figures 1F–1H). We further examined the uptake kinetics of A-Cells and the results indicated a rapid and strong uptake by macrophages in comparison with lipos (Figures 1I–1K and S2C). Additionally, the phagocytosis of macrophages in ALF was examined by cellular uptake efficiency of A-Cells. The results demonstrated that the phagocytosis of surviving Kupffer cells was not affected after treatment of APAP overdose compared with normal Kupffer cells, although the number of Kupffer cells was reduced (Figures S2D–S2F). Selective cellular uptake of A-Cells was mainly attributed to the following three aspects. First, the increase in the size of A-Cells relative to lipos prevented them from penetrating the SEF, thus achieving selective macrophage targeting. Second, the exposure of PS on the outer layer of A-Cells increased the specific engulfment through the PS receptors on macrophages. Third, A-Cells having a high surface roughness improved their binding to macrophages and cellular internalization. Additionally, as A-Cells did not significantly affect the viability of macrophages and hepatocytes (Figures 1L and S2G), this ensured that liver injury was not exacerbated during ALF treatment.

Next, to investigate whether A-Cells were localized in the injured liver, the biodistribution of DiD-labeled A-Cells was assessed in the APAP-induced ALF mouse model using in vivo imaging system (IVIS) Spectrum (Figure S2H). The imaging and in vivo biodistribution analysis showed that the fluorescence intensity of 15% PS-containing A-Cells in the injured liver was significantly higher than that of either lipos or PS-free A-Cells. Furthermore, 83.65% of 15% PS-containing A-Cells were accumulated in the liver, while the fluorescence signals in other major organs were relatively weak (Figures 1M, 1N, S2I, and S2J). Importantly, both the overlay fluorescence images and the fluorescence intensity profiles showed that DiD-labeled A-Cells (green) almost completely colocalized with Cy3-F4/80 antibody-labeled hepatic macrophages (red) in liver tissues, suggesting that hepatic macrophages were the primary target cells of A-Cells (Figure 1O). Meanwhile, A-Cells did not increase the liver organ index, alanine aminotransferase (ALT), and aspartate aminotransferase (AST) of the mice with liver injury (Figures S2K and S2M). Taken together, these data demonstrated that A-Cells constructed by mimicry of apoptotic cells were predominantly accumulated in the injured liver and further precisely delivered to macrophages.

AI-Cells rescue ALF

To overcome poor membrane permeability of ITA and play its regulatory role in inflammation, we developed A-Cells loaded with ITA (AI-Cells) by using the pH-gradient method (Figure 2A). AI-Cells displayed the encapsulation efficiency of 63.6% ± 2.7% and the concentration of 24.44 μmol/mL for ITA. As shown in Figures 2B and 2C, the particle size and zeta potential of AI-Cells were 2,174.7 ± 331.9 nm and −18.5 ± 3.3 mV, respectively. Then, reverse-phase high-performance liquid chromatography (RP-HPLC) analysis showed that AI-Cells markedly improve the membrane permeability of ITA, while free ITA was barely taken up by macrophages (Figure S3A). Moreover, flow cytometry analysis showed that fluorescence intensity of A-Cells loaded with a fluorescent dye RhB in macrophages was markedly stronger than RhB at various concentrations (Figures 2D and 2E). These data demonstrated that AI-Cells could increase the membrane permeability and the intracellular accumulation of ITA.

Figure 2.

AI-Cells improved membrane permeability of itaconic acid and augmented therapeutic efficacy in ALF mice

(A) Schematic illustration of AI-Cells preparation by the pH-gradient method.

(B and C) Size distribution (B) and (C) zeta potential of AI-Cells and other samples (n = 6 biological replicates).

(D and E) The mean and representative fluorescence intensities of macrophages ingesting RhB or A-Cells-RhB with different RhB concentrations after incubation 4 h (n = 3 biological replicates).

(F) Representative photographs of dissected liver from treated or untreated ALF mice after euthanasia.

(G–J) Serum ALT, AST, ALP, and TBIL levels at 24 h after APAP intoxication (n = 3 biological replicates). ∗Comparison with the APAP group; #comparison with the AI-Cells group.

(K) Survival curves of ALF mice that received various treatment 24 h after the administration of APAP (n = 10 biological replicates per group).

(L–N) Organ index (n = 5 biological replicates) (L), (M) body weights (n = 5 biological replicates), and (N) changes in body temperature of ALF mice before and after treatment (n = 5 biological replicates).

(O) Representative liver H&E staining from ALF mice before and after treatment. Scale bars: 400 μm (top) and 100 μm (bottom).

(P) Quantification of necrotic areas by ImageJ software (n = 10 technical replicates).

All data are expressed as mean ± SD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, ^#p < 0.05, ^##p < 0.01, ^###p < 0.001, ^####p < 0.0001, and ns is p > 0.05 by one-way or two-way ANOVA.

Next, the therapeutic efficacy of AI-Cells in a mouse model of APAP-induced ALF was evaluated using the experimental outline presented in Figure S2H. N-acetylcysteine (NAC) used as a first-line therapy for ALF was the positive control. Severe liver injury in mice that received injections of overdose APAP was evident by a large area of hepatic congestion, increased roughness, and a substantially elevated serum levels of liver function biomarkers (ALT, AST, alkaline phosphatase [ALP], and total bilirubin [TBIL]), as shown in Figures 2F–2J. Strikingly, it was noted that the large area of hepatic congestion disappeared and the surface was as smooth as normal livers after the treatment of AI-Cells (Figures 2F, S3B, and S3G). Importantly, the liver enzyme levels of serum ALT, AST, ALP, and TBIL dropped even more sharply than those of NAC treatment, and almost went back to normal levels (Figures 2G–2J, S3C, and S3D). However, treatment with A-Cells or free ITA did not result in a statistically significant reduction in liver function markers compared with the untreated APAP group. Death from ALF usually results from loss of liver functions.³⁵ Thus, we further investigated the survival of ALF mice after treatment of AI-Cells. As shown in Figure 2K, therapeutic administration of AI-Cells after lethal APAP treatment markedly improved the survival rate up to 90%, compared with untreated ALF mice or the mice treated with free ITA. Additionally, healthy mice maintained a constant organ index and body temperature. Although the body weights in all groups were stable during the treatment, a significant increase in organ index of liver was observed in the APAP, NAC, A-Cells, and ITA groups compared with the sham group, likely because of liver hyperemia and edema. However, no obvious difference was observed in organ index between the AI-Cells group and the sham group (Figures 2L, 2M, S3E, and S3F). The body temperature of APAP-induced ALF mice continued to decrease as the disease progressed, while this progression was slowed down in the ALF mice treated with NAC and almost blocked in the ALF mice treated with AI-Cells (Figure 2N). Furthermore, histological analysis of H&E-stained liver sections confirmed that APAP caused severe liver necrosis evidenced by distinct hepatocyte disruption, while the AI-Cells treatment achieved a dramatically reduction in necrotic areas (p < 0.0001) (Figures 2O and 2P). Taken together, these results suggested that AI-Cells effectively reduced severity in liver injury and rescued ALF lethality in mice through improving membrane permeability and intracellular accumulation of ITA.

