Abstract

   The regenerative capability of the liver is remarkable, but further
   research is required to understand the role that neutrophils play in
   this process. In the present study, we reanalyzed single‐cell RNA
   sequencing data from a mouse partial hepatectomy (PH) model to track
   the transcriptional changes in hepatocytes and non‐parenchymal cells.
   Notably, we unraveled the regenerative capacity of hepatocytes at
   diverse temporal points after PH, unveiling the contributions of three
   distinct zones in the liver regeneration process. In addition, we
   observed that the depletion of neutrophils reduced the survival and
   liver volume after PH, confirming the important role of neutrophils in
   liver regeneration. CellChat analysis revealed an intricate crosstalk
   between neutrophils and macrophages promoting liver regeneration and,
   using weighted gene correlation network analysis, we identified the
   most significant genetic module associated with liver regeneration. Our
   study found that hepatocytes in the periportal zone of the liver are
   more active than in other zones, suggesting that the crosstalk between
   neutrophils and macrophages might be a potential target for liver
   regeneration treatment.

   Keywords: crosstalk, immune cells, liver regeneration, macrophages,
   neutrophils, partial hepatectomy
     __________________________________________________________________

   Neutrophils interact with macrophages increase the secretion of
   interleukin (IL)‐6 and hepatocyte growth factor (HGF). This interaction
   between macrophages and hepatocytes, in turn, promotes liver
   regeneration. Neutrophils also stimulate the differentiation of
   monocytes into macrophages, which accelerates liver proliferation.
   Neutrophil depletion results in a decreased interaction with monocytes
   and macrophages, which decreases liver regeneration.

   graphic file with name FEB4-14-922-g006.jpg
     __________________________________________________________________

Abbreviations

   ALT
          alanine aminotransferase

   AST
          aspartate aminotransferase

   Ccl
          C‐C motif chemokine ligand

   Cxcl
          C‐X‐C motif chemokine ligand

   HCC
          hepatocellular carcinoma

   HFD
          high fatty diet

   IL
          interleukin

   KC
          Kupffer cell

   KEGG
          Kyoto Encyclopedia of Genes and Genomes

   Lcn2
          lipocalin‐2

   Mt
          metallothionein

   NPCs
          non‐parenchymal cells

   Orm
          orosomucoid

   PH
          partial hepatectomy

   qRT‐PCR
          quantitative real‐time PCR

   scRNA‐seq
          single‐cell RNA sequencing

   t‐SNE
          t‐distributed stochastic neighbor embedding

   WGCNA
          weighted gene correlation network analysis

   Despite the array of available treatment modalities and potential
   combination therapies, the prognosis for hepatocellular carcinoma (HCC)
   remains dismal, with a 5‐year survival rate of approximately 18%
   [[40]1, [41]2, [42]3]. Liver transplantation and surgical resection are
   the most effective approaches for treating HCC [[43]4]. The
   extraordinary regenerative prowess of the adult liver allows for
   recuperation following radical resection or transplantation [[44]5].
   However, because most patients with HCC present with underlying liver
   disease [[45]6], liver resection or transplantation necessitates a
   concerted effort to preserve as much parenchymal tissue as possible,
   thus mitigating the risk of liver failure [[46]7]. Therefore, exploring
   the mechanism underlying liver regeneration is of profound
   significance.

   The immune system plays an indispensable role in the orchestration of
   liver regeneration. A multifaceted interplay of immune components,
   including immune cells, cytokines, complements, immunoglobulins and
   cell adhesion molecules, substantially contributes to liver
   regeneration [[47]8, [48]9]. Moreover, the immune microenvironment is
   pivotal in liver inflammatory diseases and immunocompromised liver
   conditions [[49]10]. Neutrophils play a pivotal role in liver diseases,
   including cancer, making them attractive targets for the treatment of
   these diseases [[50]11]. In the present study, we investigated the role
   of neutrophils in liver regeneration. Neutrophils possess sophisticated
   machinery for essential functions in immunity and inflammation.
   Toll‐like receptor 5‐mediated signaling accelerates neutrophil and
   monocyte recruitment during the priming phase, expediting liver
   regeneration [[51]12]. Furthermore, in the progression phase,
   neutrophils actively phagocytose apoptotic extracellular vesicles,
   leading to the secretion of hepatocyte growth factor and fibroblast
   growth factor‐2, further promoting liver regeneration [[52]13].
   Notably, after partial hepatectomy (PH) or associated liver partition
   and portal vein ligation for staged hepatectomy, neutrophils and
   macrophages undergo a phenotypic transition from pro‐inflammatory to
   pro‐regenerative states as they diligently clear cell debris [[53]14,
   [54]15].

   In the present study, we conducted a meticulous reanalysis of
   single‐cell RNA sequencing (scRNA‐seq) data derived from a mouse PH
   model, employing cutting‐edge methodologies, including weighted gene
   correlation network analysis (WGCNA), and pseudotime analysis to
   unravel the regenerative capacity of hepatocytes at diverse temporal
   points and the contributions of three distinct zones in the liver
   regeneration process. CellChat analysis (Suoqin Jin, CA, USA,
   [55]https://github.com/sqjin/CellChat) aimed to trace the intricate
   cellular transitions and heterogeneities within immune cells during the
   initiation and progression peak phases of liver regeneration.

