Abstract

   Rationale: Understanding the immune mechanisms associated with liver
   transplantation (LT), particularly the involvement of tissue-resident
   memory T cells (TRMs), represents a significant challenge.

   Methods: This study employs a multi-omics approach to analyse liver
   transplant samples from both human (n = 17) and mouse (n = 16),
   utilizing single-cell RNA sequencing, bulk RNA sequencing, and
   immunological techniques.

   Results: Our findings reveal a comprehensive T cell-centric landscape
   in LT across human and mouse species, involving 235,116 cells. Notably,
   we found a substantial increase in CD8^+ TRMs within rejected grafts
   compared to stable ones. The elevated presence of CD8^+ TRMs is
   characterised by a distinct expression profile, featuring upregulation
   of tissue-residency markers (CD69, CXCR6, CD49A and CD103^+/-,), immune
   checkpoints (PD1, CTLA4, and TIGIT), cytotoxic markers (GZMB and IFNG)
   and proliferative markers (PCNA and TOP2A) during rejection.
   Furthermore, there is a high expression of transcription factors such
   as EOMES and RUNX3. Functional assays and analyses of cellular
   communication underscore the active role of CD8^+ TRMs in interacting
   with other tissue-resident cells, particularly Kupffer cells,
   especially during rejection episodes.

   Conclusions: These insights into the distinctive activation and
   interaction patterns of CD8^+ TRMs suggest their potential utility as
   biomarkers for graft rejection, paving the way for novel therapeutic
   strategies aimed at enhancing graft tolerance and improving overall
   transplant outcomes.

   Keywords: liver transplantation, tissue-resident memory T cells,
   multi-omics, graft rejection, immune tolerance

Introduction

   The enduring success of liver transplantation (LT) is intricately
   linked to the intricate dynamics within the transplant immune
   microenvironment. Post LT, the orchestration of interactions between
   donor-derived cells and the recipient's immune system is pivotal for
   graft acceptance [53]^1^, [54]^2. Allograft immunity represents a
   multifaceted process, arising from intricate interactions among various
   immune cell types, including lymphocytes, monocytes, macrophages, and
   dendritic cells. In this dynamic scenario, recipient alloreactive T
   cells identify donor alloantigens, presented either by donor-derived or
   recipient antigen-presenting cells. This recognition triggers an
   adaptive inflammatory immune response, a critical pathway leading to
   allograft rejection [55]^3. T cells, composed of diverse subgroups,
   play a pivotal role in modulating the immune response, particularly in
   enhancing anti-donor reactions [56]^4^, [57]^5. These subgroups include
   subsets such as CD4^+ regulatory T cells (Tregs), related to immune
   tolerance, and cytotoxic CD8^+ T cells, often associated with
   transplant rejection [58]^6. Understanding the distinct functions of
   these T cell subsets with different locations is essential in
   transplant immunology. Therapeutic strategies targeting these cells are
   gaining momentum in transplant medicine. The quest to identify cellular
   markers and T cell subsets driving this rejection is vital for devising
   targeted monitoring strategies and therapeutic interventions, marking a
   substantial leap forward in transplant immune.

   Tissue-resident memory T cells (TRMs), a specialized subset of the
   immune arsenal, have emerged as pivotal mediators in the immunology of
   nonlymphoid tissues [59]^7^, [60]^8 and solid organ transplantation,
   influencing the fate of transplanted hearts [61]^9, kidneys [62]^10,
   and lungs [63]^11, frequently tipping the balance toward rejection.
   TRMs are distinct both phenotypically and transcriptionally from
   circulating memory T cells. They are characterized by the expression of
   activation, retention, and adhesion markers such as C-type lectin CD69,
   integrin CD103, and are capable of producing effector cytokines like
   interferon-γ (IFN-γ), granzyme B (GZMB), and proliferating locally
   [64]^12^, [65]^13. Both CD4^+ TRMs and CD8^+ TRMs have been identified
   as playing pivotal, yet differing, roles during the course of organ
   transplantation. Notably, long-lived alloreactive CD4^+ TRMs, residing
   in organs or tissues other than the transplant site, have been shown to
   significantly contribute to organ allograft rejection, as evidenced in
   skin and heart transplantation studies [66]^9. Further research has
   demonstrated that antigens and IL-15, presented by dendritic cells, are
   crucial for the maintenance of CD8^+ TRMs in scenarios as observed in a
   mouse kidney transplantation model [67]^14. The liver, recognized for
   its immunotolerant environment, presents a unique set of
   transplantation challenges distinct from these organs [68]^15.
   Liver-resident macrophages, named Kupffer cells, reside in the
   transplanted liver and contribute to the maintenance of transplant
   immune. While our previous studies have mapped the single-cell
   environment of the transplanted liver, highlighting the significant
   roles played by T cell and myeloid cell populations [69]^16, there
   remains a notable gap in our understanding of the specific
   contributions of TRMs to liver transplant outcomes.

   Our current investigation seeks to bridge this gap by dissecting the
   phenotypic properties and critical functions of TRMs within LT
   contexts, employing a comprehensive multi-omics approach that spans
   single-cell RNA sequencing, bulk RNA sequencing, and benchmark
   experimental assays. This study is poised to shed light on the complex
   roles of TRMs, delving into their potential to trigger, sustain, or
   regulate immune responses post-transplantation. By elucidating these
   roles, we aim to deepen our understanding of the intricate equilibrium
   between immune-mediated rejection and tolerance of LT.