AI-Cells regulate macrophage inflammatory responses to reduce the production of pro-inflammatory cytokines

To understand the role of the AI-Cells treatment in recovery from ALF, we co-cultured AI-Cells with hepatocyte cells or macrophages exposed to APAP in vitro, including AML-12 cells (a non-tumorigenic mouse hepatocyte cell line), LO2 cells (a human normal hepatocyte cell line), BMDMs (mouse primary bone marrow-derived macrophages), Raw-264.7 cells (a murine macrophage cell line), and LX2 cells (a human hepatic stellate cell line), followed by cell viability assays. AI-Cells had no significant impact on the proliferation of the above cells, indicating that AI-Cells did not directly promote the proliferation of hepatic parenchymal cells exposed to APAP, or suppress the proliferation of hepatic stellate cells to achieve the therapeutic effect on ALF (Figures 3A–3C and S4A–S4C). Given that the esterified derivatives of ITA limits the inflammatory response in activated macrophages,^25,26,36 we next focused on macrophages to investigate how AI-Cells alleviated ALF. LPS stimulated Raw-264.7 cells were treated with different concentrations of AI-Cells, and then the levels of the pro-inflammatory cytokines IL-1β, IL-6, and TNF-α were measured. The qRT-PCR assays showed that AI-Cells treatment induced activated macrophages to significantly inhibit the expression of IL-1β, IL-6, and TNF-α mRNA in a dose-dependent manner, compared with that of the LPS-activated macrophages (Figures 3D–3F). Of note, although free ITA could also reduce the levels of IL-1β, IL-6, and TNF-α, the effective concentration of ITA was much greater than AI-Cells (Figures S4D–S4F), suggesting that the use of AI-Cells greatly increased the accumulation of ITA in macrophages because of the enhanced transmembrane diffusion capacity of AI-Cells relative to free ITA, and exhibited excellent anti-inflammatory properties at low concentrations (Figures 2D, 2E, and S3A). Moreover, the mRNA expression of these pro-inflammatory factors (IL-1β, IL-6, IL-3, inducible nitric oxide synthase [iNOS], TNF-α, and IL-12) in the APAP-injured liver of mice after treatment with AI-Cells was also significantly suppressed (Figures 3G and 3H). On the other hand, we found no statistically significant difference in the mRNA expression of anti-inflammatory cytokines between AI-Cells and APAP-treated mice (Figure 3I). Consistent with these findings, the protein expression of these pro-inflammatory factors (IL-1β, IL-6, TNF-α, and iNOS) in liver tissues was also markedly downregulated after AI-Cells administration, whereas anti-inflammatory cytokines (IL-10 and Arg-1) levels were not changed compared with the APAP treatment (Figures 3J and 3K). Next, to address the role of AI-Cells treated macrophages in reduced expression of the pro-inflammatory cytokines in injured livers, we determined changes in macrophage surface markers using flow cytometry (Figure 3M). APAP-injured livers showed a marked increase in population of CD86⁺ pro-inflammatory hepatic macrophages. Following AI-Cells treatment, the proportion of CD86⁺ pro-inflammatory macrophages sharply reduced and went back to that of normal livers 24 h after the APAP intoxication (Figures 3L, 3N, S4G, and S4H). Additionally, either APAP injury or AI-Cells treatment did not show meaningful changes in the proportion of CD206⁺ anti-inflammatory macrophages. Taken together, these findings suggested that AI-Cells orchestrated liver homeostasis by limiting the expression of pro-inflammatory cytokines in activated macrophages.

Figure 3.

AI-Cells modulated pro-inflammatory macrophages to limit the production of pro-inflammatory cytokines

(A–C) Cell viability of (A) AML-12 cells, (B) BMDMs, and (C) LX2 cells pretreated with APAP (5 mM) and then incubated with A-Cells, ITA, or AI-Cells for 24 h, respectively (n = 4 or 8 biological replicates).

(D–F) Down-regulation of (D) IL-1β, (E) IL-6, and (F) TNF-α mRNA expression in macrophages by AI-Cells (n = 3 biological replicates).

(G and H) Down-regulation of pro-inflammatory cytokine mRNA levels of mouse liver tissues (n = 3 biological replicates).

(I) mRNA levels of anti-inflammatory cytokines in mouse liver tissues (n = 3 biological replicates). The mice were treated with AI-Cells and sacrificed 24 h after APAP intoxication. Levels of mRNA were normalized to those of β-actin.

(J) Western blot of IL-1β, IL-6, TNF-α, iNOS, IL-10, Arg-1, and β-actin with protein lysates from liver tissues.

(K) Densitometric analysis of proteins by ImageJ software (n = 3 biological replicates).

(L) Flow cytometry analysis of CD86⁺ or CD206⁺ macrophages in F4/80⁺ hepatic macrophages 24 h after APAP intoxication.

(M) Schematic illustration of the isolation of hepatocytes and non-parenchymal cells from the livers collected from mice of sham, APAP, and AI-Cells groups. MΦ, macrophages; HSC, hepatic stellate cells, Hep, hepatocytes; SEC, sinusoidal endothelial cells.

(N) Quantification of the percentage of CD86⁺ or CD206⁺ cells in F4/80⁺ cells (n = 3 biological replicates).

All data are expressed as mean ± SD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001, and ns is p > 0.05 by one-way or two-way ANOVA.

AI-Cells ameliorate hepatocyte injury by regulating pro-inflammatory activation of macrophages

Given that AI-Cells did not directly improve the survival of APAP-injured hepatocytes, but had an immunomodulatory effect on macrophages, we further investigated whether AI-Cells could rescue injured hepatocytes via macrophage regulation. To test this, the highly inflammatory hepatic macrophage population including recruited BMDMs and tissue-resident Kupffer cells was used in vitro model. As shown in Figure 4A, primed BMDMs or Kupffer cells were treated with LPS for 3 h; the medium was then replaced with medium containing A-Cells, ITA, or AI-Cells for 45 min followed by addition of ATP for another 45 min; and finally BMDM or Kupffer cell conditioned medium was collected and incubated with APAP-injured AML-12 cells for 24 h. As expected, both BMDM conditioned medium (B-LPS&ATP CM) and Kupffer cell conditioned medium (K-LPS&ATP CM) from the model group significantly reduced survival of AML-12 cells (Figures S5A and S5B). Notably, after AI-Cells treatment of LPS&ATP-activated macrophages, BMDM conditioned medium (B-AI-Cells CM) and Kupffer cell conditioned medium (K-AI-Cells CM) markedly reduced the cell death of AML-12 hepatic cells, compared with conditioned medium from untreated LPS&ATP-activated BMDMs (B-LPS&ATP CM) (Figures 4B–4D) or Kupffer cells (K-LPS&ATP CM) (Figures 4E and 4F). Additionally, the decrease in cell death of AML-12 cells was not observed in the B-A-Cells CM and B-ITA CM groups. Next, we analyzed the apoptosis/necrosis of AML-12 cells exposed to APAP followed by B-AI-Cells CM treatment by flow cytometry. We found that the B-A-Cells CM and B-ITA CM groups had relatively high levels of apoptosis with up to about 85%, similar to that of the B-LPS&ATP CM group. However, the B-AI-Cells CM group suppressed APAP-induced apoptosis of AML-12 hepatic cells and the apoptosis rate decreased by about 10% compared with the B-LPS&ATP CM group (Figures 4G, 4H, and S5C), which was consistent with the results of cell viability assays. Furthermore, cell proliferation on the basis of immunohistochemical Ki67 biomarker expression on liver tissues demonstrated that AI-Cells treatment induced a marked increase in proliferating cells in the APAP-injured liver (Figures 4I–4J). Collectively, these results provided evidence that AI-Cells indirectly promoted hepatocytes survival and proliferation via the direct regulation of microphages.

Figure 4.

Conditioned medium from BMDMs/Kupffer cells treated with AI-Cells reduced cell death and improved cell proliferation in recipient AML-12 cells

(A) Scheme illustrating experimental setup. After BMDMs/Kupffer cells had been exposed to LPS for 3 h and incubated with A-Cells, ITA, or AI-Cells for 45 min, ATP was added. After 45 min, the supernatant medium was collected and administrated to APAP-injured AML-12 cells.

(B) Quantification of viable AML-12 cells treated with various BMDM conditioned medium by MTT assay (n = 5 biological replicates). B-LPS&ATP CM was set up as a control.