Materials and methods

scRNA‐seq data processing

   The scRNA‐seq data of the PH samples were processed using r, version
   4.2.1 (R Foundation, Vienna, Austria) and the Seurat package [[56]16].
   The harmony package (Ilya Korsunsky, MA, USA,
   [57]https://github.com/immunogenomics/harmony) was used to remove batch
   effects. After eliminating genes expressed in < 10 cells and cells
   expressing < 200 genes, cell samples with a mitochondrial gene
   proportion < 15% were included in the subsequent analysis and
   logarithmically normalized. We conducted principal component analysis
   on the scaled data and implemented a graph‐based clustering approach.
   We harnessed t‐distributed stochastic neighbor embedding (t‐SNE) to
   visualize and explore the datasets. monocle, version 2.0
   (cole‐trapnell‐lab.github.io) was used for pseudotime analysis in
   accordance with the online documentation [[58]17].

Analysis of differential gene enrichment of clusters

   The marker genes were determined using the ClusterProfilerand
   org.Mu.eg.db packages
   ([59]https://bioconductor.org/packages/release/bioc/html/ClusterProfile
   r.html;
   [60]https://www.bioconductor.org/packages/release/data/annotation/html/
   org.Hs.eg.db.html). Gene Ontology (http://geneontology.org) and Kyoto
   Encyclopedia of Genes and Genomes (KEGG) (https://www.genome.jp/kegg)
   enrichment analyses were performed, using q < 0.05 to determine
   statistically significant enrichment. We used the scMetabolism package
   to analyze the enrichment of metabolism pathways [[61]16].

Identification of modules of correlated genes and link modules with traits

   WGCNA can be used to find modules of correlated genes and identify
   disease‐related biomarkers. We used the WGCNA package [[62]18] to
   identify the regenerative‐associated marker genes. The analysis
   involved generating similarity, adjacency and topological overlap
   matrices through Pearson correlation analysis. A soft threshold power
   of 18 was chosen to define the adjacency matrix based on the criterion
   of approximate scale‐free topology, and 0.8 was used as the scale
   independence. Subsequently, a tree diagram was plotted and hierarchical
   clustering grouped the modules related to clinical traits. The final
   correlation module (P < 0.05) was the basis for the subsequent
   analyses.

Cell–cell communication analysis

   CellChat was used to analyze cell‐to‐cell communication in a single
   dataset. We inspected the cell type‐specific RNA expression of various
   ligands in the liver secretome and their corresponding receptors,
   mapping numerous unique clusters of ligand‐receptor pairs with cell
   type–specific expression patterns. We also used CellChat to analyze and
   visualize the data [[63]19].

Animal experiments

   Six‐ to eight week‐old male C57BL/6 mice were bought from the Model
   Animal Research Center of Nanjing University (Nanjing, China) and
   randomly assigned to each group (n = 10). All animal experiments were
   performed up to standard followed by the Ethics Committee of the
   Institutional Animal Care and Use Committee (IACUC) (approval number:
   ZJCLA‐IACUC‐20050066). We used a 70% PH model as first reported by
   Higgins and Anderson and modified by Boyce [[64]20, [65]21]. PH is a
   widely used model of liver regeneration that closely resembles clinical
   liver recession, and its structure is similar to that of clinical liver
   recession. During the surgical procedure, when mice were under
   anesthesia (inhalation anesthesia with 5% isoflurane during induction
   and 3% isoflurane for maintenance), the right median (30%), left median
   (10%) and left lateral (30%) lobes of the liver were removed. Each of
   these lobes was tied off at its base using a preformed silk loop, which
   was manipulated around the lobes before being secured. We killed the
   mice 48 h after PH because the peak of the mice's regeneration phase
   occurred between 36 and 48 h [[66]20, [67]21]. The liver‐to‐body weight
   ratio was calculated using the weight of lobes to delineate the
   increasing rate of liver volume (liver‐to‐body weight ratio = liver
   weight/body weight, g·g^−1).

   InVivoMAb anti‐mouse LY6G (BP0075‐1; Bio X Cell, Lebanon, NH, USA),
   InVivoMAb anti‐mouse Ly6G/Ly6C (Gr‐1) (BE0075; Bio X Cell) and anti‐rat
   IgG2a (BE0122; Bio X Cell) were used to deplete neutrophils, as
   previously reported [[68]22]. Rat IgG2a isotype was used as a control
   (BP0089; Bio X Cell) used as control (control: 50 μg·day^−1 7 days
   prior and once after PH; anti‐Gr1: 50 μg·day^−1 7 days prior and once
   after PH, combo: 50 μg·day^−1 anti‐Ly6G + 50 μg·day^−1 anti‐rat IgG2a
   7 days prior and once after PH).

Cell culture

   Monocytes were harvested from the bone marrow of C57BL/6 mice, then
   purified using gravity centrifugation (520 g for 5 min) and seeded in
   an incomplete medium. The unattached cells were removed and the
   adherent cells were added to a complete 1640 medium after 6 h of
   incubation under 5% CO[2] at 37 °C. Mouse primary neutrophils were
   isolated following a protocol reported for the isolation of mouse
   neutrophils [[69]23]. Neutrophils were purified using Histopaque1077
   (Sigma, St Louis, MO, USA) and Histopaque1119 (Sigma) under gravity
   centrifugation (800  g for 30 min). Neutrophils were incubated with a
   complete 1640 medium (5% CO[2] at 37 °C). Complete 1640 medium
   comprised 10% fetal bovine serum (Gibco, Waltham, MA, USA),
   100 IU·mL^−1 penicillin and 100 μg·mL^−1 streptomycin.