Results

Construction of the Single-Cell Landscape of Human Transplanted Liver

   To delineate the cellular dynamics of human LT, we established a
   comprehensive single-cell landscape from 17 samples, encompassing both
   liver (n = 13) and blood (n = 4), including cases with (n = 9) and
   without rejection (n = 8) ([70]Figure S1A, [71]Table S1). To enhance
   data analysis and minimize batch effects, we integrated samples from
   both DNBelab C4 and 10X Genomics sequencing platforms. This landscape
   comprises 115,026 high-quality cells, categorized into 23 clusters
   post-filtration, and annotated with canonical markers. These markers
   include CD3D, CD3E, and CD3G for T cells (69.99%); CD68, CD14, and
   CD163 for myeloid cells (15.78%); COL1A2, COL3A1, and ACTA2 for
   fibroblasts (1.25%); ERG, ENG, and CD34 for endothelial cells (4.51%);
   ALB, TF, and TTR for hepatocytes (1.86%) (Figure [72]1A, C, and
   [73]Table S1). The analysis revealed a distinct distribution of
   diagnostic categories and tissue types across the cellular landscape
   (Figure [74]1B). Notably, T cells and myeloid cells constituted the
   majority of this landscape, displaying variability across the
   samples—particularly the T cells (Figure [75]1D-E, [76]Figure S1B).

Figure 1.

   [77]Figure 1
   [78]Open in a new tab

   Single-Cell Atlas of Human Liver Transplantation. A) UMAP plot
   depicting cell type identification of 115,026 high-quality cells in
   human liver transplantation. B) UMAP plots colored by spatial
   distribution of cells among diagnoses (rejection and non-rejection) and
   tissues (liver and blood). C) Heatmap displaying the top 5 genes of
   each cell type. D) Bar plots illustrating the proportion of cell types
   in each sample. E) UMAP plots showing the expression of CD3D, CD68, and
   CD49A across the landscape. F) HE and immunohistochemistry of CD4 and
   CD8 in rejection-transplanted liver. Insets highlight areas of T cell
   infiltration. Whole image scale bar, 200 μm. Inset scale bar, 100 μm.
   G) Multi-immunohistochemistry of rejection and non-rejection
   transplanted liver for DAPI (blue), CD4 (white), CD8 (green), FOXP3
   (cyan), and PD1 (red). White arrows point to cells co-expressing CD8
   and PD1. Whole image scale bar, 200 μm. Inset scale bar, 50 μm. H) KM
   analysis showing proportion of CD3^+ and CD4^+ T cells in the
   peripheral blood of LT patients for one month with prognosis.

   Immunohistochemistry (IHC) analyses indicated heightened expression
   levels of CD4 and CD8 in rejection samples (Figure [79]1F).
   Multi-immunohistochemistry (mIHC) further highlighted that CD8^+PD1^+ T
   cells were prevalent in samples from rejection cases (Figure [80]1G and
   [81]Figure S1C). Furthermore, we conducted statistical analyses to
   examine circulating T cells and their correlation with patient outcomes
   following transplantation. Our flow cytometry analysis revealed that a
   higher proportion of CD3^+ and CD4^+ T cells in the peripheral blood of
   LT patients for one month is associated with a more favorable prognosis
   (Figure [82]1H).

CD4^+ TRMs Exhibit Dual Phenotypic Roles in Liver Transplantation

   Our exploration of CD4^+ TRMs highlights their potential to modulate
   immune responses in LT. We identified nine CD4^+ T cell subclusters,
   including Naïve T, Treg, Th2, Th17, and TRM cells, each exhibiting
   unique tissue distributions (Figure [83]2A-B). The predominant cluster,
   characterized by the upregulation of immune checkpoints (PD1, CTLA4,
   TIGIT), was designated as PD1^+CTLA4^+ T cells. Based on gene
   expression profiles, we identified two distinct phenotypes of CD4^+
   TRMs: C2-AREG^+ TRM and C4-IFNG^+ TRM. Both expressed the markers CD69,
   CXCR4, and PCNA, typically associated with CD4^+ TRMs, while C4-IFNG^+
   TRM showed higher expression of IFNG, indicating a TRM1 phenotype. TRM
   scoring across subclusters affirmed that both C2-AREG^+ TRM and
   C4-IFNG^+ TRM phenotypes possessed significantly elevated TRM scores
   compared to others, though no notable differences were observed between
   the two (Figure [84]2C).

Figure 2.

   [85]Figure 2
   [86]Open in a new tab

   Identifying CD4^+ T Cell Subsets in Human Transplanted Liver and Blood.
   A) tSNE plot for re-clustering and cell type identification of 8,836
   CD4^+ T cells. B) tSNE plot colored by spatial distribution of cells
   among tissues (liver and blood). C) Dot plots showing the expression of
   TRM-related genes among clusters. Violin plots illustrating the TRM
   score among AREG^+ TRM, IFNG^+ TRM, and other CD4^+ T subsets. D)
   Fractions of CD4^+ T subsets among tissues (liver and blood) in
   rejection samples and diagnoses (rejection and non-rejection) in liver
   samples. E) Venn plot and violin plots showing the common and different
   marker genes between two CD4^+ TRMs. F) Heatmap of the t-value for the
   area under the curve score of expression regulation by transcription
   factors, as estimated using SCENIC. G) Bar plots illustrating the
   cellular communication between CD4^+ TRMs and other cells. Red pathways
   indicate rejection, while blue pathways indicate non-rejection in
   prevalence. H) Multi-immunohistochemistry of rejection-transplanted
   liver for DAPI (blue), CD4 (red), CD69 (cyan), CD103 (white), PD1
   (purple), and CD68 (green). White arrows point to cells co-expressing
   CD4, CD69, and CD103. Whole image/insert scale bar, 100 μm.

   The CD4^+ subset showed different distribution among tissues and
   diagnoses. In rejection status, C0-PD1^+CTLA4^+ T, C2-AREG^+ TRM and
   C4-IFNG^+ TRM were predominantly found in liver, while C3-Th17 and
   Naïve T (C1-Naive_1 and C5-Naive_2) showed high proportion in blood. In
   liver tissue, C0-PD1^+CTLA4^+ T showed high proportion in rejection
   samples while C3-Th17 was downregulated in rejection. However, no
   significant differences were detected between C2-AREG^+ TRM and
   C4-IFNG^+ TRM when comparing rejection to non-rejection liver tissues
   (Figure [87]2D).