(C and D) Representative microscopy images and quantification of AML-12 cells treated with various BMDM conditioned medium and stained with crystal violet (n = 3 biological replicates). Scale bars: 400 μm. B-LPS&ATP CM was set up as a control.

(E and F) Representative microscopy images and quantification of AML-12 cells treated with various Kupffer cell conditioned medium and stained with crystal violet. Scale bars: 1000 μm. B-LPS&ATP CM was set up as a control.

(G) Detection of apoptosis in AML-12 cells treated with various BMDM conditioned medium by annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) staining.

(H) Quantitative apoptosis analysis of AML-12 cells (n = 3 biological replicates).

(I) Ki67 staining of liver tissues from the sham, APAP, NAC, A-Cells, ITA, or AI-Cells-treated ALF mice after euthanasia. Scale bars: 50 μm.

(J) Quantification of the Ki67-positive cells (n = 10 technical replicates). The number of positive cells was calculated for at least ten random fields for each slide in a blinded fashion.

All data are expressed as mean ± SD; ∗∗p < 0.01, ∗∗∗∗p < 0.0001, and ns is p > 0.05 by Student’s t test or one-way ANOVA.

Neither TNF-α nor IL-6 is the crucial regulator for rescuing APAP-injured hepatocytes by AI-Cells

Having established that AI-Cells ameliorated hepatocyte injury by macrophage regulation, we next sought to identify the crucial cytokines regulated by AI-Cells in activated macrophages. Previous evidence showed that reduced expression of certain pro-inflammatory cytokines, such as TNF-α, IL-6, IL-1α, IL-1β, and IL-22, could protect mice against acute liver injury (Table S1). Combined with our experimental results related to down-regulation of pro-inflammatory cytokines after AI-Cells treatment (Figures 3G, 3H, 3J, 3K, and S5D), we focused on TNF-α, IL-1β, and IL-6 as the potential key pro-inflammatory cytokines regulated by AI-Cells. First, we used the macrophages^ΔASC population to investigate the roles of TNF-α and IL-6 regulated by AI-Cells in rescuing APAP-injured hepatocytes. Macrophages^ΔASC is non-response to IL-1β because of the lack of apoptosis-associated speck-like protein containing a CARD (ASC) and the inability to activate intracellular NOD, LRR, and pyrin domain-containing protein 3 (NLRP3) inflammasome in the absence of signal 2 (such as ATP, nigericin, and so on) stimulation, resulting in the failure of IL-1β secretion.^37,38 Macrophages^ΔASC was stimulated with LPS and the changes in the secretion levels of these three inflammatory cytokines in the supernatant of LPS-primed macrophages^ΔASC treated with AI-Cells were determined by ELISA. Indeed, the results demonstrated that the levels of TNF-α and IL-6 secreted by macrophages^ΔASC elevated significantly after LPS stimulation, while the level of IL-1β did not change markedly (Figures 5A–5C), suggesting a cell model of inflammation in response to TNF-α and IL-6 but not IL-1β was established (Figure 5D). However, both TNF-α and IL-6 production by macrophages^ΔASC dramatically decreased after 24 h treatment with AI-Cells (Figures 5A and 5B). Next, macrophages^ΔASC conditioned medium from various groups was applied to recipient cells, and the M^ΔASC-LPS CM markedly decreased survival of AML-12 hepatic cells (Figure S5E). Surprisingly, the survival rate of the M^ΔASC-AI-Cells CM group was not markedly different from that of the M^ΔASC-LPS CM, M^ΔASC-A-Cells CM, and M^ΔASC-ITA CM groups, suggesting that the reduction of TNF-α and IL-6 in macrophages^ΔASC conditioned medium did not improve AML-12 cell survival (Figure 5E). Subsequently, to further investigate the effect of TNF-α and IL-6 regulated by AI-Cells on injured hepatocytes, APAP-injured AML-12 cells were dual cultured with LPS-activated macrophages^ΔASC treated with AI-Cells in two chambers separated by a 0.4 μm Transwell membrane, as shown in Figure 5F. After 24 h dual-culture period, a live/dead cell staining assay was performed to directly visualize cell viability by fluorescence microscopy. There was no significant difference between the M^ΔASC-AI-Cells CM and M^ΔASC-LPS CM groups (Figure 5G), which was consistent with the results of macrophages^ΔASC conditioned medium in monoculture. Correspondingly, there was no notable difference in apoptosis among each group (Figures 5H and 5I). Collectively, the results clearly illustrated that TNF-α and IL-6 were not the crucial pro-inflammatory cytokines for rescuing injured hepatocytes by AI-Cells.

Figure 5.

TNF-α and IL-6 were not key pro-inflammatory cytokines in the effect of macrophages on hepatocytes

(A–C) The levels of (A) TNF-α, (B) IL-6, and (C) IL-1β in the supernatant of macrophages^ΔASC in various groups (n = 3 biological replicates).

(D) Schematic representation of a model of inflammation in macrophages^ΔASC responding to TNF-α and IL-6 but not IL-1β.

(E) Quantification of viable AML-12 cells by MTT assay (n = 6 biological replicates). M^ΔASC-LPS CM was set up as a control.

(F) Schematic diagram depicting the cell dual culture.

(G) Calcein-AM staining of AML-12 cells with different treatments. Scale bars: 400 μm.

(H) Annexin V-FITC/PI assay for apoptosis detection of AML-12 cells under treatment with M^ΔASC-LPS CM, M^ΔASC-A-Cells CM, M^ΔASC-ITA CM, or M^ΔASC-AI-Cells CM.

(I) Quantitative apoptosis analysis of AML-12 cells (n = 3 biological replicates).

All data are expressed as mean ± SD; ∗p < 0.05, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, and ns is p > 0.05 by one-way ANOVA.

AI-Cells inhibit the priming of macrophage NLRP3 inflammasomes and block mature IL-1β release

Given that reduction of neither TNF-α or IL-6 exerted the crucial actions in attenuation of liver injury, we further identified the key cytokines regulated by AI-Cells in activated hepatic macrophages. To determine whether the reduction of the IL-1β by AI-Cells promotes hepatocyte survival, we used the highly inflammatory macrophages (BMDMs and Kupffer cells) activated by LPS and ATP that could respond to TNF-α, IL-1β, and IL-6 simultaneously (Figure 4A). As shown in Figure 6A, the protein levels of TNF-α, IL-1β, and IL-6 were markedly elevated in the supernatant of LPS&ATP-primed BMDMs or Kupffer cells (B-LPS&ATP CM or K-LPS&ATP CM) compared with unprimed controls. However, when activated BMDMs and Kupffer cells were treated with AI-Cells, the level of IL-1β, but not TNF-α and IL-6 was dramatically decreased in the B-AI-Cells CM and K-AI-Cells CM groups compared with the B-LPS&ATP CM group and the K-LPS&ATP CM group, suggesting that AI-Cells blocked the secretion of IL-1β from activated macrophages (Figure 6B). Next, APAP-injured AML-12 hepatic cells were exposed for 24 h to the above-mentioned conditioned medium of macrophages. As a response of the B-AI-Cells CM exposure, the cell viability of AML-12 cells was markedly increased compared with the B-LPS&ATP CM group. To further confirm that the reduction of IL-1β secreted by activated hepatic macrophages could effectively promote hepatocyte survival, the B-AI-Cells CM was supplemented with recombinant IL-1β protein and then administrated to AML-12 cells. Strikingly, exposure to the B-AI-Cells CM containing exogenous IL-1β subsequently led to a significant decrease in the viability of AML-12 cells compared with the B-AI-Cells CM group (Figure 6C). In contrast, using anti-IL-1β antibody to neutralize IL-1β in K-LPS&ATP CM markedly augmented survival of AML-12 cells (Figures 6D and 6E). Collectively, these data suggested that AI-Cells reducing IL-1β production by activated hepatic macrophages could effectively rescue injured hepatocytes.

