Eosinophil accumulation in the liver persists during liver repair following hepatic IR injury

Our previous study demonstrated that eosinophils were recruited to the liver within the first 24 hours after hepatic IR injury in mice and exerted a profound protective effect.31 Surprisingly, when we examined the kinetics of eosinophil accumulation, we observed that these cells persisted in the liver beyond the injury phase. Eosinophils were identified as CD45⁺CD11b⁺Ly6G⁻SSC^highSiglec-F⁺CCR3⁺ cells by flow cytometry. Additionally, immunohistochemical staining was conducted to detect eosinophils using an antibody recognising major basic protein (a generous gift from Dr Elizabeth Jacobsen, Mayo Clinic, Arizona, USA).32 33 Compared with sham mice, those subjected to IR injury had elevated accumulation of eosinophils in the liver up to 5 days after IR injury, with a peak on day 3 (figure 1A–E), coinciding with the peak of hepatocyte proliferation in wild-type C57Bl/6J (WT) mice (figure 1F,G). These data suggested that eosinophils might contribute to liver repair and regeneration after IR injury.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

The kinetics of hepatic eosinophil accumulation and the time course of liver repair after IR injury in WT mice. Male C57Bl/6J mice were subjected to hepatic ischemia (60 mins) followed by reperfusion and sacrificed at various times (n=3/group). Sham mice underwent the same surgery without vascular blockage and were sacrificed on day 7. (A-C) The percentage and numbers of eosinophils (CD45⁺CD11b⁺Ly6G^-Siglec-F⁺CCR3⁺cells) in the liver were measured by flow cytometry and quantified. (D, E) IHC staining for eosinophils by anti-mouse major basic protein (MBP) antibody and the numbers of MBP⁺ cells quantified. (F, G) Proliferating hepatocytes were stained by anti-Ki67 antibody, and the numbers of positive cells were quantified. (H, I) Liver necrosis (N, outlined areas) was evaluated and quantified. (J) Liver pathology was assessed by using the Suzuki’s scoring system. Two-tailed unpaired Student’s t-test with Welch’s correction was performed in B, C, E, G, I and J.

Eosinophils play an indispensable role in tissue repair after hepatic IR injury

By day 7 after IR injury, the liver nearly completely recovered in WT mice, based on the quantification of necrotic areas and the Suzuki scores (figure 1H–J). To determine the role of eosinophils in liver repair, we used two strains of eosinophil-deficient mice (ΔdblGATA1 and PHIL mice). Compared with WT mice, extensive areas of necrosis (>30% of the liver) remained in both strains on day 7 after IR injury (online supplemental figure S1B–D, F–H). However, one caveat in interpreting the data is that eosinophil-deficient mice had much more severe liver IR injury (online supplemental figure S1A,E), which could contribute to delayed tissue repair. To circumvent this issue, we used inducible eosinophil-deficient (iPHIL) mice, which were generated by knocking-in the human diphtheria toxin (DT) receptor gene into the endogenous eosinophil peroxidase genomic locus.34 We first established a DT dosing regimen that can achieve eosinophil depletion. The data demonstrated that after two consecutive injections of DT on day 1 and day 2, the number of eosinophils in the blood reduced markedly on day 3 and maintained at low levels of abundance until at least day 8 (online supplemental figure S2A). Based on these results, we administered the first dose of DT to iPHIL mice and WT littermates 16 hours prior to hepatic ischaemia surgery and injected the second dose 6 hours after surgery (figure 2A). This protocol ensured comparable hepatic IR injury between DT-treated WT littermates and iPHIL mice (online supplemental figure S2B) and achieved successful depletion of eosinophils in iPHIL mice after injury and during the time of repair (online supplemental figure S2C,D). We observed a significant delay in tissue repair in iPHIL mice compared with WT littermates, evidenced by larger areas of necrosis and higher Suzuki scores on day 7 after IR surgery (figure 2B–D). These changes coincided with the dramatic reduction of hepatocyte proliferation markers (Ki67 and proliferating cell nuclear antigen (PCNA)) on day 3 (figure 2E–H).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Eosinophils promote liver repair after IR injury. (A-H) Male iPHIL mice and WT littermates were subjected to hepatic IR surgery. Mice were administered (i.p.) the first dose of DT at 16h prior to IR surgery and the second dose at 6h after surgery. Mice were sacrificed on day 3 (n=4/group) or day 7 (n=6/group) after IR surgery. (B, C) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7 after IR surgery. (D) Liver pathology was determined by the Suzuki score on day 7 after IR surgery. (E, F) Liver tissue sections were stained for Ki67 to quantify proliferating hepatocytes on day 3 after IR surgery. (G, H) PCNA protein expression was detected by Western blotting and quantified. (I-P) Male ΔdblGATA-1 mice were subjected to hepatic IR surgery. After 24h, half of the mice were i.v. injected with bmEos (10x10⁶) and the other half injected with saline as control. Mice were sacrificed on day 3 (n=4/group) or day 7 after IR surgery (n=6/group). (J, K) Liver necrosis (N, outlined area) was evaluated and quantified on day 7 after IR surgery. (L) Liver pathology was determined by the Suzuki score on day 7 after IR surgery. (M, N) Proliferative hepatocytes were stained for Ki67 and quantified on day 3 after IR surgery. (O, P) PCNA protein expression was detected by Western blotting and quantified. Two-tailed unpaired Student’s t-test with Welch’s correction was performed in C, D, F, H, K, L, N and P.

