A total of 65 study samples ( Supplementary Table 1 ) including 34 ACLF (paired sample, n = 18; blood only, n = 8; perfusate only, n = 8) and 14 decompensated liver cirrhosis (DLC) (paired sample, n = 14) during liver transplantation were recruited between 2021 and 2022 in Renji Hospital, School of Medicine, Shanghai Jiao Tong University. Patients with ACLF were diagnosed in accordance with Asian Pacific Association for the Study of the Liver (APASL) ACLF Research Consortium in 2019 (5): patients with compensated cirrhosis (diagnosed or undiagnosed) and those with non-cirrhotic chronic liver disease, who have a first episode of acute liver deterioration due to an acute insult directed to the liver. The criteria were established by (1) jaundice (total bilirubin levels of ≧5 mg/dl); (2) coagulopathy (International Normalized Ratio (INR) of ≧1.5, or prothrombin activity of less than 40%), within 4 weeks by clinical ascites, hepatic encephalopathy, or both. Etiology of enrolled patients with ACLF included HBV infection (n = 25), alcoholic-related liver injury (n = 4), drug-induced liver injury (DILI, n = 1), and autoimmune liver disease (n = 4; primary biliary cholangitis, autoimmune hepatitis, and primary sclerosing cholangitis). The exclusion standards were (1) malignancies, such as HCC; (2) Budd–Chiari syndrome; (3) extra-hepatic portal venous obstruction. Most enrolled patients had accepted artificial liver support. Patients who were admitted with DLC was either biopsy-affirmed or measured by a combination of clinical symptoms, routine laboratory tests, endoscopic examination, type-B ultrasonic, and CT/MRI scan. The enrolled patients with DLC include HBV infection (n = 5) and autoimmune liver disease (n = 9). Seventeen healthy living donors in liver transplantation were used as controls. The demographic and clinical characteristics of the study subjects were described in Table 1 . The study was conducted and approved by the Ethics Committee of Renji Hospital, Shanghai Jiao Tong University School of Medicine. All enrolled subjects gave written informed consent under the guidance of the Declaration of Helsinki.

All patients and matched cohorts accepted usual laboratory assessments for liver diseases, which include clinical evaluations, complete blood count, liver function tests (AST, ALT, ALP, γ-GGT, Tbil, Dbil, and albumin), renal function (serum creatinine), and coagulation function (international normalized ratio). The tests were performed at the clinical laboratory in Renji hospital, Shanghai Jiao Tong University School of Medicine.

All liver samples were obtained during orthotropic liver transplantation. For immunohistochemistry (IHC), 10% formalin-fixed and paraffin-embedded liver tissues were cut into 4-μm sections. IHC staining was carried out as previously described (24). Briefly, after heat-mediated antigen retrieval with sodium citrate buffer (pH 6.0) for 20 min and 3% H₂O₂ incubation (Beyotime, Shanghai, China, P0100A) for 15 min, liver sections were blocked with 10% goat serum (Solarbio, Beijing, China, SL038) for 30 min at room temperature and then incubated with rabbit anti-human CD19 (Abcam, Cambridge, UK, ab134114; 1:250) or rabbit anti-CD138 (Abcam, Cambrige, UK, ab128936; 1:8,000) overnight at 4°C. After washing in 1× Phosphate Buffered Saline (PBS) (GENOM, Haining, China, GNM20012) for three times, the liver slides were incubated with a horseradish peroxidase–conjugated secondary antibody (Long Island, Shanghai, China, D-3004) at room temperature for 30min and detected by 3,3′-diaminobenzidine (MXB Biotechnologies, Fuzhou, China, MAX007) and imaged by lighted microscope. Liver infiltration of CD19⁺ cells were assessed by counting absolute numbers of CD19⁺ cells per high power 400× field (CD19⁺ number).

Confocal staining assay was performed as mentioned above. After antigen retrieval and 3% H₂O₂ incubation, liver sections were blocked with 10% donkey serum (Solarbio, Beijing, China, SL050) for 30 min at room temperature. Then, liver sections were incubated with rabbit anti-human CD19 (Abcam, Cambridge, UK, ab134114; 1:100) and mouse anti-human IgM (Abcam, Cambridge, UK, ab200541; 1:50) or mouse anti-human CD19 (Abcam, Cambridge, UK, ab270715; 1:100) and rabbit anti-human CD11c (Abcam, Cambridge, UK, ab52632; 1:100) overnight at 4°C. After washing in PBS, the sections were incubated with different fluorochrome-conjugated secondary antibodies (Invitrogen, Carlsbad, CA, USA, 1:500) for 30 min at room temperature. Confocal scanning was performed using an LSM-710 laser scanning confocal microscope (Carl Zeiss, Jena, Germany).