   Differential gene expression analysis between the two CD4^+ TRM
   clusters identified 52 and 36 uniquely expressed genes for C2-AREG^+
   TRM and C4-IFNG^+ TRM, respectively, with a commonality of 13 signature
   genes, including transcription factors FOSB, FOS, JUN, and NR4A1, as
   corroborated by transcription factor analysis. Compared with CD4^+ Treg
   and Naïve T cells, distinct expression patterns were noted, where
   C2-AREG^+ TRM upregulated UBE2S, AREG, and RGR1, while C4-IFNG^+ TRM
   showed heightened levels of GZMA, IFNG, GZMK, STAT1, STAT4, and CXCR3,
   indicative of a TRM1 phenotype (Figure [88]2E-F, [89]Figure S2A and
   [90]Table S2).

   Functional enrichment analysis was performed to distinguish the
   difference between the two clusters ([91]Table S2). Gene Ontology (GO)
   pathway analysis revealed overlapping pathways for both CD4^+ TRM
   clusters in myeloid cell differentiation, negative regulation of
   protein phosphorylation, and activation of the stress-activated MAPK
   cascade. Unique to C2-AREG^+ TRM were pathways involving epithelial
   cell migration, regulation of epithelial cell proliferation, and
   response to fibroblast growth factor. Conversely, C4-IFNG^+ TRM was
   associated with the extrinsic apoptotic signaling pathway, positive
   regulation of hepatocyte proliferation, and modulation of steroid
   metabolic processes, alongside enhancement of cytokine and
   interleukin-1 production ([92]Figure S2B-C, [93]Table S2).

   Cellular communication analysis indicated that pathways in rejected
   tissues predominantly involved TGFβ, NOTCH, GAS, and NECTIN signaling,
   whereas both rejection and non-rejection samples shared pathways
   related to IFN-II, CXCL, and TNF signaling (Figure [94]2G). To validate
   the role of CD4^+ TRM in human LT, we conducted mIHC on LT samples
   exhibiting rejection. Our analysis revealed that CD4^+ TRM,
   characterized by high expression of CD69 and CD103 and low expression
   of PD-1, were actively involved in various stages of rejection,
   including mild acute, severe acute, and chronic rejection.
   Additionally, these cells were observed to interact significantly with
   myeloid cells (Figure [95]2H, [96]Figure S2D).

CD8^+ TRMs Predominate in Rejected Transplanted Liver Tissues

   Our comprehensive analysis of CD8^+ T cells, facilitated by
   re-clustering within the established single-cell landscape, delineated
   10 discrete clusters. We annotated these clusters according to their
   functional attributes, identifying cytotoxic T cells (Tc),
   tissue-resident memory T cells (TRM), effector memory T cells (TEM),
   exhausted T cells (TEX), mucosal-associated invariant T cells (MAIT),
   and double negative T cells (DNT) (Figure [97]3A, [98]Table S3). Among
   the cytotoxic cohorts (Tc), five clusters were characterized—C0-XCL2^+
   Tc, C1-GZMH^+ Tc, C5-JUN^+ Tc, C6-GZMA^+ Tc, and C8-GNLY^+ Tc—all
   expressing classic cytotoxicity-associated markers such as GZMA, GZMH,
   and NKG7 ([99]Table S3). Notably, each Tc subset exhibited distinct
   cytokine profiles, with C0-XCL2^+ Tc marked by elevated cytokines
   (XCL2, XCL1, CCL4, CCL3, IFNG), C5-JUN^+ Tc by transcriptional
   regulators (JUN, FOSB, FOS, GADD45B, EGR1), and C8-GNLY^+ Tc by
   effector molecules (GNLY, FGFBP2, CX3CR1).

Figure 3.

   [100]Figure 3
   [101]Open in a new tab

   Identifying CD8^+ T Cell Subsets in Human Transplanted Liver and Blood.
   A) tSNE plot for re-clustering and cell type identification of 46,117
   CD8^+ T cells. B) tSNE plot colored by spatial distribution of cells
   among tissues (liver and blood). C) Fractions of CD8^+ TRMs among
   tissues (liver and blood) in rejection samples and diagnoses (rejection
   and non-rejection) in liver samples. D) Dot plots showing the
   expression of TRM-related genes among clusters. Violin plots
   illustrating the TRM score between CD8^+ TRMs and other CD8^+ T
   subsets. E) Violin plots showing the top DEGs of CD8^+ TRMs. Bar plots
   illustrating the GO pathway enrichment analysis for DEGs of CD8^+ TRMs.
   F) Dot plots showing the cellular communications between CD8^+ TRMs and
   Myeloid cell subsets between non-rejection (green) and rejection (red)
   samples. G) Multi-immunohistochemistry of rejection-transplanted liver
   for DAPI (blue), CD8 (orange), CD69 (cyan), CD103 (white), PD1
   (purple), and CD68 (green). White arrows point to cells co-expressing
   CD8, CD69, CD103, and PD1. Whole image/insert scale bar, 100 μm.

   The tissue distribution analysis of CD8^+ T cell subclusters revealed a
   liver dominance, with significant enrichment in rejected transplants
   (Figure [102]3B). Cluster 3, identified as CD8^+ TRM, was particularly
   overrepresented in rejected liver tissues, characterized by an
   upregulation of tissue residency (CD69, CD103, CXCR6 and CD49A),
   cytotoxic (GZMB, IFNG), immune checkpoint (PD1, CTLA4, TIGIT) and
   proliferative (PCNA) markers. To corroborate the TRM designation, we
   employed TRM scoring, confirming that C3-TRM displayed substantially
   higher scores in comparison to other clusters, denoting its specificity
   (Figure [103]3C-D).