Flow cytometry

   Single cells were isolated and digested using a modified liver
   collagenase perfusion method from the livers of the control and PH mice
   [[70]24]. We used a 70‐mm filter (BD Biosciences, Franklin Lakes, NJ,
   USA) to perfuse livers. Total immune cells were purified using
   medium‐speed gravity centrifugation (520 g  for 5 min × 3). Live cells
   were stained using Fixable Viability Stain 780 (dilution 1 : 1000;
   565388; BD Biosciences) and performed in FACS buffer (4 °C for 30 min).
   Primary antibodies against CD45 (dilution 1 : 100; 75‐0451‐U100; TONBO,
   Tokyo, Japan), CD11b (dilution 1 : 100; 85‐0112‐U100; TONBO), Ly6G
   (dilution 1 : 100; 127633; BioLegend, San Diego, CA, USA), F4/80
   (dilution 1 : 100; 123137; BioLegend), LY6C (dilution 1 : 100;
   20‐5932‐U100; TONBO) and CLEC4F (dilution 1 : 100; 156804; BioLegend)
   were carried out in FACS buffer (4 °C for 60 min). Cells were sorted
   using a BD Aria III, and data were analyzed using flowjo, version 10
   ([71]https://www.flowjo.com).

Quantitative real‐time PCR

   We utilized the Trizol Reagent (Invitrogen, Waltham, MA, USA) to
   extract the total RNA and the PrimeScript RT Master Mix kit (Takara,
   Shiga, Japan) to reverse‐transcribe into cDNA. Quantitative real‐time
   PCR (qRT‐PCR) was performed using the Q6 real‐time PCR system (Applied
   Biosystems, Waltham, MA USA) with the SYBR Green Master Mix (Takara).
   Target genes were normalized to GAPDH expression levels. Each group had
   three biological replicates, and each sample had three technical
   replicates. The primers used for qRT‐PCR were: Gapdh (forward:
   5′‐AGGTCGGTGTGAACGGATTTG‐3′, reverse: 5′‐GGGGTCGTTGATGGCAACA‐3′), Ccl2
   (forward: 5′‐TTAAAAACCTGGATCGGAACCAA‐3′, reverse:
   5′‐GCATTAGCTTCAGATTTACGGGT‐3′), Ccl3 (forward:
   5′‐TTCTCTGTACCATGACACTCTGC‐3′, reverse: 5′‐CGTGGAATCTTCCGGCTGTAG‐3′),
   Ccl5 (forward: 5′‐GCTGCTTTGCCTACCTCTCC‐3′, reverse:
   5′‐TCGAGTGACAAACACGACTGC‐3′), Cxcl1 (forward:
   5′‐CTGGGATTCACCTCAAGAACATC‐3′, reverse: 5′‐CAGGGTCAAGGCAAGCCTC‐3′),
   Sox9 (forward: 5′‐GAGCCGGATCTGAAGAGGGA‐3′, reverse:
   5′‐GCTTGACGTGTGGCTTGTTC‐3′), Ki67 (forward:
   5′‐GAGGAGAAACGCCAACCAAGAG‐3′, reverse: 5′‐TTTGTCCTCGGTGGCGTTATCC‐3′),
   hepatocyte growth factor (forward: 5′‐ATGTGGGGGACCAAACTTCTG‐3′,
   reverse: 5′‐GGATGGCGACATGAAGCAG‐3′), Pcna (forward:
   5′‐TTTGAGGCACGCCTGATCC‐3′, reverse: 5′‐GGAGACGTGAGACGAGTCCAT‐3′).

Immunofluorescence staining and immunohistochemical staining

   The liver sections (4 μm thick) were embedded, sliced, dried, dewaxed
   and rehydrated for the experiments. Next, they were quenched with 3%
   H[2]O[2] (if needed), followed by heat‐induced epitope retrieval in
   citrate buffer (10 mm, pH 6, 95 °C for 20 min). We used 5% BSA (A7906;
   Sigma) to block the non‐specific antigens. Anti‐Ly6G antibody (dilution
   1 : 200, ab238132; Abcam, Cambridge, MA, USA), anti‐Ki67 antibody
   (dilution 1 : 500, ab15580; Abcam) and anti‐Pcna antibody (dilution
   1 : 500, ab18197; Abcam) were incubated overnight at 4 °C. Horseradish
   peroxidase‐conjugated goat‐anti‐rabbit IgG (dilution 1 : 100, PR30011;
   Proteintech, Rosemont, IL, USA) and CoraLite488‐conjugated
   goat‐anti‐rabbit IgG or rhodamine‐conjugated goat‐anti‐mouse IgG
   (dilution 1 : 100, SA00013‐2, SA00007‐2; Proteintech) were used as
   secondary antibody at 4 °C for 1 h. Immunofluorescence used
   4′,6‐diamidino‐2‐phenylindole for nucleus staining (Abcam), whereas
   immunohistochemical staining of the nucleus was performed using
   hematoxylin. Slides were mounted and visualized using a microscope
   (Olympus, Tokyo, Japan).

Serum analysis

   Mouse serum alanine aminotransferase (ALT) and aspartate
   aminotransferase (AST) were measured using an ALT and AST Assay kit
   (E‐BC‐K235‐M, E‐BC‐K236‐M; Elabscience, Wuhan, China) in accordance
   with the manufacturer's instructions.