Intrahepatic lymphocytes (IHLs) were isolated as previously described (25). Briefly, once the organ was excised, liver sections were flushed with saline solutions to remove any residual blood. Small pieces of liver sections were perfused with isotonic fluid supplemented with EDTA (5 mM/L), and then, cells were resuspended in RPMI-1640 medium (Gibco). The suspension was filtered through a 70-μm mesh. After a 5-min centrifugation at 50g, the non-parenchymal hepatic cells in the supernatant were collected and harvested by another 10-min centrifugation at 500g. Subsequently, IHL were further lysed with the Red Blood Cell Lysis Buffer (Sigma) for 3 min. After gentle washing, separated IHLs were labeled with cell surface and intracellular cytokine markers for flow cytometric analysis.

We obtained peripheral blood mononuclear cells (PBMCs) by a gradient of Ficoll-Paque (GE Healthcare, USA). The freshly enriched PMBCs and IHLs were washed with PBS twice. For intracellular cytokines detection, cells were stimulated in complete RPMI-1640 containing 10% Fetal Bovine Serum (FBS) and Lipopolysaccharide (LPS) (1μg/ml) (Sigma-Aldrich, St. Louis, CA, USA, L2630) plus Leukocytes Activation Cocktail with GolgiPlug (BD Biosciences, San Diego, CA, USA, 550583) in a 37°C humidified CO₂ incubator for 5 h. Then, cells were stained with live/dead surface markers, fixed with the Fix/Perm kit (BD Biosciences, San Diego, CA, USA, 554714), and incubated with antibodies against intracellular cytokines. The monoclonal antibodies (mAbs) include APC-Cy7–anti-CD19 (BioLegend, USA, Cat#302218); PE-Cy7–anti-CD21 (BioLegend, USA, Cat#354912), PerCP-Cy5-5–anti-CD27 (BD Biosciences, Cat#560612); APC–anti-CD38(BD Biosciences, Cat#555462); PE–anti-IgD (BioLegend, USA, Cat#348204); FITC–anti-IgM (BioLegend, USA, Cat#314506); PerCP-Cy5-5–anti-CCR6 (BD Biosciences, Cat#560467); BV421–anti-CD11c(BD Biosciences, Cat#562561); BV605–anti-CD273 (BD Biosciences, Cat#752600); BV650–anti-CD27 (BD Biosciences, Cat#563228); APC–anti-IgD (BD Biosciences, Cat#561303); BB515–anti-CXCR5 (BD Biosciences, Cat#564624); APC-R700–anti-CD69 (BD Biosciences, Cat# 565154); BV510–anti-CD38 (BD Biosciences, Cat#563251); BV711–anti-IgM (BD Biosciences, Cat#743327); BV786–anti-CD19 (BD Biosciences, Cat#563325); PE-CF594–anti-CXCR3 (BD Biosciences, Cat#562451); PE-Cy7–anti-CD21 (BD Biosciences, Cat#561374); FITC–anti-IL-6 (BD Biosciences, Cat#554544); BV421–anti–TGF-β (BD Biosciences, Cat#562962); BV605–anti-CD38 (BD Biosciences, Cat#562665); BV650–anti-CD80(BD Biosciences, Cat#743866); APC–anti–GM-CSF (BD Biosciences, Cat#502310); Alexa Fluor 700–anti-CD21 (BioLegend, USA, Cat#354918); PE–anti-IL-10 (BD Biosciences, Cat#555088); BV510–anti–TNF-α (BioLegend, USA, Cat#502950); and PE-CF594–anti–granzyme B (BD Biosciences, Cat#562462). All experiments were performed by flow cytometry (Celesta, BD Biosciences and LSRFortessa X-20, BD Biosciences), and data were analyzed with FlowJo software version 10.8 (Tree Star, Ashland, OR USA). Gating strategy was shown in Figure 1A .