   The top genes of CD8^+ TRMs include RGS1, TNFRSF9, TTN, ICOS, and
   DUSP4, which were partly reported to function with TRMs
   [104]^17^-[105]^19. Functional pathway analysis revealed that
   differentially expressed genes (DEGs) within CD8^+ TRMs were
   significantly involved in pathways germane to T cell activation,
   proliferation, and differentiation, cytokine production, antigen
   processing and presentation, and notably, allograft rejection (Figure
   [106]3E, [107]Figure S3). Interrogation of cellular communication
   networks elucidated that TRMs engaged in extensive dialogue with
   myeloid cells, including dendritic cells, Kupffer cells, and
   monocyte-derived macrophages. In the context of rejection, TRMs
   exhibited an augmented secretion profile of cytokines (TGFB1, IFNG,
   CCL5, CCL4, CCL3L1) and demonstrated enhanced reception of chemokines
   (CXCL12, CXCL16) from myeloid cells (Figure [108]3F). Furthermore,
   utilizing mIHC, we investigated the presence of CD8^+CD103^+/- TRMs in
   LT samples undergoing different stages of rejection, encompassing mild
   acute, severe acute, and chronic rejection. These TRMs were identified
   by their high expression levels of CD69, CD103, and PD-1. Our analysis
   also highlighted their interactions with myeloid cells within these
   various rejection contexts (Figure [109]3G, [110]Figure S3B).

CD8^+ TRM Marker Expression Distinguishes Rejection in Liver Transplantation

   In investigating the potential of TRMs as biomarkers for LT outcomes,
   we utilized bulk RNA sequencing on an independent cohort to discern the
   expression patterns associated with transplant rejection. The cohort
   included multiple liver biopsy samples from cases both with (n = 37)
   and without (n = 37) rejection. Differential expression analysis
   indicated that a majority of CD8^+ TRM markers were upregulated in
   rejection cases. This included markers for CD8^+ T cells (CD3D, CD8),
   tissue residency (CD69, CXCR6), immune checkpoints (CTLA4, TIM3, TIGIT,
   LAG3), cytotoxic (IFNG, GZMB), and proliferation (PCNA, TOP2A), along
   with costimulatory molecules CD86 and CD80 (Figure [111]4A), which were
   validated by another independent cohort (n = 7) ([112]Figure S4). These
   findings were in agreement with our single-cell analyses, confirming
   the prevalence of CD8^+ TRMs in rejected graft tissues (Figure [113]3).

Figure 4.

   [114]Figure 4
   [115]Open in a new tab

   CD8^+ TRM Marker Expression Distinguishes Rejection in Liver
   Transplantation. A) Gene expression of TRM-related markers between
   non-rejection (n = 37) and rejection (n = 37) samples using bulk
   RNA-seq. B) Immunohistochemistry of CD8, CD69, and CD103 in
   rejection-transplanted liver. Whole/insert image scale bar, 100 μm. C)
   Analysis of immune infiltration between non-rejection and rejection
   transplanted liver. D) Bar plots and dot plots illustrating the GO and
   KEGG pathway enrichment analysis for DEGs between non-rejection and
   rejection tissues.

   Contrastingly, certain markers such as CD103, CD49A, and PD1 did not
   exhibit significant differential expression between rejected and
   non-rejected samples. Concurrently, markers indicative of CD4^+ Th17
   cells, specifically CD4 and IL17A, were significantly downregulated in
   rejected samples, corroborating our earlier observations (Figure
   [116]2D). Immunohistochemical (IHC) validation suggested an inclination
   of CD8, CD69, and CD103 expressions towards rejected samples (Figure
   [117]4B).

   Using a predictive model for immune infiltration in post-transplant
   tissues, we observed that tissues from rejected transplants harbored
   significantly higher proportions of CD8^+ T cells and CD8^+ memory T
   cells. They also demonstrated elevated immune and microenvironmental
   scores compared to non-rejected tissues. However, populations of CD4^+
   T cells, naïve CD4^+ T cells, and Treg cells displayed no significant
   variance between rejected and non-rejected samples (Figure [118]4C),
   aligning with our previous single-cell analyses (Figure [119]2A-B,
   Figure [120]3A-B).

   Further differential expression analysis of genes associated with
   rejection revealed an upregulation of T cell-mediated pathways,
   encompassing the regulation of T cell activation, proliferation, and
   differentiation. Cytokine-mediated pathways were also prominent,
   particularly those governing the positive regulation of cytokine
   production and the response to interferon-gamma (Figure [121]4D,
   [122]Table S4).

Dynamic Shifts of CD8^+ TRMs in Mouse Liver Transplantation Models

   Our observations in human LT underscore the pivotal role of TRMs,
   particularly CD8^+ TRMs. To expand upon these insights, we explored the
   CD8^+ TRM profile within a mouse model of LT (Figure [123]5A). Data
   were gathered from seven time points across the LT timeline (pre-LT,
   3h, 6h, 12h, 3d, 5d, 7d post-LT), corresponding to the pre-transplant
   (LT0D), acute (LT3h, LT6h, LT12h), and stable (LT3d, LT5d, LT7d) phases
   of the mouse LT model in a tolerance setting. Our analysis yielded 30
   distinct clusters (Figure [124]5B, [125]Figure S5A-B), and we
   constructed a comprehensive cellular atlas of mouse LT, comprising
   63,049 cells, marked by canonical cell-type identifiers such as Cd3d
   for T cells, Cd79a for B cells, Lyz2 for monocytes, Adgre1 for
   macrophages, Ly6g for granulocytes, Bmp2 for endothelial cells, Acta2
   for fibroblasts, and Apoa1 for hepatocytes (Figure [126]5C-D,
   [127]Figure S5C).

Figure 5.

   [128]Figure 5
   [129]Open in a new tab

   CD8^+ TRMs Showing Dynamic Timeline Phenotypes in Mouse Liver
   Transplantation Models. A) C57BL/6 liver allografts were transplanted
   into C57BL/6 recipients, and single-cell RNA-seq analysis was performed
   using 10X genomics. B) UMAP plots showing the distribution of cells
   among dynamic timelines (pre-LT, 3h, 6h, 12h, 5d, 7d post-LT) and
   phases (pre, acute, and stable). C) Single-cell atlas of mouse LT in
   dynamic timelines, comprising different immune and stromal cells. D)
   Expression of canonical cell markers including Cd3d, Lyz2, Apoa1, Ly6g,
   Acta2, Bmp2, and Cd79a. E) tSNE plots for re-clustering and cell type
   identification of 11,480 T cells, especially CD8^+ T cells. Dot plots
   showing the expression of TRM-related genes among CD8^+ T clusters. F)
   Violin plots showing the TRM score among dynamic timelines and phases.
   G) Pseudotime analysis with identified CD8^+ proliferative T cells,
   TRM, and TEX cells. H) Heatmap of the t-value for the area under the
   curve score of expression regulation by transcription factors between
   CD8^+ TRM and TEX cells.