Statistical analysis

   No pre‐processing of the data was carried out. For all experiments,
   mice were randomly assigned to groups of 10 mice. Only dead mice were
   excluded from the experiments. The immunohistochemical and
   immunofluorescence staining experiments were conducted in a
   double‐blinded manner, ensuring the unbiasedness and reliability of the
   results. We used image j, version 1.53t (NIH, Bethesda, MD, USA) to
   quantify each liver sample, and all liver sections were repeated at
   least three times. Data were analyzed using prism, version 8.0
   (GraphPad Software Inc., San Diego, CA, USA) and the results are
   presented as the mean ± SD. All the immunofluorescence,
   immunohistochemistry, flow cytometry and qPCR data were repeated at
   least thrice to quantify liver tissues. We used the Shapiro–Wilk test
   [[72]25] to check the normal distribution of characteristics
   (P > 0.05). An unpaired two‐tailed Student's t‐test was used to
   determine statistical significance between two groups and multiple
   group data were analyzed using one‐way analysis of variance when the
   data conformed to a normal distribution. A chi‐square test was used to
   analyze abnormal distributions. P < 0.05 was considered statistically
   significant.

Results

Hepatocyte subsets analysis during liver regeneration

   We first analyzed scRNA‐seq from adult mouse livers 0, 24 and 48 h
   after PH. In total, we recovered 199 229 single cells that passed
   quality control. We identified 19 clusters in liver samples. Clusters
   0–9 were identified as hepatocytes based on the presence of parenchymal
   cell marker genes (Alb, Cyp2e1 and Oat), and clusters 10–18 were
   non‐parenchymal cells (NPCs) (for definition details, see below)
   (Fig. [73]1). During PH, the main cell subsets were clusters 0, 1 and 2
   at 0, 24 and 48 h, respectively (Figs [74]2A,B and [75]3A). Moreover,
   cell correlation analysis showed clusters 0, 7, 8 and 9 were different
   from clusters 1, 2, 3, 4, 5 and 6. The above comment shows that
   clusters 0, 7, 8 and 9 were quiescent cells, whereas clusters 1–6 were
   proliferative cells (Fig. [76]3B). Cluster 0 (Sult2a8, Cyp2e1 and Alb)
   is the main hepatocyte subset that exists in normal liver tissue
   (Fig. [77]3C); mup11, mup3 and mup17 were highly expressed in cluster 0
   and expressed at low levels in clusters 1 and 2, indicating that
   synthetic liver function may be damaged at 24 and 48 h after PH
   [[78]26]. Interestingly, cluster 1 (Mt1/2, Gstm3, Oat) and cluster 2
   (Mt1/2, Saa1/2, Ifith2) were the major components at 24 and 48 h after
   PH (Fig. [79]3D,E), depending on the mitochondrial activity in
   hepatocytes. The lipocalin‐2 (Lcn2) gene was highly expressed in
   clusters 1 and 2, whereas metallothionein (Mt) and orosomucoid (Orm)
   were overexpressed in cluster 2 compared to cluster 1. Metabolic
   pathway enrichment analysis was used for hepatocytes, and it was found
   that a higher inflammatory response after PH and the immune response
   contribute widely to liver regeneration (Fig. [80]3F).

Fig. 1.

   Fig. 1
   [81]Open in a new tab

   The definition of 19 clusters. Clusters 0–18 were identified using
   enrichment genes. The intensities and directions of the correlations
   are indicated on the right side of the diagram. The color and cluster
   names are shown on the right.

Fig. 2.

   Fig. 2
   [82]Open in a new tab

   Cell type composition of a regenerating liver and dynamically changed
   hepatocytes during liver regeneration. (A) Cells were classified into
   18 subsets using the t‐SNE algorithm. (B) Cells were separated at 0, 24
   and 48 h after PH, using the t‐SNE algorithm.

Fig. 3.

   Fig. 3
   [83]Open in a new tab

   The regenerative capacity of hepatocytes after PH. (A) Proportions of
   the cells for clusters 0–9 at 0, 24 and 48 h after PH. (B) Heat map of
   relative expression of the top genes enriched in hepatocytes. The
   intensity and direction of the correlations are indicated on the right
   side of the heatmap. (C–E) Expression patterns of critical genes in
   cluster 1 and cluster 2 are shown as volcano plots. (F) Distribution of
   the inflammatory response metabolic pathways in clusters 0–9, using the
   t‐SNE algorithm.

   In summary, hepatocytes in the quiescent phase after PH mainly maintain
   normal liver function, whereas those in the active phase experience
   proliferative and protective effects induced by the above pathways.

Pseudotime analysis of hepatocytes

   We constructed discrete cell‐state trajectories from the initiation
   progression phase to the peak phase of regeneration of each state cell
   (PH 0 h → PH 24 h → PH 48 h). We found that PH 0 h, PH 24 h and PH 48 h
   quiescent cells resided on separate arms of the trajectory
   (Fig. [84]4A,B), similar to the previous study in Fig. [85]2B. Cluster
   0 accounted for the major part 0 h after PH. Clusters 8 and 9 were
   transitioning cells and cluster 7 was highly expressed at 24 and 48 h
   after PH.

Fig. 4.