To further investigate the role of eosinophils in liver repair, we adoptively transferred bmEos to eosinophil-deficient mice after IR injury (figure 2I). To evaluate the feasibility, we prepared bmEos from B6-CD45.1 mice and intravenously injected to C57Bl/6J (B6-CD45.2) recipients and tracked the cells by flow cytometry. The data showed rapid infiltration of CD45.1⁺bmEos into the recipient’s liver within 30 min, reaching its peak at 6 hours postinjection and persisting for at least 3 days postinjection (online supplemental figure S3A,B). After validating the protocol, we adoptively transferred bmEos from WT mice (WT-bmEos) to ΔdblGATA1 and PHIL mice on day 1 after IR surgery. We used this time point to avoid the effects of WT-bmEos on the extent of injury. As shown in figure 2J–L and online supplemental figure S4A–C, while large areas of necrosis remained after IR surgery in control eosinophil-deficient mice injected with saline, liver histology returned to nearly normal in those that received WT-bmEos. Moreover, hepatocyte proliferation was restored in ΔdblGATA1 mice injected with WT-bmEos (figure 2M–P).

To further validate the pro-repair function of eosinophils, we extended the study to two additional models of acute liver injury caused by acetaminophen (APAP) or carbon tetrachloride (CCl₄). We treated iPHIL mice and their WT littermates with DT 16 hours before and 8 hours after APAP or CCl₄ challenge. As shown in online supplemental figure S5, the WT and iPHIL mice developed similar degrees of liver injury. However, in both acute liver injury models, a significant delay in tissue recovery was observed in iPHIL mice compared with WT littermates. This was evidenced by the presence of larger areas of necrosis and higher Suzuki scores on day 6 after APAP or CCl₄ treatment. Together, these data provide strong evidence that eosinophils play an essential role in the liver repair process after acute injury.