   A refined clustering of T cells unveiled subsets including CD8^+ Tc,
   CD8^+ TRM, CD8^+ TEM, CD8^+ TEX, CD8^+ Proliferative T cells, CD4^+ T
   cells, and DNT cells. The CD8^+ TRM cluster (c11) notably expressed
   high levels of Cd69, Cxcr6, Gzmb, Ifng, Runx3, and Jun, and low levels
   of Cd103 (Figure [130]5E, [131]Figure S5C). Temporal TRM scoring
   revealed dynamic fluctuations, with the acute phase of mouse LT
   displaying significantly elevated TRM scores compared to both
   pre-transplant and stable phases (Figure [132]5F). Pseudotime
   trajectory analysis suggested that CD8^+ TRMs occupy a middle-to-late
   stage, akin to CD8^+ TEX, whereas the CD8^+ Proliferative subset was
   positioned in the early phase, indicating a transition of effector
   CD8^+ T cells to functional TRMs, which might sustain localized
   rejection in the transplanted liver (Figure [133]5G).

   Transcription factor profiling of CD8^+ TRMs in the mouse model
   unveiled an abundance of transcripts, including Eomes, Runx3, Maf,
   Irf4, and Stat1 (Figure [134]5H). Investigating cellular communication,
   we mapped the interplay between TRMs and other tissue-resident cells
   including Kupffer cells, endothelial cells, and fibroblasts across time
   ([135]Figure S5E-F). Significantly, our analysis of cellular
   communications revealed variations in the number and intensity of
   signals between TRMs and Kupffer cells across different phases of LT.
   In the acute phase, there was an increased prevalence of signals,
   notably Pd-l1/Pd-1 interactions and CC chemokines, such as Ccl3, Ccl4,
   and Ccl5, predominantly originating from Kupffer cells. Conversely, in
   the stable phase, a higher incidence of signals involving Cxcl
   chemokines, including Cxcl4, Cxcl13, and Cxcl16, was observed
   ([136]Figure S5F). Complementing these findings, bulk RNA sequencing
   data revealed that MHC-mismatched mouse LT models exhibited higher
   expression of TRM-related genes compared to MHC-matched counterparts,
   demonstrating notable dynamic changes post-transplantation ([137]Figure
   S5G).

Single-cell Transcriptomics and Cellular Communication analysis of
CD103^-CD8^+ TRMs in the Rejection and Tolerance Phases of Mouse Liver
Transplantation

   To elucidate the roles of CD8^+ TRMs across different immunological
   outcomes in a mouse model of LT, we initially performed scRNA-seq on
   samples from mice undergoing LT with rejection (C57BL/6 to C3H/He; n =
   3), tolerance (C3H/He to C3H/He; n = 3), and from normal, healthy
   controls (C3H/He; n = 3) and constructed a immune landscape of 57,041
   cells, including T cells, B cells, myeloid cells, granulocytes,
   hepatocytes, fibroblasts, and endothelial cells, for mouse LT (Figure
   [138]6A, B). We focused on the T cell compartment, comprising 28,408
   cells, and annotated subpopulations including CD8^+ Tc, CD8^+ TRM,
   CD8^+ TEM, CD8^+ Proliferative T cells, CD4^+ T cells, and DNT cells
   (Figure [139]6C). Using a panel of established markers (Cd69, Pd1,
   Cxcr6, Gzmb, Ifng, Runx3, Jun), we identified CD8^+ TRMs, which showed
   low expression of Cd103 and were significantly more abundant in
   rejection samples compared to those from tolerance and normal states
   (Figure [140]6D-E).

Figure 6.

   [141]Figure 6
   [142]Open in a new tab

   Single-Cell Omics Analysis of CD8^+ TRMs in Rejection and Tolerance
   Phases of Mouse Liver Transplantation. A) C57BL/6J (n = 3) or C3H/He (n
   = 3) liver allografts were transplanted into C57BL/6 recipients, and
   single-cell RNA-seq analysis was performed using 10X genomics. B)
   Single-cell atlas of mouse LT in different diagnoses, comprising
   different immune and stromal cells. C) tSNE plots for re-clustering,
   cell type identification, and tissue distribution of 28,408 T cells,
   especially CD8^+ T cells. D) Dot plots showing the expression of
   TRM-related genes among CD8^+ T clusters. E) Fractions of CD8^+ TRMs
   among diagnoses (rejection, non-rejection, and normal). F) Bar plots
   and dot plots illustrating the cellular communication between CD8^+
   TRMs and Kupffer cells. Red indicating rejection and cyan indicating
   non-rejection.

   Cellular communication analysis in the rejection state highlighted an
   intensified signaling activity of TRMs. This included enhanced
   interactions via MHC-1, CD86, SN, PD-L1, CD39, and ICAM pathways.
   Notably, Kupffer cells in the rejection group exhibited robust
   signaling with CD8^+ TRMs, particularly through Cxcl16-Cxcr6,
   Cxcl10-Cxcr3, and Pdl1-Pd1 pathways (Figure [143]6F, [144]Figure S6A).
   Additionally, CD8^+ TRMs in rejection samples demonstrated increased
   secretion of molecules like Vcam1, Tigit, Ifng, and Fasl to endothelial
   cells and fibroblasts, and a higher engagement with Nectin2 and Cxcl
   pathway signals, as compared to tolerance and normal conditions
   ([145]Figure S6B).