   Fig. 4
   [86]Open in a new tab

   The dynamic change of hepatocytes after PH. (A) Pseudotime plot
   indicating cellular trajectories of hepatocytes from clusters 0, 7, 8
   and 9 after PH. Single‐cell trajectories were constructed, and
   pseudotime values were calculated using monocle 2. The trajectories are
   colored according to the pseudotime. (B) Trajectory demonstrating the
   three distinct states of hepatocytes in clusters 0, 7, 8 and 9. (C)
   Pseudotime plot indicating cellular trajectories of hepatocytes from
   clusters 1–6 after PH. Single‐cell trajectories were constructed, and
   pseudotime values were calculated using monocle 2. The trajectories are
   colored according to pseudotime. (D) Trajectory demonstrating the three
   distinct states of hepatocytes in clusters 1–6. (E) Heat maps
   representing modules of genes in clusters 1–6 that co‐vary along the
   pseudotime at 0, 24 and 48 h after the PH renaturation phases.

   Using the same approach, proliferative cells were analyzed, as shown in
   Fig. [87]4C, and were separated into six branches of the trajectory.
   Cluster 5 was located in the PH 24 h pseudotime arm. Clusters 4 and 6
   were located in the PH 48 h pseudotime arm (Fig. [88]4D). In addition,
   to identify the continuous changes within differential genes along the
   pseudotime, we performed a dynamic analysis of the top 10 genes that
   varied as a function of progress through the pseudotime. Along the
   0–48 h pseudotime trajectory, the expression of Hamp, Oat and Apoa2
   declined, and S100a8, S100a9, mt, Rnase4 and Mup3 simultaneously
   increased. Interestingly, genes for Gstp1, Eabp1, Eabp5, Ifitm2, Gata3
   and Rgn first increased at 24 h and then decreased at 48 h
   (Fig. [89]4E).

   Our analysis successfully captures the cellular plasticity of the
   hepatocytes. This result highlights that not all hepatocytes undergo
   reprogramming after PH. The dramatic change in hepatocytes after PH
   balances mature liver function and liver regeneration.

Hepatocytes performed a diversified proliferative function in three zones

   Based on the different microenvironments, including oxygen levels,
   nutrients and secreted factors, the liver lobule can be roughly divided
   into three distinct zones: the periportal zone (zone 1), the
   pericentral zone (zone 3), and the mid‐lobular zone (zone 2) between
   zone 1 and zone 3 [[90]27]. A recent study has reported that a
   subpopulation of hepatocytes located in zone 2 preferentially
   contributed to new hepatocytes during liver homeostasis and
   regeneration [[91]28, [92]29]. Therefore, we aimed to explore the
   heterogeneity of hepatocytes in the three zones after PH. We identified
   zone 1 (Cdh1), zone 2 (Pklr) and zone 3 (Cyp2e1) based on their
   specific marker genes as previously reported (Fig. [93]5A–C). As shown
   in Fig. [94]5D, the mt‐Atp6, mt‐Co1, mt‐Co2 and mt‐Co3 genes were
   up‐regulated in zone 1, which indicates that mitochondrial metabolism
   is boosted in zone 1 to afford liver regeneration. However, hepatocytes
   in zones 2 and 3 showed high expression of Mup3, 7, 11, 17 and 20,
   suggesting that the hepatocytes may function in normal liver metabolism
   and rarely contribute to the proliferation in zones 2 and 3
   (Fig. [95]5E,F). The hepatocytes in zone 1 were more active than in
   zones 2 and 3, whereas hepatocytes in zones 2 and 3 balanced the normal
   liver function from 0 to 48 h after PH.

Fig. 5.

   Fig. 5
   [96]Open in a new tab

   Hepatocytes performed a diversified proliferative function in three
   zones. (A) Zone 1 hepatocytes were identified based on the zone 1
   marker gene Cdh1, using the t‐SNE algorithm. (B) Zone 2 hepatocytes
   were identified based on the zone 2 marker gene Pklr, using the t‐SNE
   algorithm. (C) Zone 3 hepatocytes were identified based on the zone 3
   marker gene Cyp2e1, using the t‐SNE algorithm. (D) The expression
   patterns of key genes in zone 1 are shown as volcano plots. (E) The
   expression patterns of key genes in zone 2 are shown as volcano plots.
   (F) The expression patterns of key genes in zone 3 are shown as volcano
   plots.

Neutrophils increased after PH

   Non‐parenchymal cells, particularly immune cells, have been reported to
   contribute to liver regeneration. We analyzed the relationship in each
   cluster of NPCs (Fig. [97]6A). Cluster 11 was identified as macrophages
   (Eabp5, Gstp1). Cluster 11 also highly expressed diazepam binding
   inhibitor during liver regeneration (Figs [98]1A and [99]6B), which has
   been reported to stimulate mitochondrial fatty acid oxidation and
   regulate the cell cycle [[100]30]. Cluster 12 was identified as
   neutrophil (sp100a, sp100b) and was dramatically increased from 0 to
   48 h after PH as shown in Fig. [101]6C. Cluster 12 also highly
   expressed Lcn2, interleukin (IL)1b, Ccl6 and galectin 3, which were
   inflammatory‐related genes (Fig. [102]6D). Figure [103]6E shows that
   NPCs were enriched in amino sugar and nucleotide sugar metabolism,
   indicating that NPCs also participate in liver regeneration by
   regulating metabolic pathways. The dramatic changes of neutrophils at
   24 and 48 h after PH raised our attention with expect to exploring the
   function of neutrophils in liver regeneration.