To understand how eosinophils promote liver repair after IR injury, we performed a cytokine array analysis (Proteome Profiler Mouse Cytokine Array Kit). The data demonstrated IL-4 and IL-13 as the most significantly upregulated cytokines on adoptive transfer of WT-bmEos to ΔdblGATA1 mice (online supplemental figure S6). To further investigate the cellular source of IL-4 and IL-13, we performed flow cytometric analyses using IL-4/GFP-enhanced transcript (4Get) mice to measure IL-4 expression and intracellular staining to detect IL-13. The data showed that eosinophils accounted for >86% of IL-4-positive cells and >90% of IL-13-positive cells in the liver (online supplemental figure S7A,B), suggesting that IL-4 and IL-13 are mainly produced by eosinophils during tissue repair. To investigate the role of eosinophil-derived IL-4 and/or IL-13 in liver repair, we generated mice with eosinophil-specific deletion of both IL-4 and IL-13 (Il-4/13^fl/fleoCre^+/-). We obtained bmEos from these mice and their WT littermates (Il-4/13^fl/fleoCre^−/−) and adoptively transferred the cells to ΔdblGATA1 mice on day 1 after hepatic IR injury (figure 3A). While WT-bmEos restored liver repair in ΔdblGATA1 mice, IL-4/13-deleted bmEos failed to improve liver repair compared with control ΔdblGATA1 mice injected with saline (figure 3B–D). Congruent with this observation, the adoptive transfer of WT-bmEos, but not IL-4/13-deleted bmEos, significantly increased hepatocyte proliferation in ΔdblGATA1 mice on day 3 after IR injury (figure 3E,F). These data suggest that eosinophil-derived IL-4 and/or IL-13 are critically involved in liver repair and regeneration after IR injury.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Eosinophil-derived IL-4 plays a critical role in liver repair after IR injury. (A-F) Male ΔdblGATA-1 were subjected to hepatic IR surgery. After 24h, the mice were divided into three groups and i.v. injected with saline, WT-bmEos (10x10⁶) or Il-4/13-deleted bmEos (10x10⁶). Mice were sacrificed on day 3 (n=4/group) or day 7 (n=6/group) after IR surgery. (B, C) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7 after IR surgery. (D) Liver pathology was assessed by using the Suzuki’s scoring system on day 7 after IR surgery. (E, F) Proliferative hepatocytes were stained for Ki67 and quantified on day 3 after IR surgery. (A, G-K) Male ΔdblGATA-1 were subjected to hepatic IR surgery. After 24h, the mice were divided into three groups and i.p. injected with PBS, IL-4c (5 μg recombinant IL-4 complexed to 25 μg anti-IL-4 antibody) or IL-13c (5 µg of recombinant IL-13 complexed to 25 µg anti-IL-13 antibody). Mice were sacrificed on day 3 (n=4/group) or day 7 (n=6/group). (G, H) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7 after IR surgery. (I) Liver pathology was assessed by using the Suzuki’s scoring system quantified on day 7 after IR surgery. (J, K) Proliferative hepatocytes were stained for Ki67 and quantified on day 3. (A, L-P) Male C57Bl/6J mice were subjected to hepatic IR surgery. On day 1 and day 3 after surgery, the mice were i.p. injected with anti-IL-4 antibody (10 μg/mouse) or anti-IL-13 antibody (10 μg/mouse). Control mice were injected with IgG. Mice were sacrificed on day 3 (n=4/group) or day 7 (n=6/group) after IR surgery. (L, M) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7 after IR surgery. (N) Liver pathology was assessed by using the Suzuki’s scoring system on day 7 after IR surgery. (O, P) Proliferative hepatocytes were stained for Ki67 and quantified on day 3. (A, Q-S) Male ΔdblGATA-1 mice were subjected to hepatic IR surgery. After 24h, half of the mice were i.v. injected with WT-bmEos (10x10⁶) and the other half injected with Il-4-deleted bmEos (10x10⁶). Mice were sacrificed on day 7 after IR surgery (n=4/group). (Q, R) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7 after IR surgery. (S) Liver pathology was assessed by using the Suzuki’s scoring system on day 7 after IR surgery. Two-tailed unpaired Student’s t-test with Welch’s correction was performed in R and S. One-way ANOVA was performed in C, D, F, H, I, K, M, N and P.

Eosinophils promote liver tissue repair through IL-4 production

To further determine whether the delayed liver repair in ΔdblGATA1 mice was due to the lack of eosinophil-derived IL-4 or IL-13, we injected IL-4/anti-IL-4 mAb complex (IL-4c) or IL-13/anti-IL-13 mAb complex (IL-13c) to ΔdblGATA1 mice (figure 3A). We used the antibody-conjugated cytokines to prolong the half-life of the cytokines as reported in many studies.35–38 Our data showed that IL-4c injection could significantly increase hepatocyte proliferation (figure 3J,K) and accelerate liver repair (figure 3G–I). Moreover, blockade of IL-4 by a neutralising antibody significantly delayed liver repair in WT mice, as demonstrated by decreased numbers of Ki67⁺ hepatocytes on day 3 after IR injury as well as increased Suzuki scores and extents of live necrosis on day 7 (figure 3L–P). In contrast, we did not observe any improvement of liver repair in ΔdblGATA1 mice treated with IL-13c, nor any inhibitory effect by IL-13 blockade on liver repair in WT mice (figure 3G–P). To further corroborate these results, we adoptively transferred bmEos obtained from IL-4^−/− mice to ΔdblGATA-1 mice on day 1 after IR surgery (figure 3A). Our data showed that IL-4-deleted bmEos, although expressing IL-13, could not restore liver repair in ΔdblGATA1 mice (figure 3Q–S). Taken together, these data strongly suggest that eosinophil-derived IL-4, but not IL-13, plays a critical role in promoting liver regeneration and repair after IR injury.