Figure 6.

AI-Cells controlled liver injury by inhibiting macrophage NLRP3 inflammasome activation and reducing IL-1β production

(A) The levels of IL-1β, TNF-α, and IL-6 in the supernatant of BMDMs/Kupffer cells in various groups (n = 3 biological replicates).

(B) Schematic representation of a BMDM/Kupffer cell inflammation model responding to TNF-α, IL-6, and IL-1β. AI-Cell treatment significantly reduced the concentration of IL-1β but had no significant effect on TNF-α and IL-6.

(C) Quantification of viable AML-12 cells after IL-1β exposure (n = 5 biological replicates). The AML-12 cells were cultured in the B-AI-Cells CM supplemented with recombinant IL-1β. B-LPS&ATP CM was set up as a control.

(D) Survival of AML-12 cells after K-LPS&ATP CM and IL-1β antibody exposure (n = 3 biological replicates). K-LPS&ATP CM was set up as a control.

(E) Representative microscopy images of AML-12 cells treated with various conditioned medium from Kupffer cells and stained with crystal violet. Scale bars: 1,000 μm.

(F) Western blot of NLRP3, pro-Casp1, Casp1 p20, and β-actin with protein lysates from liver tissues of sham, APAP, and AI cells groups.

(G) Western blot of IL-1β with protein lysates from liver tissues of sham, APAP, and AI-Cells groups.

(H–K) Densitometric analysis of NLRP3, pro-Casp1, Casp1 p20, and IL-1β proteins by ImageJ software (n = 3 biological replicates).

(L) IL-1β staining of liver tissues from the sham, APAP, or AI-Cell-treated ALF mice after euthanasia. Scale bars: 200 μm.

(M) Schematic diagram of AI-Cells blocking activation of NLPR3 inflammasome.

(N) Docking mode of NLRP3 (white surface and green cartoon) with itaconic acid (cyan sticks).

(O) Three-dimensional (3D) interaction mode of NLRP3 (green cartoon) and itaconic acid (cyan sticks). The key residues are shown as green sticks and the hydrogen bond is shown in yellow dashed lines.

All data are expressed as mean ± SD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, and ns is p > 0.05 by one-way ANOVA.

Activation of NLRP3 inflammasomes is necessary to induce the maturation and release of IL-1β to trigger an inflammatory response.^39,40 Therefore, we next investigated the effects of AI-Cells on NLRP3/ASC/caspase-1/IL-1β signaling pathway. The levels of key proteins involving the this signaling pathway in the liver, including NLRP3, pro-caspase-1 (pro-Casp1), caspase-1 p20 (Casp1 p20), and IL-1β, were significantly increased after APAP intoxication. Interestingly, treatment with AI-Cells dramatically attenuated the expression of Casp1 p20 and IL-1β, but not NLRP3 and pro-Casp1, compared with that with APAP (Figures 6F–6K). Also, treatment with AI-Cells led to the reduction of IL-1β levels in the liver as assessed by IHC analysis (Figure 6L). These data revealed the inability of Casp1 p20 to be activated to cleave pro-IL-1β, which ultimately reduced IL-1β maturation (Figure 6M). AI-Cells could rapidly regulate hepatic macrophages within 24 h in vivo, likely because of rapid inhibition of NLRP3 inflammasomes in macrophages.^41,42 However, current evidence indicated that ITA was unlikely to directly inhibit Casp1 p20 activity, as only modest inhibition could be observed with the treatment of nonphysiologically high concentrations of 20 mM ITA.⁴³ Notably, we presented evidence via a molecular docking study that ITA in AI-Cells directly bound NLRP3 by forming a covalent bond with residue CYS548 on NLRP3, and a hydrogen bond interacted with GLY547 at a distance of 2.5 Å (Figures 6N and 6O). Recent work has demonstrated that itaconate interacting with NLRP3 resulted in disrupting the NLRP3-NEK7 (the mitotic kinase NIMA-related kinase 7) interaction, and subsequently inhibiting the self-assembly of NLRP3 inflammasomes.³⁸ Our study also verified that suppression of NLRP3 inflammasome activation reduced IL-1β production, and subsequently attenuated inflammatory cell recruitment (Figures S6A–S6C). Overall, these findings provided evidence that AI-Cells inhibited the priming of macrophage NLRP3 inflammasomes and blocked the release of mature IL-1β, which ultimately reversed severe liver injury.

AI-Cells train anti-inflammatory memory-like hepatic macrophages to protect against liver reinjury

During liver reinjury, activation of Kupffer cells and neutrophils induces the production of large amounts of inflammatory cytokines, thereby exacerbating liver injury.⁴⁴ In addition, patients with primary liver failure are more susceptible to liver reinjury, resulting in increased mortality.^8,9 Therefore, we sought to investigate whether AI-Cells therapy could offer protection against liver reinjury by driving innate immune cells to acquire anti-inflammatory memory. BMDMs isolated from healthy mice were stimulated with LPS&ATP as the first challenge and subsequently treated with AI-Cells or PBS. AI-Cells trained macrophages while defending against the initial challenge. Next, trained BMDMs were obtained and then exposed to the second challenge with LPS&ATP after 5 days of rest (Figure 7A). Given that the pro-inflammatory cytokine IL-1β was a key cytokine affected by AI-Cells and played a crucial role as a regulator in inflammatory responses, IL-1β released into the supernatant of BMDMs before and after each challenge was detected. Notably, the level of IL-1β in the supernatant from BMDMs treated with AI-Cells or PBS returned to the baseline after 5 days of resting. Strikingly, in response to LPS and ATP, which served as the second challenge in the trained immunity process, the production of IL-1β was remarkably less in AI-Cells-treated BMDMs compared with PBS-treated BMDMs (Figure 7B). These data implied that AI-Cells-induced beneficial effects could be reactivated owing to training anti-inflammatory memory-like macrophages.

Figure 7.

AI-Cells induced anti-inflammatory memory-like macrophages to protect recovered mice against liver reinjury

(A) Schematic of in vitro anti-inflammatory memory-like macrophages experimental outline.

(B) The anti-inflammatory memory-like macrophage marker IL-1β levels in the PBS or AI-Cells groups at baseline, after first challenge/training, before second challenge, and after second challenge (n = 4 biological replicates).

(C) Schematic of in vivo liver reinjury experimental outline.

(D and E) Representative pictures and quantification showing changes in body temperature of mice treated with saline or AI-Cells at baseline, 6 h after first challenge, after first challenge, before second challenge, 6 h after second challenge, after second challenge, before third challenge, 6 h after third challenge, and after third challenge (n = 3 biological replicates).

(F) Changes in the serum IL-1β levels of mice treated with saline or AI-Cells at baseline, after first challenge/training, before second challenge, after second challenge, before third challenge, and after third challenge (n = 3 biological replicates).

(G) The heatmap of serum IL-1β levels in the AI-Cells group at baseline, after first challenge/training, before second challenge, after second challenge, before third challenge, and after third challenge (n = 3 biological replicates).

(H) Expression heatmaps of functional genes participated in drug, glucose, fatty acid, and cholesterol metabolism in the livers of sham, model, and AI-Cells (1 and 15 days) groups (n = 3 biological replicates).

All data are expressed as mean ± SD. All statistical tests used in this figure are one-way or two-way ANOVA, and p < 0.05 is considered to indicate statistical significance.