Fig. 6.

   Fig. 6
   [104]Open in a new tab

   Neutrophils increased after PH. (A) Heat map of the relative expression
   of the top genes enriched in non‐parenchymal cells. The intensities and
   directions of the correlations are indicated on the right side of the
   heatmap. (B) Expression patterns of the critical genes in cluster 11
   are shown in volcano plots. (C) Proportions of neutrophils in 0, 24 and
   48 h after PH using the sc‐RNAseq database. (D) Expression patterns of
   the critical genes in cluster 12 are shown in volcano plots. (E)
   Heatmap showing KEGG pathway analysis results of marker genes for
   clusters 10–18.

Depletion of neutrophils damaged liver regeneration

   First, we verified the variation tendency of neutrophils using flow
   cytometry and immunofluorescence, and it had the same results as our
   analysis (Fig. [105]7A,B). Next, we investigated the function of
   neutrophils in liver regeneration. We used neutrophil‐specific
   depletion antibody to delete the mouse liver neutrophils and performed
   a mouse PH model to explore the role of neutrophils (Fig. [106]7C–E).

Fig. 7.

   Fig. 7
   [107]Open in a new tab

   Neutrophils increased after PH. (A) Flow cytometry showed the
   percentage of neutrophils (Ly6G^+CD11b^+) among total immune cells
   (CD45^+) after PH (n = 6, *P < 0.05). (B) Immunofluorescence staining
   of Ly6G^+ neutrophils time after PH (n = 3, *P < 0.05). (C) Schematic
   representation of neutrophil depletion by PH (control: rat IgG2a
   isotype; anti‐Gr1: anti‐Gri antibody; combo: anti‐Ly6G antibody with
   anti‐rat IgG2a). (D) Flow cytometry showed the percentage of
   neutrophils (Ly6G^+CD11b^+) in total immune cells in different groups
   (CD45^+) (n = 6, *P < 0.05). (E) Immunofluorescence staining of Ly6G^+
   neutrophils in different groups (n = 3, *P < 0.05).

   As shown in Fig. [108]8A–C, the survival rate, the liver‐to‐body weight
   ratio, and the serum liver transaminase–ALT and AST levels were
   severely damaged after PH when neutrophils were depleted. The number of
   proliferating hepatocytes (Ki67^+ and Pcna^+ hepatocytes) decreased
   after depletion (Fig. [109]8D–F). In addition, the steatosis area
   increased, indicating the metabolic dysfunction in the absence of
   neutrophils after PH (Fig. [110]8E). The mRNA level also showed that
   proliferative rate and stem cell marker levels decreased when
   neutrophils depleted (Fig. [111]8F). All of these results suggest that
   neutrophils play an indispensable role in liver regeneration and
   maintenance of normal function.

Fig. 8.

   Fig. 8
   [112]Open in a new tab

   Depletion of neutrophils damaged liver regeneration. (A) Kaplan–Meier
   analysis was used to determine the survival of mice 48 h after PH
   (control: rat IgG2a isotype; anti‐Gr1: anti‐Gri antibody; combo:
   anti‐Ly6G antibody with anti‐rat IgG2a, n = 10, *P < 0.05). (B) The
   liver‐to‐body weight ratio after 48 h. Liver to body weight
   ratio = liver weight/body weight (g·g^−1) (n = 5, *P < 0.05). (C) Serum
   levels of transaminase‐ALT and AST are shown (n = 5, *P < 0.05). (D)
   Immunofluorescence staining of marker of proliferation Ki‐67 (Ki67) and
   the number of Ki67^+hepatocytes are shown (n = 3, *P < 0.05). (E)
   Immunohistochemical staining of proliferating cell nuclear antigen
   (Pcna) and the number of Pcna^+ hepatocytes and steatosis areas are
   shown (white arrow: Pcna^+ hepatocytes; black arrow: Pcna^+ hepatocytes
   around the bile duct in zone 1; red arrow: lipid droplet in the liver,
   n = 3, *P < 0.05). (F) The mRNA expression levels of Ki67, Pcna, Sox9
   and Tert were examined using qRT‐PCR. Genes expression was normalized
   to the GAPDH mRNA levels in each sample (n = 3, *P < 0.05).

Crosstalk between neutrophils and macrophages promotes liver regeneration

   Finally, we used CellChat analysis to explore the interactions between
   NPCs (Fig. [113]9A). The number and strength in cluster 12 (S100a8,
   S100a9) with respect to other clusters were the strongest, indicating
   that neutrophils had the most substantial relationship with other
   cells. The interactions between clusters 11 and 12 were the strongest,
   indicating that neutrophils had stronger crosstalk with macrophages.

Fig. 9.