IL-4Rα signalling in hepatic macrophage orchestrates liver repair after IR injury

The direct effect of IL-4 on hepatocytes has been investigated in the context of liver regeneration after non-injurious partial hepatectomy, but the results are controversial. One study suggested that IL-4 could stimulate hepatocyte proliferation39; however, another study showed that IL-4 did not directly affect hepatocyte proliferation.40 To investigate if hepatocytes are a direct target of IL-4, we generated a mouse line with hepatocyte-specific IL-4rα-deletion (Il-4rα^fl/flAlb-Cre^+/−). After IR surgery, these mice showed similar degrees of liver injury with no effect on tissue recovery (figure 4A–F). These data ruled out a direct effect of IL-4 on hepatocytes during liver regeneration and repair after IR injury.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

IL-4Rα signaling in liver macrophages, rather than hepatocytes, promotes liver repair after IR injury. (A-F) Male Il-4rα^fl/flAlb-Cre^+/- mice and WT littermates (Male Il-4rα^fl/flAlb-Cre^-/-) were subjected to hepatic IR surgery and sacrificed after 3 days (n=4/group) or 7 days (n=6/group). (G-L) Male Il-4rα ^fl/flClec4f-Cre^+/- mice and WT littermates (Il-4rα^fl/flClec4f-Cre^-/-) were subjected to hepatic IR surgery. After 24h, the Il-4rα^fl/flClec4f-Cre^+/- mice were divided into two groups and injected (i.v.) with PBS or IL-4c (n=6/group) and and sacrificed after 3 days (n=4/group) or 7 days (n=6/group) after IR surgery. (A, G) Serum ALT levels 6h after IR surgery. (B, C, H, I) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7 after IR surgery. (D, J) Liver pathology was assessed by using the Suzuki’s scoring system on day 7 after IR surgery. (E, F, K, L) Proliferative hepatocytes were stained for Ki67 and quantified on day 3 after IR surgery. Two-tailed unpaired Student’s t-test with Welch’s correction was performed in A, C, D, F, G and L. One-way ANOVA was performed in I and J.

To search for IL-4 target cells, we screened for IL-4Rα expression in NPCs. Our data showed that >70% of IL-4Rα-positive cells were macrophages (F4/80⁺), and the rest were liver sinusoidal endothelial cells (CD31⁺) (online supplemental figure S8). To further address whether IL-4 promotes liver repair through IL-4Rα signalling in macrophages, we generated macrophage-specific Il-4rα-deleted mice using Clec4f-Cre mice, which have been used to successfully generate liver macrophage-specific gene knockout mice.41–43 We subjected Il-4rα^fl/flClec4f-Cre^+/− mice and WT littermates (Il-4rα^fl/flClec4f-Cre^−/−) to hepatic IR surgery and observed similar degrees of liver injury (figure 4G). However, liver regeneration was significantly delayed in the Il-4rα^fl/flClec4f-Cre^+/− mice, evidenced by larger areas of necrosis, higher Suzuki scores and lower numbers of Ki67⁺ hepatocytes than WT littermates (figure 4H–L). Interestingly, treatment of Il-4rα^fl/flClec4f-Cre^+/− mice with IL-4c did not improve liver repair (figure 4H–J), despite the presence of intact IL-4Rα on other cells. Taken together, these data demonstrate that the pro-repair effects of IL-4 after IR injury are mediated through IL-4Rα signalling in hepatic macrophages.

IL-4-induced HB-EGF production by hepatic macrophages plays an essential role in liver repair after IR injury