Next, to further investigate whether AI-Cells could train anti-inflammatory memory-like macrophages to protect against liver reinjury, we re-challenged ALF mice with overdose APAP on days 7 and 14 after the first challenge (day 0), and AI-Cells were administered intravenously only on the first challenge (Figure 7C). Body temperature is an important indicator reflecting the survival status of mice after APAP challenge. The results demonstrated that upon the first challenge of a lethal dose of APAP, the body temperature of mice in the model group continued to decrease within 24 h, which was significantly lower than the normal body temperature of mice, and the mortality rate of mice was close to 100% within 48 h. However, the body temperature of ALF mice in the AI-Cells group decreased slightly at 6 h after the first challenge and returned to normal at 24 h, and the mortality rate of mice was reduced to approximately 50%. Furthermore, the AI-Cells-treated mice that survived after the first challenge showed increased resistance to liver reinjury. Specifically, all these mice were survived and the body temperature of the mice was maintained at a normal level before and after the challenge (Figures 7D and 7E). We next examined whether this effect was correlated with IL-1β production and release which were indicative of anti-inflammatory memory-like macrophages. Upon the first challenge, serum IL-1β levels in the AI-Cells group were significantly lower than those in the APAP model group. Concurrently, serum IL-1β levels in the AI-Cells group were maintained at a relatively stable level before and after the second and third challenges, which prevented over-activation of the inflammatory response and protected the mice from liver reinjury, implying that the anti-inflammatory memory-like macrophages was established. Although serum IL-1β levels after challenge were slightly lower than those before challenge in the AI-Cells group, they were not significantly different from the baseline of IL-1β in normal mice (Figures 7F and 7G). Additionally, the liver in the AI-Cells-treated group almost recovered to the normal level, and the liver function was relatively intact at the mRNA level in terms of drug metabolism, glucose metabolism, fatty acid metabolism, and cholesterol metabolism (Figure 7H). Taken together, these data suggested that AI-Cells enabled to train anti-inflammatory memory-like macrophages to prevent liver reinjury via modulation of IL-1β.

Discussion

Current strategies for ameliorating ALF predominantly focus on the rescue of hepatocytes.^45,46,47 For example, widely used NAC in the treatment of ALF converts into glutamyl cysteine in hepatocytes, subsequently generating glutathione under the action of glutathione synthase, thereby achieving hepatocyte detoxification.⁴⁸ Emerging evidence suggests that hepatic macrophages are essential for the progression of ALF.^49,50,51 Excessive inflammation by macrophages worsens liver damage and failure.⁵² In the present study, a triple-mediated precision drug delivery system for hepatic macrophages was constructed by artificial apoptotic cells loaded with ITA, AI-Cells, wherein the compositions of the synthetic plasma membrane and surface topology were rationally engineered. AI-Cells were predominantly localized to the liver and further precisely delivered to macrophages. Notably, AI-Cells robustly drove pro-inflammatory macrophages to restore homeostasis and even induced anti-inflammatory memory-like hepatic macrophages through IL-1β modulation to rescue ALF and prevent reinjury. We hereby provided evidence of macrophages being a therapeutic target and a precision drug delivery strategy for liver disease treatment.

ITA, a derivative of the macrophage TCA cycle, is derived from IRG-1-mediated decarboxylation of cis-aconitic acid in the mitochondrial matrix.⁵³ As an endogenous anti-inflammatory metabolite, ITA has been demonstrated to control inflammatory responses during macrophage activation and ameliorate sepsis and psoriasis in animal models.^25,26 Unfortunately, ITA has extremely strong polarity, resulting in poor transmembrane diffusion ability without the presence of an active transporter, so the majority of studies on ITA have applied esterified derivatives as equivalent substitutes. Nevertheless, the differences between ITA and its esterified derivatives in terms of intracellular accumulation, electrophilicity, and immunomodulatory functions have emerged with further research.^29,38 Here, we applied the artificial apoptotic cells to deliver ITA, which perfectly solved the problems of poor transmembrane property, low bioavailability, and requiring esterification for application. In cellular and mouse models, we demonstrated that AI-Cells were significantly more effective than free ITA in terms of macrophage uptake, inflammatory cytokine expression, macrophage phenotype modulation, lesion targeting, liver injury reduction, liver proliferation, and liver function recovery. Furthermore, these studies supported that ITA was a crucial determinant of innate immune responses and had profound anti-inflammatory effects. Therefore, the AI-Cells may offer a potent approach for the future investigation and application of ITA.

Prior work demonstrated that ITA affect NLRP3 inflammasome activation.³⁸ Yet 4-OI was used instead of ITA in these studies, thereby direct roles of unmodified ITA in NLRP3 inflammasome activation in APAP-induced ALF mice need to be examined. Our studies performed in the interactions of macrophages and AML-12 hepatic cells showed that the presence of AI-Cells containing ITA led to decreased IL-1β production by macrophages and increased survival of hepatocytes, establishing the importance of the ITA-NLRP3 axis in restraining hepatic macrophage inflammation. Moreover, IL-1β but not TNF-α or IL-6 was identified as a pivotal component and a key regulator in AI-Cells orchestrating hepatic macrophage homeostasis. Further evidence from the data revealed that AI-Cells enabled to limit pathogenic inflammation and rescue mice with ALF in the context of NLRP3 inflammasome activation. Additionally, our pilot data indicated that treatment of AI-Cells resulted in a reduction in LDH release (Figure S6D), which has been used for evaluation of pyroptotic cell death.³⁸ Therefore, whether other mechanism acts synergistically with inhibition of NLRP3 inflammasome activation needs to be further investigated.

A recent study showed that the IRG1-itaconate-succunate dehydrogenase axis is a central regulatory node linking innate immune tolerance and trained immunity.²² Zhang et al. demonstrate that pre-operative exercise upregulates itaconate metabolism in Kupffer cells thereby inducing anti-inflammatory trained immunity, which overpowers inflammation and reduces local organ damage.¹⁶ Our study provided evidence that AI-Cells induced anti-inflammatory memory-like macrophages by interfering with the activation of NLRP3 inflammasomes in macrophages and thus modulating the key mediator IL-1β, which unraveled the mechanisms underlying the therapeutic benefit of AI-Cells. Our data not only confirmed the crucial role of ITA in the regulation of macrophage-trained immunity but also revealed for the first time that exogenous and unmodified ITA could protect against liver reinjury through modulating IL-1β. Collectively, our study offered insights into AI-Cells precisely delivering ITA and training anti-inflammatory memory-like hepatic macrophages through IL-1β modulation to rescue ALF and prevent the liver against reinjury.

Conclusion

In summary, AI-Cells were constructed by artificial apoptotic cells loaded with ITA, an immunometabolite that has not been used for liver disease treatment before. AI-Cells offered several advantages for robustly reversing ALF. First, AI-Cells facilitated the precise delivery of ITA to hepatic macrophages via “eat-me” signaling, surface topology, and interception of SEF. Second, AI-Cells inhibited the priming of NLRP3 inflammasomes and blocked the maturation of IL-1β in macrophages, thus reducing liver inflammation and damage. Third, AI-Cells enabled the induction of anti-inflammatory memory-like macrophages that could prevent liver reinjury by modulating IL-1β signaling. Therefore, AI-Cells selectively delivering ITA to hepatic macrophages and orchestrating liver homeostasis provide a potent therapeutic strategy for ALF and other types of liver failure.