   Fig. 9
   [114]Open in a new tab

   The interaction between neutrophils and macrophages enhances liver
   regeneration. (A) Network diagrams showing cell–cell interactions
   indicated by arrows (edges) pointing in the source‐to‐target direction.
   The thickness indicates the sum of the weighted paths between
   populations, and the color of the arrows corresponds to the source.
   Network diagrams for clusters 10–18 are shown. (B) Flow cytometry was
   used to determine the percentage of macrophages (F4/80^+CD11b^+) in
   total immune cells (CD45^+) in the different groups (n = 6, *P < 0.05).
   (C) Flow cytometry revealed the percentage of Kupffer cells (CLEC4F^+)
   in macrophages in the different groups (F4/80^+CD11b^+CD45^+) (n = 6,
   *P < 0.05). (D) Flow cytometry revealed the percentage of monocytes
   (Ly6C^+CD11b^+) among the total immune cells in the different groups
   (CD45+) (n = 6, *P < 0.05). (E) The mRNA expression levels of Ccl3,
   Ccl5 and Cxcl1 were examined using qPCR analysis (the genes were
   normalized to GAPDH mRNA levels in each sample. Ccl3, C‐C motif
   chemokine ligand 3; Ccl5, C‐C motif chemokine ligand 5; Cxcl1, C‐X‐C
   motif chemokine ligand 1; n = 3, *P < 0.05). (F) Macrophages cultured
   without neutrophils and co‐culture of neutrophils and macrophages
   (n = 3, *P < 0.05).

   Interestingly, the percentages of macrophages, monocytes and Kupffer
   cells all decreased after the depletion of neutrophils
   (Fig. [115]9B–D). This can be explained by the depletion of Ly6C^+
   monocytes when anti‐Gr1 (Ly6C/Ly6G) antibody was used. In total,
   monocytes also decreased after the depletion of neutrophils, whereas
   monocytes were beneficial for liver regeneration via transfer into
   necrotic foci to phagocytose dead cell debris [[116]31]. The number of
   macrophages also decreased after PH because the neutrophils may secrete
   C‐C motif Ccl3, Ccl5 and C‐X‐C motif chemokine ligand (Cxcl)1 to
   recruit macrophages (Fig. [117]9E). The co‐culture of neutrophils and
   macrophages also confirmed the neutrophils stimulated monocytes to
   differentiate into macrophages (Fig. [118]9F). These results suggest
   that the crosstalk between neutrophils and macrophages promotes liver
   regeneration after PH.

WGCNA identified regeneration‐related modules

   Because most HCC patients present at a high age and experience some
   liver diseases [[119]6], such as hyperlipidemia and even non‐alcoholic
   fatty liver disease, the metabolic environment also influences liver
   regeneration [[120]7]. Here, we analyze the dynamic changes of the
   liver with different clinical traits, such as aging, high fatty diet
   (HFD) and regeneration following the PH mouse model to explore the
   different proliferative capabilities of these populations.

   We constructed a WGCNA to determine the relationship between clinical
   traits and liver regeneration (Fig. [121]10A). The module‐trait heatmap
   represents the correlations between the module genes and clinical
   characteristics (Fig. [122]10B). We identified seven modules shown as
   different colors in Fig. [123]10C (the gray module represents genes do
   not belong to any modules). The yellow and brown modules were the most
   significantly correlated with liver regeneration. Next, we performed
   KEGG pathway analysis of different modules to determine the pathway
   involved in each characteristic. Blue module was characterized as
   normal liver. In Fig. [124]11A, several greatly enhanced pathways in
   the blue module, including fatty acid metabolism and drug metabolism
   cytochrome P450, are needed to balance the normal liver function after
   PH. Together with the T cell, B cell receptor signaling
   cytokine‐cytokine pathways, the most relevant modules related to
   activating liver regeneration are essential for controlling hepatocyte
   proliferation and paracrine cell interactions [[125]32]. Brown module
   was most closely related to proliferation. As shown in Fig. [126]11B,
   DNA replication, spliceosome and cell cycle pathways were significantly
   enriched to adapt to the increasing need for cell regeneration after PH
   in the brown module.

Fig. 10.

   Fig. 10
   [127]Open in a new tab

   Construction of weighted gene co‐expression network identification of
   modules. (A) Gene expression pattern‐based DEGs clustering and module
   screening. The top is the gene dendrogram, and the bottom is the gene
   modules with different colors. In total, seven modules were identified
   in this study (the gray module represents genes that do not belong to
   any modules). (B) Sample dendrogram and trait heatmap of each clinical
   trait. (C) Relationship between consensus module eigengenes and
   clinical traits. The module is displayed on the left side of each cell.
   The numbers in the table indicate the corresponding module eigengenes
   and traits correlations, with the P‐values printed below the
   correlations in parentheses. The table is color‐coded according to the
   color legend. The intensities and directions of the correlations are
   indicated on the right side of the heatmap.

Fig. 11.

   Fig. 11
   [128]Open in a new tab

   The metabolic pathways related to the clinical traits. (A, B) KEGG
   pathway enrichment analysis of the (A) blue and (B) brown module genes.
   Pathway names are shown on the left.

Discussion

   In the present study, we have tracked the cellular transitions and
   heterogeneities of the liver using a scRNA‐seq database in a mouse PH
   model. We found that the dramatic change of hepatocytes after PH
   balanced mature liver function and liver regeneration. Importantly, we
   found that immune cells play an important role during liver
   regeneration, and the interaction between neutrophils and macrophages
   enhanced liver regeneration after PH.