To further understand how IL-4/IL-4Rα signalling in hepatic macrophages contributes to tissue repair and regeneration after IR injury, we performed a Reverse Phase Protein Array (RPPA) experiment using liver samples harvested on day 3 after IR injury. The data revealed a reduced level of phosphorylated(p)-epidermal growth factor receptor (p-EGFR) in the liver of Il-4rα^fl/flClec4f-Cre^+/− mice compared with WT littermates (table 1). This result was confirmed by western blot analyses (figure 5A,B), and suggested that the IL-4/IL-4Rα signalling in hepatic macrophages could regulate one or more EGFR ligands. Thus, we purified hepatic macrophages from Il-4rα^fl/flClec4f-Cre^+/− mice and WT littermates on day 3 after IR injury and measured mRNA levels of all seven known EGFR ligands, including Egf, transforming growth factor alpha, Hb-egf, amphiregulin, epigen, epiregulin and betacellulin. The data showed that only the expression of Hb-egf was reduced in hepatic macrophages from Il-4rα^fl/flClec4f-Cre^+/− mice (figure 5C). To determine if IL-4, through IL-4Rα, directly triggers macrophages to produce HB-EGF, we treated bone marrow-derived macrophages (BMDM) from WT and Il-4rα^−/− mice with IL-4. The data showed that IL-4 dramatically upregulated hb-egf mRNA and protein levels in WT-BMDM, but not in IL-4rα-deleted BMDM (figure 5D,E). Given our data showing that liver sinusoidal endothelial cell (LSEC) also express IL-4Rα (online supplemental figure S8), we isolated LSECs from naïve mice and treated them with IL-4. Interestingly, IL-4 did not induce hb-egf expression in LSECs (online supplemental figure S9), suggesting that IL-4/IL-4Rα-mediated signalling in LSECs does not significantly contribute to HB-EGF production. Taken together, these results suggest that (1) hepatic macrophages represent the predominant source of HB-EGF during liver IR injury and (2) the IL-4/IL-4Rα signalling is critical in HB-EGF production by macrophages.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

IL-4/IL-4Rα signaling in hepatic macrophage induces HB-EGF expression. (A-C) Male Il-4rα^fl/flClec4f-Cre^+/- mice and WT littermates (Male Il-4rα^fl/flClec4f-Cre^-/-) were subjected to hepatic IR surgery and sacrificed after 3 days. Liver tissues were collected and hepatic macrophages were purified by magnetic-associated cell sorting (MACS) using anti-F4/80 antibody. (A, B) phospho(p)-EGFR and total (t)-EGFR protein levels in the liver tissue were detected by Western blotting and quantified (n=4/group). (C) mRNA levels of all EGFR ligands expressed by hepatic macrophages were measured by q-PCR (n=3/group). (D, E) WT-BMDM and Il-4rα-deleted BMDM were cultured with IL-4 for 6h or 24h. Control groups were treated with PBS (n=3/group). mRNA and protein levels of HB-EGF were measured at 6h by q-PCR and 24h by ELISA, respectively. (F-H) rmHB-EGF was i.p. injected to male Il-4rα^fl/flClec4f-Cre^+/- mice on day 1 and day 3 after liver IR surgery. Control mice were injected with PBS. All mice were sacrificed on day 7 after IR surgery (n=4/group). (F, G) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7. (H) Liver pathology was assessed by using the Suzuki’s scoring system. Two-tailed unpaired Student’s t-test with Welch’s correction was performed in B, C, G and H. One-way ANOVA was performed in D and E.

Male Il-4rα^fl/flClec4f-Cre^+/− mice and WT littermates (male Il-4rα^fl/flClec4f-Cre^−/−) were subjected to hepatic IR surgery and sacrificed after 3 days. Liver tissue samples were processed and used for the RPPA screening.

To investigate the involvement of HB-EGF in tissue repair after IR injury, we injected recombinant mouse HB-EGF into Il-4rα^fl/flClec4f-Cre^+/− mice on days 1 and 3 after IR injury. As shown in figure 5F–H, exogenous HB-EGF restored liver repair by day 7 in these mice. To further determine whether HB-EGF produced by hepatic macrophages, rather than other cells, plays a critical role in liver repair, we generated a mouse with hepatic macrophage-specific deletion of hb-egf (Hb-egf^fl/flClec4f-Cre^+/−). Figure 6A shows that these mice developed similar degrees of liver injury compared with WT littermates after IR surgery. However, whereas WT mice recovered by day 7 after IR injury, Hb-egf^fl/flClec4f-Cre^+/− mice continued to show extensive areas of necrosis (>35% of the liver) with higher Suzuki scores (figure 6B–D). In line with this observation, hepatocyte proliferation, as quantified by Ki67 staining and hepatic PCNA expression, was also dramatically decreased in Hb-egf^fl/flClec4f-Cre^+/− mice on day 3 after IR injury (figure 6E–H). In addition, western blot analyses showed that the p-EGFR levels were significantly lower in Hb-egf^fl/flClec4f-Cre^+/− mice than WT littermates (figure 6I–J). Moreover, treatment of Hb-egf^fl/flClec4f-Cre^+/− mice with IL-4c could not restore liver repair after IR injury (figure 6B–D), suggesting that the pro-repair function of IL-4 is dependent on hepatic macrophage-derived HB-EGF. Collectively, these data demonstrate that IL-4/IL-4Rα signalling in hepatic macrophages results in HB-EGF production which is essential in liver regeneration and repair after IR injury.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Liver repair after IR injury is impaired when hb-egf is deleted in hepatic macrophages. Male Hb-egf^fl/flClec4f-Cre^+/-mice and WT littermates (Male Hb-egf^fl/flClec4f-Cre^-/-) were subjected to hepatic IR surgery. After 24h, Hb-egf^fl/flClec4f-Cre^+/- mice were divided into two groups and injected (i.p.) with PBS or IL-4c (n=6/group). (A) Serum ALT levels at 6h after IR surgery. (B, C) Liver necrosis (N, outlined areas) was evaluated and quantified on day 7 after IR surgery. (D) Liver pathology was assessed by using the Suzuki’s scoring system on day 7 after IR surgery. (E, F) Proliferative hepatocytes were stained for Ki67 and quantified on day 3 after IR surgery (n=4/group). (G, H) PCNA protein expression was detected by Western blotting and quantified. (I, J) p-EGFR and t-EGFR protein expression levels were detected by Western blotting and quantified. Two-tailed unpaired Student’s t-test with Welch’s correction was performed in A, F, H and J. One-way ANOVA was performed in C and D.