Limitations of the study

Although AI-Cells have shown great promise for treating ALF, some challenges and limitations remain to be addressed in future studies. The interactions between the IL-6-dependent genes, the NLPR3-IL-1β axis and other factors involved in liver injury and repair need to be clarified. Moreover, the contribution of IL-1β reduction to improving in vivo liver phenotypes needs to be further strengthened. Solving these issues would facilitate the clinical translation of AI-Cells therapy for various types of liver failure.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-IL-1β antibody (mouse)	Abcam	Cat# ab9722; RRID:AB_308765
Anti-TNF-α antibody (rabbit)	Abcam	Cat# ab6671; RRID:AB_305641
Anti-IL-6 antibody (mouse)	Abcam	Cat# ab229381; RRID:AB_2861234
Anti-IL-10 antibody (rabbit)	Abcam	Cat# ab9969; RRID:AB_308826
Anti-iNOS antibody (rabbit)	Abcam	Cat# ab3523; RRID:AB_303872
Anti-Arg-1 antibody (rabbit)	Abcam	Cat# ab233548; RRID:AB_2895715
Anti-NLPR3 antibody (mouse)	Abcam	Cat# ab263899; RRID:AB_2889890
Anti-β-actin antibody (mouse)	Abcam	Cat# ab8226; RRID:AB_306371
Anti-pro-Casp1/anti-Casp1 p20 antibody (mouse)	Cell Signaling Technology	Cat# 89332; RRID:AB_2923067
FITC anti-mouse F4/80 antibody	Biolegend	Cat# 123107; RRID:AB_893500
APC anti-mouse CD206 antibody	Biolegend	Cat# 141707; RRID:AB_10896057
PE anti-mouse CD86 antibody	Biolegend	Cat# 105007; RRID:AB_313150
Chemicals, peptides, and recombinant proteins
Itaconic acid (>99%)	Shanghai Macklin Biochemical Co., Ltd.	I811765; CAS: 97-65-4
Lecithin (99.8%)	Shanghai Macklin Biochemical Co., Ltd.	L812368; CAS: 8002-43-5
Stearylamine (90%)	Bide Pharmatech Ltd.	BD17851; CAS: 124-30-1
Phosphatidylserine (50%)	Shanghai Macklin Biochemical Co., Ltd.	S832149; CAS: 51446-62-9
Cholesterol (99%)	Shanghai Macklin Biochemical Co., Ltd.	C6213; CAS: 57-88-5
N-acetyl-L-cysteine (NAC)	Shanghai Macklin Biochemical Co., Ltd.	N800425
Acetaminophen (APAP)	Bide Pharmatech Ltd.	BD112079; CAS: 103-90-2
DiI	Invitrogen	V22885
DiD	Invitrogen	V22887
Rhodamine B	Invitrogen	S1307
Recombinant mouse IL-1β	MedChem Express	HY-P7073A
Insulin-transferrin-sodium (ITS) liquid media supplement	Sigma-Aldrich	I3146
Lipopolysaccharide (LPS)	Sigma-Aldrich	L2880
Adenosine triphosphate (ATP)	Sigma-Aldrich	A6419
Hoechst 33342	Sigma-Aldrich	H1399
Critical commercial assays
Annexin V-FITC/PI apoptosis detection kit	Multi Sciences Biotech, Co., Ltd.	70-AP101-100
RNAex Pro Reagent (Total RNA extraction reagent)	AG	21102
Calcein-AM staining kit	YEASEN Biotechnology	40747ES76
Experimental models: Cell lines
AML-12 cells	Dr. Huichang Bi, Sun Yat-sen university	N/A
Raw-264.7 cells	Laboratory Animal Center of Sun Yat-sen University	N/A
LX2 cells	Laboratory Animal Center of Sun Yat-sen University	N/A
L929 cells	Laboratory Animal Center of Sun Yat-sen University	N/A
Experimental models: Organisms/strains
BALB/c mice (male)	Laboratory Animal Center of Sun Yat-sen University	N/A
Oligonucleotides
Primers used for quantitative PCR	Sangon Biotech	Table S2
Software and algorithms
GraphPad Prism	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
FlowJo	FlowJo Software	https://www.flowjo.com/
ImageJ	NIH	https://imagej.nih.gov/ij/

Resource availability

Lead contact

Further information and requests for resources and regents should be directed to and will be fulfilled by the lead contact, Min Feng (fengmin@mail.sysu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Experimental models and subject details

Animal studies

Male BALB/c mice (6–8 weeks old) were obtained from the Laboratory Animal Center, Sun Yat-sen University (Guangzhou, China). Mice were housed in laminar airflow cabinets under specific pathogen-free conditions with 12-h light/12-h dark schedule and fed autoclaved standard chow and water ad libitum. All animal studies were conducted in strict compliance with the Guiding Principles for the Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of Sun Yat-sen University. The ethical approval number was No.44008500026874.

Cell line

A non-tumorigenic mouse hepatocyte cell line AML-12 cells, a murine macrophage cell line Raw-264.7 cells, a human hepatic stellate cell line LX2 cells, and a mouse fibroblast cell line L929 cells were sourced from the American Type Culture Collection (ATCC, Manassas, VA, USA). AML-12 cells were cultured in DMEM/F12 medium containing 0.005 mg/mL insulin, 0.005 mg/mL transferrin, 5 ng/mL selenium, 40 ng/mL dexamethasone, 10% FBS, and 1% penicillin/streptomycin in a 37°C humidified incubator (Thermo, USA) with 5% CO₂ and 95% air.

Primary cells

Mouse primary bone marrow derived macrophages (BMDMs) were isolated from bone marrow collected from BALB/c mouse femur and tibia. BMDMs were plated in a 10 cm tissue culture dish and cultured for one week in DMEM medium supplemented with 30% L929 cell conditioned medium, 10% FBS and 1% penicillin/streptomycin. Kupffer cells were isolated essentially as described previously.⁵⁴

Method details

Preparation and characterization of AI-Cells

The empty liposomal formulations (SA Lipos and PS Lipos) were constructed using a lipid film hydration method. SA Lipos was composed of lecithin/stearylamine (SA)/cholesterol (68.5/8.5/23, mass ratio) and PS Lipos was composed of lecithin/phosphatidylserine (PS)/cholesterol (52.5/22.5/25, mass ratio). The lipid preparation was dissolved in chloroform at 60°C, and then the chloroform was removed by rotary evaporation under reduced pressure to obtain the lipid dry film. The prepared lipid film was further hydrated in phosphate buffered saline (PBS, pH 7.4, 50 mM) for 1 h with stirring. The empty PS Lipos were subsequently sonicated using a BILON-650E probe sonicator (Shanghai Bilon Instrument Co., Ltd., Shanghai, China) before passing through 0.8, 0.45, and 0.22 μm polycarbonate filters. The empty SA Lipos were passed through a 0.8 μm polycarbonate filter. Itaconic acid was loaded into SA Lipos (SA Lipos-ITA) and PS Lipos (PS Lipos-ITA) using the pH gradient method. Briefly, the prepared SA Lipos or PS Lipos were dialyzed in double-distilled water (ddH₂O) for 1.5 h to obtain gradient blank liposomes, and then which were incubated with glycine-HCl buffered solution (pH 3.3) of itaconic acid (lipid/drug ratio 10:1, mass ratio) at 50°C for 5 min, and placed in an ice water bath for 2 min to terminate drug loading. Afterward, AI-Cells were prepared by mixing PS Lipos-ITA and SA Lipos-ITA in a ratio of 2:1. The preparation of A-Cells was the same except that SA Lipos and PS Lipos were used.

The particle size and zeta potential of the liposomes were determined by a Malvern Zetasizer Nano ZS instrument (Worcestershire, U.K.). The morphology of the A-Cells was examined by transmission electron microscopy (TEM) using a model JEM-1400 Flash microscope (JEOL, Tokyo, Japan). Fluorescence resonance energy transfer (FRET) was used to study the electrostatic adsorption of SA Lipos and PS Lipos. SA Lipos were stained with DiI and PS Lipos were stained with DiD. The fluorescent spectrum of samples ranged between 550 and 700 nm with an excitation wavelength at 525 nm. The fluorescent attenuation monitored the electrostatic adsorption of SA Lipos and PS Lipos. The encapsulation efficiency (EE%) of AI-Cells was quantified by ultrafiltration centrifugation and determined by LC-20AT HPLC (Shimadzu, Japan) at 205 nm. The EE% was calculated using the following equation:

$$EE\%=\frac{\mathrm{W}}{{\mathrm{W}}_0}$$

where W and W₀ are the amount of ITA in the AI-Cells after and before ultrafiltration centrifugation, respectively.