Bulk RNA-seq and Flow Cytometry Confirm the Dominance of CD103^-CD8^+ TRMs in
Rejected Mouse Liver Transplants

   Bulk RNA-seq of the three groups provided insights at the transcript
   level (Figure [146]7A). Rejection samples revealed 2,116 DEGs when
   compared to normal, 844 DEGs when compared to tolerance, and 762 genes
   common to both sets (Figure [147]7B). The genes identified played roles
   in T cell activation, proliferation, differentiation, and the
   regulation of cytokine and interferon-gamma (Ifng) production (Figure
   [148]7C). TRM-associated markers such as Cd8a, Runx3, Gzmb, Ctla4,
   Eomes, Top2a, Pd1, and Cxcr6 were expressed at higher levels in
   rejection than in tolerance or normal samples (Figure [149]7D),
   indicating a correlation with graft outcome.

Figure 7.

   [150]Figure 7
   [151]Open in a new tab

   Bulk RNA-seq analysis of Rejected Liver Transplants. A) PCA of Bulk
   RNA-seq for Three Groups, Including Rejection, Tolerance, and Normal,
   in Mouse. B) Volcano plots and Venn plots illustrating the DEGs of
   rejection/normal and rejection/tolerance groups using bulk RNA-seq. C)
   Dot plots presenting the GO pathway enrichment analysis for DEGs of
   rejection/normal and rejection/tolerance groups. D) Heatmap displaying
   the expression of TRM-related genes among diagnoses.

   To validate the presence and prevalence of CD8^+ TRMs within our mouse
   model of LT, we employed flow cytometry to examine tissues under
   different dynamic conditions. Our results indicated that CD3^+ T cells
   were dominant in blood compared to liver and spleen, while CD8^+ T
   cells showed a higher proportion in liver compared to spleen and blood.
   Specifically, CD103^-CD8^+ TRMs were more abundant than their CD103^+
   counterparts in the livers of mice with rejected transplants (Figure
   [152]8A, [153]Figure S8).

Figure 8.

   [154]Figure 8
   [155]Open in a new tab

   Flow Cytometry and mIHC Confirm Dominance of CD103^-CD8^+ TRMs in
   Rejected Liver Transplants. A) Flow cytometry depicting the
   distribution of CD3^+, CD8^+, CD69^+CD103^+/- T cells among tissues
   (liver, spleen, and blood). B) Flow cytometry showing the dynamic
   changes of CD69^+CD103^- and PD1^+ CD69^+CD103^- T cells in liver after
   different LT timelines (1W, 2W and 3W). C) Flow cytometry showing the
   dynamic changes of CD69^+CD103^- and PD1^+CD69^+CD103^- T cells in
   spleen and blood after different LT timelines (1W, 2W and 3W). D)
   Multi-immunohistochemistry of different LT timelines (1W, 2W, 4W and
   3M) in liver for CD8 (cyan), CD69 (purple), CD103 (red) and PD1
   (yellow). Inset scale bar, 50 μm.

   We then compared one week (LT1W, n = 6), two weeks (LT2W, n = 5), and
   three weeks (LT3W, n = 4) post-transplantation in donor (C57, n = 6)
   and recipient (C3H, n = 6) mice. We found that increased T cell
   repertoire crosstalk during rejection influenced T cell subsets in the
   liver, spleen, and blood. CD3^+ T cells and CD8^+ T cells were
   significantly upregulated in the liver after LT ([156]Figure S7A).
   CD103^-CD8^+ TRMs in the liver increased with the duration of
   transplantation, and PD1 showed higher levels within CD103^-CD8^+ TRMs
   compared to normal samples (Figure [157]8B).

   In the spleen, CD3^+ T cells increased after LT while CD8^+ T cells
   decreased ([158]Figure S7B). The percentage of CD103^-CD8^+ TRMs and
   PD1^+CD103^-CD8^+ TRMs increased with the duration of transplantation
   (LT1W, LT2W, and LT3W) in the spleen (Figure [159]8C). In the blood,
   CD3^+ T cells increased at LT2W and decreased at LT3W, while CD8^+ T
   cells showed a slight increasing tendency, though not significantly
   ([160]Figure S7C). Blood samples showed a low presence of CD103^-CD8^+
   TRMs, even in LT models; however, PD1 expression was high in the
   limited CD103^-CD8^+ TRMs present (Figure [161]8C).

   Additionally, we performed mIHC for liver samples at LT1W, LT2W, LT4W,
   and LT3M to show the presence and tendency of PD1^+CD8^+ TRMs. Samples
   at four weeks and three months showed a lower presence of TRMs,
   indicating a tolerance status (Figure [162]8D, [163]Figure S7D).

Discussion

   Recent studies have underscored the significance of CD4^+/CD8^+ TRMs in
   solid organ transplantation, particularly in modulating transplant
   immunity [164]^9^, [165]^11 and influencing outcomes in various liver
   diseases [166]^20^-[167]^22. Given the liver's distinctive
   immunotolerant properties, understanding the role and mechanisms of
   TRMs in LT is crucial [168]^23. Through a robust multi-omics approach,
   our study provides an in-depth analysis of T cells in LT, emphasizing
   the complex phenotypic characteristics and significant roles of TRMs
   within the liver transplant microenvironment. We observed a pronounced
   dominance of CD8^+ TRMs with distinct features in cases of liver
   transplant rejection, along with two distinct phenotypes of CD4^+ TRMs.
   Notably, CD8^+ TRMs exhibited dynamic levels during LT and engaged in
   significant signaling, particularly with Kupffer cells, involving
   pathways like PD-L1 and various chemokines, which are instrumental in
   sustaining rejection [169]^24.