Discussion

The current studies demonstrate that eosinophils, persisting in the liver beyond the injury phase, orchestrate liver repair after IR injury and acute injury caused by APAP and CCl₄. Eosinophil-derived IL-4, but not IL-13, activates hepatic macrophages through IL-4Rα signalling to produce HB-EGF, thereby facilitating hepatocyte proliferation.

Tissue repair after injury is a complex process critical for the survival of the organism. Subsets of innate immune cells are recruited to the wound site to clear debris and dead cells,44–46 triggering the proliferation of epithelial cells or progenitor cells47–50 and inducing vascularisation.51 The involvement of eosinophils in tissue repair has emerged as a previously unrecognised function. Recent studies of skeletal muscle injury, peritonitis and myocardial infarction have demonstrated the indispensable role of these cells in tissue repair.24–26 28–30 52 Our current study, using congenic and iPHIL mice and adoptive transfer of bmEos, provides the first evidence to support the indispensable role of eosinophils in liver repair after IR injury and in other models of acute liver injury, caused by APAP and CCl₄. An observation common in the published studies and ours is that structural cell damage results in eosinophil infiltration to the injured tissues.31 53 This is probably because that injured tissues compel macrophages to release eotaxin-2 (CCL24), a crucial eosinophil chemoattractant,53 and type 2 innate lymphoid cells (ILC2s) are suggested to be potent inducers of eosinophil migration, either through their production of IL-5 or potentially through the production of eotaxin-1 (CCL11).54–56 In contrast, when cellular injury does not occur, such as in partial hepatectomy (PHx), we did not observe eosinophil infiltration to the liver (data not shown). However, there is one report describing that eosinophil-derived IL-4 could promote hepatocyte proliferation after PHx.39 Given that PHx causes intestinal perturbation, it is possible that eosinophils, which are abundant in the gut, may contribute to liver regeneration. The notion that IL-4 can directly stimulate hepatocyte proliferation is contradicted by another study of PHx,40 which showed that IL-4 did not directly affect hepatocyte proliferation. We generated mice with hepatocyte-specific IL-4Rα deletion to investigate if hepatocytes were a direct cellular target of IL-4. Our data ruled out this possibility and instead demonstrated that hepatocyte proliferation was attenuated when IL-4Rα was deleted in hepatic macrophages.

Macrophages are critically involved in the tissue repair process. A recent study of concanavalin A-induced acute liver injury demonstrated that hepatic macrophages encircled the necrotic areas to prevent further damage of hepatocytes and eventually eliminate necrotic lesions.57 It is known that macrophages adopt an inflammatory phenotype when they first enter an injured or infected tissue. They then switch to a pro-repair phenotype with the production of anti-inflammatory cytokines, growth factors and angiogenic factors.58–60 Although it is widely considered that macrophage activation through IL-4Rα contributes to tissue repair, the experimental evidence had been circumstantial. In vivo studies of macrophage-specific deficiency of IL-4Rα or identifying macrophage-derived pro-repair molecules are lacking. The current study filled a knowledge gap regarding how macrophage plasticity is regulated. Our data demonstrate that eosinophils play a crucial role, through IL-4, in promoting macrophages to produce HB-EGF that is necessary for liver repair after IR injury.