Serum cytokines level were quantitated by a BD Human Inflammation CBA (cytometric bead array) kit. Briefly, the standards were prepared by serial dilution according to the manufacturer’s instructions (BD Biosciences, San Jose, CA, USA). The serum samples and standards were incubated with specific capture beads for 2 h at room temperature. The mixture was incubated for additional 1 h with the detection reagent. After a final wash, beads were analyzed on a BD FACSCanto II cytometer. Data analysis was performed using the FCAP Array Software (BD Biosciences, San Jose, CA, USA).

The statistical testing was performed with GraphPad Prism 9.0 (GraphPad Software Inc., USA). Plotted data were expressed as mean ± standard error of mean (SEM). Mann–Whitney U-test and paired t-test were conducted for continuous variables. Correlations were determined by Pearson or Spearman correlation analysis. One-way analysis of variance (ANOVA) was used for multiple comparisons. All tests were two-tailed, and P < 0.05 was considered as statistically significant.

First, we sought to examine whether there was a change in CD19⁺ total peripheral B cells and no difference was observed between ACLF, cirrhosis, and healthy control (HC) cohorts ( Figure 1B ). In ACLF, we found a drastic reduction in naïve B cells compared with cirrhosis and HC, indicating that patients with ACLF had more antigen-experienced B cells in blood (p < 0.05 for ACLF vs. cirrhosis, p < 0.001ACLF vs. HC) ( Figure 1C ). No discrepancy was detected in RM between ACLF and cirrhosis (p < 0.05 for ACLF vs. HC and for cirrhosis vs. HC) ( Figure 1D ). Flow Cytometry (FCM) analysis showed an enrichment of AM in ACLF, whereas cirrhosis and HC showed no difference ( Figure 1E ). atMBC, also known as tissue-like memory B cells or aged B cells, were notably expanded in comparison with the remaining two groups (P < 0.001 for ACLF vs. cirrhosis, P < 0.0001 for ACLF vs. HC, P < 0.01 for cirrhosis vs. HC) ( Figure 1F ). We further took advantage of IgM to characterize CD27⁺IgD⁻ IgM⁺ memory B cells (IgM⁺ MBC). Unexpectedly, IgM⁺ memory B-cell subset, accounting for an average of 15% of total B cells, displayed a marked increase compared with liver cirrhosis and HC ( Figure 1G ). We also noticed a marked elevation of CD27⁺CD38⁺⁺ plasma cells in patients with ACLF (p < 0.05 for ACLF vs. cirrhosis and for ACLF vs. HC) ( Figure 1H ). In summary, we managed to provide evidence for altered peripheral B-cell subsets including naïve B cells, AM, atMBC, IgM⁺ MBC, and CD27⁺CD38⁺⁺ plasma cells in patients with ACLF.

To determine the distinct profile of intrahepatic B cells in ACLF, we next analyzed B-cell subsets in ACLF liver in comparison with cirrhosis and HC. Aside from concurrent abundant lymphocyte liver infiltrate, IHC staining demonstrated an absolute increase in CD19⁺ B-cell number, suggesting accumulated intrahepatic B cells in ACLF ( Figure 2A ). However, cytometry analysis showed no difference in the percentage of infiltrating CD19⁺ B cells within live lymphocytes between ACLF and cirrhosis ( Figure 2B ). In the liver of patients with ACLF, the fraction of naïve B cells was the lowest ( Figures 2C, D ). Hepatic AM was similarly presented in cirrhosis and HC, proposing that this subset might be associated with aberrant liver dysfunction ( Figure 2E ). Liver-infiltrating atMBC appeared highest in ACLF group, followed by liver cirrhosis and healthy donors ( Figure 2F ). Within the intrahepatic IgM⁺ MBC population, we detected that ACLF liver harbored an elevated percentage of IgM⁺MBC compared with cirrhotic patients and HCs ( Figure 2G ). By performing double immunohistochemical staining, we verified the presence of IgM⁺CD19⁺B cells in ACLF liver ( Figure 2H ). In addition, IHC and FACS analysis showed that ACLF liver was specifically rich in CD27⁺CD38⁺⁺plasma B cells ( Figures 2I, J ), raising the possibility that ACLF liver dysfunction might be ascribed to an increase in intrahepatic CD27⁺CD38⁺⁺plasma B cells. In sum, we could conclude that patients with ACLF displayed unique intrahepatic B-cell subsets that were, to some extent, divergent from a cirrhotic setting and physical homeostasis.