Selective macrophage targeting within hepatic sinusoidal capillaries

A transwell system with a pore size of 0.4 μm was used to simulate the sinusoidal endothelial fenestrae. The selective targeting of macrophages and hepatocytes by conventional liposomes and A-Cells in hepatic sinusoidal capillaries was studied. AML-12 cells were plated in the lower compartments of 24-well transwell plates and Raw-264.7 cells were plated in the upper chambers (Corning, CA, USA). DiI-labeled conventional liposomes or A-Cells were added to the upper chambers and incubated for 1 h before the nuclei were stained with Hoechst 33342. Qualitative and quantitative analysis of macrophage and hepatocyte uptake was performed using a digital inverted microscope (EVOS, Fisher Scientific, USA) and a flow cytometry (CytoFLEX S, Beckman Coulter, USA).

Cellular internalization test

Cellular internalization of A-Cells-DiI

In order to investigate the differences in macrophage uptake between conventional liposomes and A-Cells, conventional liposomes and A-Cells were labeled with DiI fluorescent dye. Raw-264.7 cells were cultured in 12-well plates and incubated with conventional liposomes labeled with DiI (Lipos-DiI) or A-Cells labeled with DiI (A-Cells-DiI) for different times (from 5 min to 4 h) at 37°C to study the uptake kinetics of liposomes. After incubation, cells were harvested to analyze the intracellular fluorescent intensity by flow cytometry.

Cellular internalization of A-Cells-RhB

In order to study the differences in cellular internalization between the water-soluble drug ITA and AI-Cells, the water-soluble fluorescent dye Rhodanine B (RhB) was selected to simulate ITA, and the RhB-loaded A-Cells (A-Cells-RhB) were prepared. Raw-264.7 cells were cultured in 12-well plates and incubated with different concentrations of RhB solution or A-Cells-RhB for 4 h at 37°C. Cells were harvested and resuspended in PBS for flow cytometry.

Cytotoxicity assay

Raw-264.7 cells (1 × 10⁴ per well) were plated into 96-well plates (Costar, USA) and incubated for 12 h. The cells were then exposed to varying concentrations of Lipos or A-Cells for a further 24 h. After incubation, 10 μL of 3-(4,5-dimethylthiazol-2-yl)-2,5 diphenyltetrazolium bromide (MTT) solution (5 mg/mL) was added to the cell cultures and treated for a further 4 h at 37°C. Finally, the medium was replaced with 100 μL DMSO and the absorbance at 490 nm was measured in a microplate reader (Bio-Tek, USA).

APAP-induced acute liver failure (ALF) mouse model

The acetaminophen (APAP)-induced ALF mouse model was established for male BALB/c mice. After 16 h fasting, fresh APAP solution (300 mg/kg in PBS) was administrated to the mice by intraperitoneal injection. In 3 h of APAP injection, the preparation was injected into the mice by tail vein. The mice were fasted without water during the experiment, and the mice were euthanized 24 h after APAP treatment. In parallel, the sham group was intraperitoneally injected with saline.

In vivo imaging and immunofluorescence analysis

To check the biodistribution of the injected liposomes, DiD-labeled Lipos (Lipos-DiD) or DiD-labeled A-Cells (A-Cells-DiD) were injected into APAP-induced ALF mouse model by intravenous injection. After 24 h of APAP injection, the in vivo biodistribution of DiD was visualized using the in vivo imaging system (IVIS) Spectrum (MA, USA). To examine the ex vivo tissue distribution of DiD, the main organs were dissected from the mice, imaged using the IVIS Spectrum and quantified by the Living Image software. Liver tissues were snap frozen with liquid nitrogen and incubated with Cy3 anti-mouse F4/80 primary antibody overnight at 4°C. Liver sections were stained with DAPI dye for nuclear localization, and imaged using Olympus fluoview3000 software (FV31S-SW).

In vivo safety evaluation

To evaluate the in vivo biocompatibility of Lipos and A-Cells, Lipos or A-Cells were injected into the mice by tail vein. After 24 h of APAP injection, blood samples were taken from the mice and the serum levels of ALT and AST were measured using an automated biochemical analyzer to assess hepatotoxicity. The kidneys were fixed in 4% paraformaldehyde for paraffin sectioning. Then, 5 μm thick sections were prepared and followed by hematoxylin-eosin (H&E) staining to evaluate nephrotoxicity.

In vivo therapeutic efficacy

The APAP-induced ALF mice were randomized into five groups (5 mice per group). The treatment was initiated at 3 h post APAP intraperitoneal injection. The mice received injection of saline, NAC, A-Cells, ITA or AI-Cells. The other groups were injected intravenously except for the NAC group. The dose of NAC was 300 mg/kg and the dose of ITA was 30 mg/kg. In addition, the sham group received injection of saline as the negative control group. The mouse survival rate was assessed for a period of 24 h. APAP-induced liver injure symptoms were quantitated by monitoring body temperature and body weight. Mice were humanely euthanized by exposure to CO₂ at 24 h post APAP injection. Then, the major organs were harvested and weighed to calculate organ index. To assess the injurious effects of APAP on liver function, serum samples were collected from mice. ALT, AST, ALP, and TBIL levels were measured in blood specimens using an automated biochemical analyzer.

Histology and immunohistochemistry analysis

The liver segments were fixed in 4% paraformaldehyde and then stained with hematoxylin-eosin (H&E). Histological changes were quantified by ImageJ 6.0 software (Media Cybernetics, Siler Spring, MD, USA). The percentage of necrosis area was calculated according to the following formula: (areas of necrosis tissue/total tissue area) × 100%. Ten distinct sections were used for calculation.

Paraffin-embedded liver sections were treated with citrate buffer for antigen retrieval and inactivated in 0.3% hydrogen peroxide solution to quench endogenous peroxidase activity. Afterward, liver sections were blocked in diluted normal goat serum, followed by anti-Ki67 primary antibody or anti-IL-β primary antibody incubation. Then, liver sections were incubated with the Vector Elite ABC kit (Vector Laboratories, USA) according to the instructions, visualized with the stable DBA reagent (Invitro, USA) and counterstained with hematoxylin. After staining, ten random fields per slide were observed in a blinded fashion. The average proportion of positive cells was calculated using ImageJ 6.0 software.

Direct effect on cell viability

BMDMs at 1 × 10⁴ per well, LX2 cells at 5 × 10³ per well or AML-12 cells at 5 × 10³ per well were plated into 96-well plates. When reaching to 70%–80% confluence, the cells were subjected to APAP solution (5 mM) for another 24 h, except for BMDMs (8 h). Then the APAP supernatant was removed, and the cells were treated with PBS, A-Cells, ITA or AI-Cells. After 24 h, the MTT assay was performed to detect cell viability. Free itaconic acid was prepared as a stock solution in PBS, and itaconic acid was eventually administered at a concentration of 2.0 μg/mL in the cellular experiments.

Quantitative real-time PCR

RNA of cells and liver tissue was extracted using a RNAiso Plus. RNA was reverse-transcribed by a PrimeScript RT Reagent Kit. Real-time PCR was performed using an IQ SYBR Green Supermix in a real-time PCR Detection System (Bio-Rad Laboratories). GAPDH was chosen as a housekeeping gene. The ΔΔCt method was used to normalized the data. Primers were shown as follows (Table S2).