   In the liver, long-lived alloreactive CD4^+ TRMs have been documented
   to produce type 1 polyfunctional cytokine responses upon stimulation
   [170]^22^, [171]^23, yet their specific role in transplant immunity
   requires further investigation. Our study explored the maintenance of
   CD4^+ TRMs in both rejected and non-rejected liver transplants, noting
   their scarce presence in the blood. We identified two distinct subsets
   of CD4^+ TRMs: AREG^+ TRMs and IFNG^+ TRMs. These subsets shared a set
   of 13 signature genes, including transcription factors such as FOSB,
   FOS, JUN, and NR4A1 [172]^25^, [173]^26. Notably, distinct expression
   patterns emerged between these subsets. AREG^+ TRMs showed an
   upregulation of UBE2S, AREG, and RGR1, whereas IFNG^+ TRMs exhibited
   elevated levels of GZMA, IFNG, GZMK, STAT1, STAT4, and CXCR3,
   suggesting a TRM1 phenotype. Interestingly, research has shown that
   Th17-derived AREG can promote intestinal fibrotic responses,
   positioning it as a potential therapeutic target for fibrosis [174]^27.
   Additionally, CXCR3^+CD4^+ TRMs in the liver have been identified as a
   novel, functionally distinct, recirculating population that contributes
   to diverse immunosurveillance, notably marked by the expression of IFNG
   [175]^22. However, our analysis did not reveal significant differences
   between these two CD4^+ TRM subsets in the context of transplant
   diagnosis, indicating that they may not play a critical role in
   differentiating between rejection and non-rejection in liver
   transplants.

   A noteworthy revelation from our research is the predominant presence
   of CD8^+ TRMs in liver transplant rejection, emphasizing their
   distinctive transcriptional identity and dynamic nature, potentially
   playing a crucial role in rejection development [176]^14. To our best
   knowledge, it's the first time to find the significant dominance of
   CD8^+ TRMs during liver transplant rejection. However, studies found
   the CD8^+ TRMs infiltration of graft using mouse kidney transplantation
   model, but not indicated the relation of rejection status [177]^10^,
   [178]^28. We explored the plasticity of TRM cells in LT. The observed
   elevation of CD8^+ TRMs during liver transplant rejection is marked by
   a unique expression profile. This profile encompasses an upregulation
   of cytotoxic markers such as GZMB and IFNG [179]^29^, [180]^30;
   tissue-residency markers, including CD69, CD103, CXCR6 and CD49A
   [181]^31; immune checkpoint molecules like PD1, CTLA4, and TIGIT
   [182]^32 and proliferative markers PCNA and TOP2A [183]^33.
   Additionally, these cells exhibit high levels of transcription factors
   EOMES and RUNX3, further emphasizing their distinct role in transplant
   rejection scenarios. Eomes was essential to molecular and functional
   attributes of small intestine and colon CD8^+ TRMs [184]^34. Runx3 was
   known to be important to CD8^+ TRMs, rather than CD4^+ TRMs [185]^35,
   and previously found to promote cytotoxic function of skin CD8^+ TRMs
   [186]^13^, [187]^36. Pseudotime analysis showed TRMs were observed in
   the late phase, indicating a transition from effector CD8^+ T cells,
   which might sustain localized rejection in the transplanted liver.
   Results This observation aligns with the notable conservation of
   certain CTRM traits across different organ transplants, highlighting
   the specialized immune environment of the transplanted liver.

   Utilizing a mouse LT model, our study constructed a dynamic T cell
   atlas, with CD8^+ TRMs, with poor expression of Cd103, exhibiting
   higher activity in the acute phase, especially in signaling
   interactions with Kupffer cells. This finding corroborates earlier
   reports suggesting that PD-L1 blockade can rejuvenate TRM cells,
   offering potential therapeutic strategies targeting the PD-L1/PD-1
   pathways during LT rejection [188]^24. CD103^- TRMs was found as the
   primary responders, reacting to secondary infection in intestinal while
   CD103^+ TRMs had limited potential [189]^37, which indicated that
   CD103^- TRMs intend to immune regulation. We also validated the
   dominance of CD103^-CD8^+ TRMs in rejection samples with high
   expression of immune checkpoint pathways, especially the PD-L1/PD-1
   pathway, using an independent cohort. Our comprehensive use of
   immunological techniques including flow cytometry,
   immunohistochemistry, and fluorescence assays provided quantitative
   insights into TRM distribution and communication, enhancing our
   understanding of TRM behavior post-transplantation. TRMs, recognized as
   key mediators of adaptive immunity within nonlymphoid tissues, are
   characterized by their retention in these tissues without circulating
   in the bloodstream [190]^38. Consistent with this, our findings
   indicate a notably low presence of TRMs in spleen tissues and blood,
   compared with liver, for LT. Furthermore, Megan Sykes found the
   increased blood-tissue crosstalk and dynamic changes of TRMs post
   intestine transplantation patients using detailed flow cytometry and
   omics in recent studies [191]^39^, [192]^40, and we also investigated
   the blood-tissue crosstalk in LT patients.

   The translational significance of our findings is considerable. The
   study demonstrated that T cell repertoire cross talk during rejection
   can influence the T cell subsets in the blood. During the period of LT
   with rejection, both CD3^+ and CD8^+ T cells increased in blood,
   followed by the high infiltration of CD3^+ and CD8^+ T cells in
   transplanted liver. Especially, by elucidating the functional dynamics
   of TRMs, especially the CD8^+ subset, our research paves the way for
   novel strategies to monitor and potentially modulate the immune
   response in LT recipients. Targeting TRM-related pathways could offer
   new therapeutic avenues for enhancing graft acceptance and longevity.
   Additionally, the distinct behavior of TRMs in rejection scenarios
   presents opportunities for early detection and intervention,
   potentially significantly improving patient outcomes.

   Limitations of our study include the lack of further basic validation
   through in vivo and in vitro studies to elucidate the transcription
   factors, cell differentiation of TRMs, and intercellular communication
   with other tissue-resident cells. This would deepen our understanding
   of TRM development and function in transplant immunity and potentially
   guide treatment strategies targeting TRMs.