With regard to activating macrophages, IL-4 and IL-13 have been considered indistinguishable in promoting M2 polarisation.38 61 62 However, a more detailed investigation of downstream effects revealed differential gene targeting by IL-4 vs IL-13. Our data demonstrate that IL-4, but not IL-13, triggers HB-EGF production by macrophages. We also found that IL-4, but not IL-13 could rescue liver repair in ΔdblGATA1 mice. Antibody-mediated neutralisation of IL-4, but not IL-13, significantly delayed liver repair after IR injury in WT mice. Although IL-4 and IL-13 share the same receptor, IL-4Rα/IL-13Rα1, IL-4 is known to also bind to IL-4Rα/γc, which may contribute to the unique functions of IL-4. In congruent with our finding, several studies of tissue repair after myocardial infarction show the involvement of IL-4, but not IL-13.28–30 Another study of traumatic muscle wounds also showed that implantation of biomaterial scaffold induced a pro-regenerative response mediated by T helper 2 activation to produce IL-4.63

Through binding to EGF receptors, HB-EGF promotes cell proliferation, migration, adhesion and differentiation.64 Owing to these functions, HB-EGF has been implicated in wound healing and repair after injury in many tissues.65 In the liver during recovery from injuries, the level of HB-EGF is elevated.66 Studies of liver injury caused by CCl₄, activation of Fas receptor or bile duct ligation show that HB-EGF treatment increases hepatocyte proliferation, decreases cell death and reduces areas of fibrosis.67–69 The current study provides the first evidence that HB-EGF plays a critical role in promoting liver repair after IR injury. More importantly, our data demonstrate that liver repair is significantly delayed when HB-EGF is specifically deleted from hepatic macrophages, suggesting that macrophages are a major source of HB-EGF. This is consistent with the fact that HB-EGF was first identified in the culture medium of human peripheral blood mononuclear cells.70 Further studies suggested that M2 polarised macrophages were a source of HB-EGF. For example, a head and neck cancer cell line (CAL17) co-cultured with M2 macrophages was resistant to radiation in an HB-EGF-dependent manner.71 Another in vitro study showed that co-culture with M2-polarised macrophages increased ovarian cancer cell proliferation and that this effect was abrogated by HB-EGF neutralising antibodies.72 A more recent in vivo study demonstrated that macrophages that infiltrated into the pancreas after cerulein-induced experimental pancreatitis produced high levels of HB-EGF. Importantly, this study showed that myeloid-specific HB-EGF deletion impaired DNA repair and decreased epithelial cell proliferation, resulting in delayed tissue recovery after pancreatic injury.73 It is known that HB-EGF is synthesised as a membrane-bound pro-HB-EGF, which is cleaved by a disintegrin and metalloprotease (ADAM) and matrix metalloproteinase, and the pro-inflammatory cytokine IL-1β to soluble HB-EGF.74 75 Although both pro-HB-EGF and soluble-HB-EGF can bind to the EGF receptor,65 76 a limitation of the current study is that we have yet to elucidate which form is at play. It may be possible that the IL-4/IL-4Rα signalling in macrophages is important in promoting HB-EGF synthesis and is involved in the shedding of pro-HB-EGF.

Another limitation of this study is the primary focus on hepatocyte proliferation during the liver repair process. It is important to recognise that the eosinophil-macrophage crosstalk may also impact other aspects of liver repair, such as resolving inflammation, remodelling extracellular matrix and promoting angiogenesis. These possibilities warrant further investigation in order to gain a better understanding of the pro-repair function of eosinophils.

In summary, we uncovered a previously unappreciated pro-repair function of eosinophils after hepatic IR injury. The experimental findings provided insights into the underlying mechanism of liver repair involving eosinophil and macrophage crosstalk through IL-4 signalling and HB-EGF production. There is currently no treatment modality to improve or accelerate tissue recovery after hepatic IR injury. Developing such a modality could enable the safe transplantation of ‘marginal’ livers and expand the donor organ pools. The current study provides strong evidence to further explore eosinophils and the IL-4/HB-EGF axis as novel approaches to improve the outcomes of liver transplantation and patients with acute liver injury.