Since we have successfully discerned the peripheral B-cell milieus in ACLF from cirrhosis and HC, we set out to investigate whether intrahepatic B-cell populations differed from their peripheral counterparts. In HC liver, we found that the percentage of naïve B cells was reduced compared to peripheral counterparts, whereas atMBC, IgM⁺MBC, and CD27⁺CD38⁺⁺plasma B cells were increased despite no difference in AM ratio ( Figures 3A–C ). In patients diagnosed with cirrhosis, we examined increased fractions of AM, atMBC, and IgM⁺ MBC in cirrhotic liver perfusate ( Figures 3A, B ). However, a similar distribution of CD27⁺CD38⁺⁺plasma B cells between blood and liver was observed ( Figures 3A–C ). To notify, we noticed altered proportions of intrahepatic B cells that can be illustrated by a lower fraction of naïve B cells and a higher fraction of AM and IgM⁺ MBC in ACLF liver ( Figures 3A–C ). Remarkably, we indicated increased percentage of hepatic atMBC and CD27⁺CD38⁺⁺plasma cells in ACLF, which was in sharp contrast to liver cirrhosis and HCs. Thus, we indicated that the discrepancy in intrahepatic and blood B cells in cirrhosis was less diverse compared to ACLF. We verified that intrahepatic ACLF B-cell populations were distinct from peripheral counterparts.

Since we have observed difference in intrahepatic B-cell subsets composition, we next investigated the functional change in these subsets. In ACLF, intrahepatic AM expressed lower CD273 (PD-L2) molecule, indicating a decreased interaction of B-T cells and probably denoting an exhausted status. However, no difference was found in the hepatic expression of CCR6, CD11c, CD69, CRCR3, CXCR5, and CD80 in AM of ACLF ( Figures 4B, C ). Among the atMBC subsets, patients with ACLF expressed more CD11c molecules and lower levels of CD80, showing aggravated cell exhaustion and enhanced local chemotaxis ( Figures 4D, E ). We also observed the spatial distribution of CD11c⁺CD19⁺ B cells ( Figure 4A ). IgM⁺ MBC that targeted HBsAg in HBV-ACLF liver was mainly CD27⁻CD21⁻ subset. The reduced percentage of CCR6⁺IgM⁺ MBC in ACLF indicated a diminished tendency of differentiation into antibody-secreting cells ( Figures 4F, G ). Notably, ACLF plasma cells expressed higher level of CD273 (PD-L2) and liver-homing chemokine CXCR3, whereas CXCR5, a lymph node–homing indicator, was similarly expressed between disease and controls ( Figure 4H ). Aside from chemokine receptors, reduced TNF-α⁺ plasma cells indicated that this group of plasma cells alleviated inflammation during ACLF disease progression ( Figure 4I ).We also demonstrated that intrahepatic plasma cells secreted more IL-10 and granzyme B, suggesting the existence of a subset of B cells that could exert regulatory function in ACLF ( Figure 4I ). Moreover, we found heightened level of serum IL-10 in ACLF compared with cirrhosis and HCs ( Figure 4J ). In summary, data from flow analysis indicated that intrahepatic AM, atMBC, IgM⁺ MBC, and plasma cells exhibited altered functions in ACLF.

To investigate whether these intrahepatic B-cell subsets were reflective of disease severity, we went on to examine the association between those cells with several clinical indices in ACLF. We did not find any association between AM and IgM⁺ MBC with clinical parameters. The percentage of naïve B cells was in an inverse relationship with alanine transaminase (ALT; r = −0.5945, p = 0.0014) and aspartate transaminase (AST; r = −0.6379, p = 0.0005) and was positively correlated with albumin (ALB; r = 0.3828, p = 0.0212) but not with alkaline phosphatase (AKP) or gamma-glutamyl transferase (γ-GGT) ( Figure 5A ). atMBC subset was significantly correlated to ALT (r = 0.5312, p = 0.0052) and AST (r = 0.6525, p = 0.0003) ( Figure 5B ), whereas no correlations were found between atMBC and other clinical indices. Moreover, we determined a significant correlation between CD27⁺CD38⁺⁺ plasma B cells with clinical parameters of liver cholestasis, measured by AKP (r = 0.5526, p = 0.0034) and γ-GGT (r = 0.4556, p = 0.0193) ( Figure 5C ). However, we failed to associate intrahepatic CD27⁺CD38⁺⁺ plasma B cells with ALT, AST, and ALB. Altogether, these findings demonstrated that intrahepatic B-cell lineages could be indicative of liver injury in ACLF.