Western blot analysis

Liver tissues were lysed in RIPA buffer containing protease inhibitor cocktail on ice for 10 min. The concentration of protein was measured by a BCA Protein Assay kit. Protein samples were separated by 10% SDS-PAGE gels and then transferred onto polyvinylidene difluoride (PVDF) membrane. Membranes were blocked in 5% (W/V) nonfat milk power in TBST buffer for 1 h at room temperature. Membranes were incubated with anti-IL-6 antibody, anti-IL-1β antibody, anti-TNF-α antibody, anti-IL-10 antibody, anti-Arg-1 antibody, anti-iNOS antibody, anti-NLRP3 antibody, anti-pro-Casp1/anti-Casp1 p20 antibody or anti-β-actin antibody overnight at 4°C, followed by the appropriate HRP-conjugated secondary antibody. Protein bands were detected by ECL substrate, visualized using the Tanon 4600 Imaging System (Tanon, China), and quantified using ImageJ 6.0 software.

In vivo hepatic macrophages phenotype conversion detection

For FACS analysis of liver tissue, the cell suspension was collected, passed through a 40 μm filter and centrifuged at 1500 g for 5 min. The cells were added with red blood cell lysis buffer and lysed for 3 min. Then the cells were stained with the following antibodies: FITC anti-mouse F4/80 antibody, PE anti-mouse CD86 antibody, and APC anti-mouse CD206 antibody. The marker expression on hepatic macrophages was detected using a flow cytometry and analyzed by FlowJo V10 software (Tritar Inc. San Carlos, California, USA).

Conditioned medium (CM)

BMDMs, Kupffer cells or Raw-264.7 cells (Macrophages^ΔASC) were plated in 6-well cell culture plates and allowed to adhere overnight. Then BMDMs or Kupffer cells were treated with LPS (100 ng/mL) for 3 h. Medium was removed, replaced with serum-free medium, and treated with PBS, A-Cells, ITA or AI-Cells for 45 min as required. Thereafter, BMDMs or Kupffer cells were processed with ATP (5 mM) for 45 min to activate the NLRP3 inflammasome. Macrophages^ΔASC were stimulated with LPS (1 μg/mL) for 12 h, and then medium was discarded. Macrophages^ΔASC were treated with PBS, A-Cells, ITA or AI-Cells for 24 h. After the incubation, the cell supernatant was collected and centrifuged at 1500 g for 5 min to obtain macrophage-derived conditioned medium (CM). Conditioned medium of each group was named B-CM, B-LPS&ATP CM, B-A-Cells CM, B-ITA CM, B-AI-Cells CM (BMDMs-derived CM), K-CM, K-LPS&ATP CM, K-AI-Cells CM (Kupffer cells-derived CM), M^ΔASC-CM, M^ΔASC-LPS CM, M^ΔASC-A-Cells CM, M^ΔASC-ITA CM, and M^ΔASC-AI-Cells CM (Macrophages^ΔASC-derived CM).

Indirect effect on cell viability

To investigate the effect of macrophage-derived conditioned medium on the survival rate of hepatocytes, B-LPS&ATP CM, B-A-Cells CM, B-ITA CM, B-AI-Cells CM, K-LPS&ATP CM, K-AI-Cells CM, M^ΔASC-LPS CM, M^ΔASC-A-Cells CM, M^ΔASC-ITA CM or M^ΔASC-AI-Cells CM was added to APAP-injured AML-12 cells and incubated for 24 h. The MTT method and the crystal violet staining method were used to determine the survival rate of AML-12 cells. In order to investigate which of TNF-α, IL-1β, and IL-6 secreted by macrophages was the main factor affecting the survival rate of hepatocytes, recombinant proteins of pro-inflammatory cytokines were used. The experimental method referred to the previous report.⁵⁵ As controls, recombinant mouse IL-1β at concentrations of 1 ng/mL was added to the B-AI-Cells CM. Twenty-four hours after exposure the cells were harvested for cell viability assay. IL-1β antibody was added to the K-LPS&ATP CM, and the survival rate of hepatocytes was assayed after 24 h of conditioned medium exposure.

Cell apoptosis assay

AML-12 cells (1 × 10⁵ per well) were plated in 12-well plates overnight, treated with APAP (5 mM) for 24 h, and incubated with B-LPS&ATP CM, B-A-Cells CM, B-ITA CM, B-AI-Cells CM, M^ΔASC-LPS CM, M^ΔASC-A-Cells CM, M^ΔASC-ITA CM or M^ΔASC-AI-Cells CM for another 18 h. After incubation, AML-12 cells were stained with Annexin-V-FITC and propidium iodide (PI) based on manufacturer’s recommendations. The cells were detected and analyzed with a flow cytometry and FlowJo V10 software.

Enzyme-linked immunosorbent assay (ELISA)

The secretion levels of TNF-α, IL-1β, and IL-6 in B-CM, B-LPS&ATP CM, B-AI-Cells CM, K-CM, K-LPS&ATP CM, K-AI-Cells CM, M^ΔASC-CM, M^ΔASC-LPS CM, and M^ΔASC-AI-Cells CM were assayed using ELISA kits in accordance with the manufacturer’s instructions. The absorbance at 450 nm was measured with a microplate spectrophotometer (Epoch, Bio-Tek, USA). Concentrations were calculated using a 4-parameter fit curve.

Co-culture assay

APAP (5 mM) damaged AML-12 cells were plated in the lower compartments of 12-well transwell plates with 0.4 μm pore size (Corning, CA, USA). Macrophages^ΔASC were plated in the upper chambers after LPS (1 μg/mL) stimulation and PBS, A-Cells, ITA or AI-Cells treatment. After 24 h of incubation, the survival of AML-12 cells in the lower chambers was measured using a Calcein-AM staining kit according to manufacturer’s protocol. A digital inverted microscope was applied to capture the live cells.

In vitro training of anti-inflammatory memory-like macrophages

BMDMs were stimulated with LPS (100 ng/mL) for 3 h and then removed. Subsequently, BMDMs in the model and AI-Cells groups were treated with PBS or AI-Cells for 45 min, respectively, followed by 45 min of treatment with ATP (5 mM). BMDMs were washed 3 times with PBS and incubated with fresh medium. After 5 days of resting, BMDMs were restimulated with 100 ng/mL of LPS for 3 h and 5 mM of ATP for 45 min. The viability of BMDMs was determined by the crystal violet method. Production of IL-1β was measured in the supernatants using an ELISA kit.

In vivo training of anti-inflammatory memory-like macrophages

Male BALB/c mice were injected intraperitoneally with a lethal dose of APAP solution (350 mg/kg in PBS) after 16 h fasting as the first challenge. The AI-Cells group mice were intravenously injected with AI-Cells for training anti-inflammatory memory-like macrophages 3 h after APAP injection. As a control, the model group mice were intravenously injected with normal saline. The mice were then rested for 6 days, during which time they resumed eating and drinking. On day 7, mice received an equal dose of APAP for a second challenge. Similarly, the mice rested again for 6 days and received the same dose of APAP for a third challenge on day 14 to establish the APAP-induced liver reinjury mouse model. The body temperature of the mice was monitored before each challenge, 6 h after the challenge, and 24 h after the challenge. Meanwhile, the serum of the mice was collected before and after each challenge and the level of IL-1β was detected with an ELISA kit.

Quantification and statistical analysis

All data were presented as mean ± SD. The two-group analysis was performed using a two-tailed unpaired Student’s t test with or without Welch’s correction for heteroscedasticity. One-way (or two-way) ANOVA with a Tukey’s test or Dunnett’s test were performed for multiple group comparisons. Statistical differences with two-tailed probability values of p < 0.05 were considered significant. All data were analyzed using GraphPad Prism 9.0.

Published: August 3, 2023

Data and code availability

• •All data supporting the findings of this study are available within the paper and its supplemental information files.
• •There are no codes generated in this paper.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.