   The regeneration capacity of hepatocytes is most important for liver
   regeneration both in homeostasis and injury. Our study found that
   quiescent cells contribute to normal liver function and the
   proliferative cells contribute to liver regeneration, and these two
   types of populated hepatocytes balanced the liver regeneration after
   PH. Cluster 2 in 48 h was more active than cluster 1 in 24 h, according
   to the high expression of the gene Mt1 and Mt2 in cluster 2. Mt is
   critical in protecting the liver from oxidative stress‐induced liver
   injury by activating NF‐κB target genes [[129]33]. Cluster 2 also
   highly expressed Orm compared to cluster 1. Orm is one of the major
   acute‐phase proteins in humans and is mainly synthesized in the liver,
   also playing an important role in liver regeneration [[130]34]. We
   found that hepatocytes in zone 1 were more active than zone 2 and 3,
   whereas hepatocytes in zone 2 and 3 mainly contributed to normal
   metabolism during 0–48 h after PH. However, recent research has
   confirmed that liver regeneration following PH is enabled by the
   proliferation of hepatocytes throughout the liver, rather than by a
   particular zone [[131]35, [132]36]. It is possible that such a
   discrepancy is a result of the different time points after PH that we
   selected. Our findings indicate that, overall, hepatocytes in zone 1
   were more active than those in zones 2 and 3 from 0 to 48 h after PH.
   In addition, we only gathered the highly expressed genes to evaluate
   the activation of the specific population of hepatocytes, which focused
   on cell division, whereas other research focuses on the differentiation
   of stem cells or other cells [[133]36, [134]37, [135]38].

   Importantly, our research confirmed the existing results that NPCs,
   especially immune cells, function as necessary in liver regeneration
   [[136]8]. Hepatocyte reprogramming is closely associated with the
   dynamic injured‐induced immune responses that play crucial roles in the
   initial time of liver regeneration. Our study focused on the dynamic
   change of neutrophil count, and we verified the indispensable role of
   neutrophils after liver resection (Fig. [137]7). In addition, we used
   CellChat analysis and revealed that neutrophils have stronger crosstalk
   with macrophages. The depletion of neutrophils impaired the recruitment
   of macrophages and Kupffer cells (KCs) and further decreased liver
   regeneration, and the co‐culture of neutrophils and monocytes
   accelerated the monocytes to differentiate into the macrophages to help
   liver regeneration after PH (Figs [138]8 and [139]9). It was also
   previously reported that neutrophils and macrophages interact with each
   other and accelerate liver regeneration [[140]39]. CD11/CD18 in
   neutrophils bind with intercellular adhesion molecule‐1 in KCs and
   induce the release of interleukin‐6 and tunor necrosis factor‐α to
   accelerate hepatocytes into the cell cycle [[141]40]. Activated KCs
   directly influence the recruitment of neutrophils through the secretion
   of Cxcl2, which binds to Cxcr2 in neutrophils [[142]41]. KCs can also
   indirectly promote neutrophil recruitment by stimulating other cell
   types, such as hepatic stellate and endothelial cells [[143]42].
   Interestingly, our results showed that neutrophils secreting chemokines
   Cxcl1, Ccl3 and Ccl5 to recruit macrophages are stimulants during PH,
   and the interaction between neutrophils and macrophages is important
   for liver regeneration. Our study also found Lcn2 upregulated at 24 and
   48 h after PH, whereas the ability of neutrophils to perform chemotaxis
   for transmigration depended on a neutrophil secondary Lcn2. LCN are
   members of a family of evolutionarily conserved genes and participate
   in non‐alcoholic steatohepatitis by promoting neutrophil‐macrophage
   crosstalk via the induction of CXCR2 [[144]39] and influence macrophage
   polarization to promote liver regeneration [[145]43].

   The liver balances cell proliferation and metabolism to prevent liver
   function disorders or failure after PH [[146]26]. Hepatocytes have a
   flexible metabolic machinery that can adapt dynamically to changes
   during tissue regeneration. Following PH, vital liver functions must be
   preserved except for cell division, such as maintaining blood glucose
   homeostasis, balancing the degradation of hepatic glycogen stores and
   gluconeogenesis [[147]44]. Simultaneously, a distinct population of
   metabolically hyperactive cells appears to compensate for the temporary
   deficits in liver function [[148]45]. Our study focused on liver
   metabolism and its dynamic regulation during liver regeneration, and we
   found that fatty acid metabolism was strongly overexpressed 24 and 48 h
   after PH to offer extra energy needed for liver regeneration.

   However, the present study also had some limitations. There was no
   in‐depth research into the mechanism of neutrophils and liver
   regeneration. In addition, WGCNA did not gather the most significant
   pathways in the HFD module and age modules, which cannot provide any
   advice for the age or HFD populations when they go through PH.

Conclusions

   Given that most patients with HCC have underlying liver disease, it is
   meaningful to promote liver regeneration after liver transplantation or
   resection. Our study unraveled the regenerative capacity of hepatocytes
   at diverse temporal points after PH, unveiling the contributions of
   three distinct zones in the liver regeneration process. Our study also
   explored the crosstalk in NPCs and the dynamic metabolism changes in
   the liver, also indicating a potential target for liver regeneration
   treatment.

Conflicts of interest

   The authors declare that they have no conflicts of interest.

Peer review

   The peer review history for this article is available at
   [149]https://www.webofscience.com/api/gateway/wos/peer‐review/10.1002/2
   211‐5463.13803.

Author contributions

   WS and XX designed and conceived the study. CYY and YYJ conducted most
   of the experiments and prepared the manuscript. LJJ and CH provided
   technical support for the animal experiments. CYY, NX and WXD performed
   the RNA‐sequencing data analysis. SZX revised the manuscript. YYJ, LJJ,
   CH, WXD and SZX reviewed the manuscript. All of the authors read and
   approved the final version of the manuscript submitted for publication.

Acknowledgements