   In conclusion, this study not only enriches the current understanding
   of transplant immunology but also opens avenues for future research.
   CD8^+ TRMs have a potential function of immune regulation in LT, which
   exhibit an activated state, secret cytotoxic factors and communicate
   with other tissue resident cells, especially for rejection. The
   therapeutic manipulation of TRM responses to foster tolerance and
   reduce rejection risks, coupled with a more nuanced approach to
   immunosuppressive therapy informed by specific immune cell populations,
   could revolutionize LT management and outcomes.

Materials and Methods

Study Subjects

   This study encompassed patients who underwent liver transplantation
   (LT) at the Organ Transplantation Center, the Affiliated Hospital of
   Qingdao University, between November 2020 and November 2023. Liver
   biopsies, including a previously published cohort [193]^16, were
   conducted post-LT. All liver grafts originated from voluntary donations
   after cardiac death or living donors. The Ethics Committee of the
   Affiliated Hospital of Qingdao University (IRB number: QYFYWZLL26550)
   granted approval for the study.

Animals

   Male C57BL/6J mice and male C3H/He mice served as donors, with male
   C3H/He mice as recipients. All animals, aged 8-10 weeks (Body Weight =
   23 ± 2 g), were procured from SiPeiFu, Beijing, Biotechnology Co., LTD,
   and were housed in a specific pathogen-free (SPF) environment.
   Orthotopic liver transplantation (OLT) surgeries were performed under
   isoflurane inhalation anesthesia, following established procedures
   [194]^41.

Tissue Dissection and Cell Suspension

   Human liver biopsies and mouse liver tissues, obtained immediately
   post-surgery, were immersed in a tissue preservation solution and
   transported for standard dissociation procedures. The tissues were
   finely diced into approximately 0.5-mm³ pieces in RPMI-1640 medium
   (Invitrogen) supplemented with 1% Penicillin/Streptomycin. Enzymatic
   digestion was performed using a mix containing 0.05% trypsin
   (INVITROGEN Cat# 25200056), 0.4% collagenase IV (INVITROGEN Cat#
   17104-019), 0.25% collagenase I (SIGMA Cat# C0130-1G), 0.13%
   collagenase II (BBI Cat# A004174-0001), and 0.1% elastinase
   (WORTHINGTON Cat# [195]LS002292), maintained at 37°C for 30-45 minutes
   with constant agitation. Following digestion, the cell suspensions were
   filtered through 70-μm and 40-μm cell strainers (BD) and centrifuged at
   300g for 10 minutes. The supernatant was discarded, and the cell pellet
   was resuspended in red blood cell lysis buffer (Thermo Fisher),
   incubated on ice for 3 minutes to lyse any remaining red blood cells.
   After two washes with PBS (Invitrogen), the cells were finally
   resuspended in PBS containing 0.04% BSA for further analysis.

Library Preparation and Sequencing

   For scRNA-seq library preparation, the DNBelab C Series High-throughput
   Single-cell System (BGI-research) was utilised. This involved
   transforming single-cell suspensions into barcoded scRNA-seq libraries,
   encompassing droplet encapsulation, emulsion disruption, collection of
   mRNA-captured beads, reverse transcription, and subsequent cDNA
   amplification and purification. The resulting cDNA was fragmented into
   segments ranging from 250 to 400 base pairs. Indexed sequencing
   libraries were then constructed following the manufacturer's
   guidelines. Quality control checks were performed using the Qubit ssDNA
   Assay Kit (Thermo Fisher Scientific) and the Agilent Bioanalyzer 2100.
   Sequencing was conducted on the DIPSEQ T1 platform (China National
   GeneBank), utilizing paired-end sequencing. The sequencing read
   structure comprised a 30-base pair (bp) Read 1, including a 10-bp cell
   barcode 1, a 10-bp cell barcode 2, and a 10-bp unique molecular
   identifier (UMI). This was followed by a 100-bp Read 2 for gene
   sequencing and a 10-bp barcode read for sample indexing.

Single-Cell RNA Sequencing Data Pre-Processing

   The sequencing data underwent comprehensive processing using an
   open-source pipeline available on GitHub
   ([196]https://github.com/MGI-tech-bioinformatics/DNBelab_C_Series_HT_sc
   RNA-analysis-software). The process initiated with sample
   de-multiplexing, barcode processing, and the counting of single-cell 3'
   unique molecular identifiers (UMIs) using the default parameters of the
   pipeline. The resulting reads were aligned to the GRCh38 human genome
   reference using STAR software (version 2.5.3). Cell validity was
   ascertained based on UMI distribution per cell, employing the
   barcodeRanks() function from the DropletUtils tool. This facilitated
   the removal of background noise and beads with UMI counts below a
   specified threshold. For the final step, we employed the PISA framework
   to quantify gene expression in cells, generating a gene-by-cell matrix
   for each library.

Single-Cell Gene Expression Quantification and Subcluster Delineation

   The analysis of our single-cell RNA sequencing data utilized the Seurat
   R package (version 4.3.0) [197]^42 to manage various stages of data
   processing. This began with raw data importation, followed by stringent
   quality control measures. Low-quality cells, defined as those with
   fewer than 501 expressed genes and exhibiting more than 25%
   mitochondrial counts, were filtered out. High-quality cells passing
   these filters were then normalized and scaled using Seurat's default
   settings.

   Next, we identified highly variable features within our dataset using
   the FindVariableFeatures function. This step was crucial for the
   subsequent principal component analysis (PCA) on the scaled data,
   focusing on these variable features. We proceeded with dimension
   reduction and clustering of the data, employing the FindNeighbors (with
   dimensions set to 1:10) and FindClusters (resolution set at 0.5)
   functions within Seurat. The final step in our analytical process was
   the application of uniform manifold approximation and projection
   (UMAP). This technique was instrumental in further exploring and
   visualizing intricate patterns within our dataset.

Cell Type Determination

   To identify differentially expressed features across each cell cluster,
   we utilized the FindMarkers and FindAllMarkers functions. Cell types
   were annotated based on known biological types, using a set of
   canonical marker genes ([198]Table S1). This annotation process was
   guided by the CellMarker database
   ([199]http://bio-bigdata.hrbmu.edu.cn/CellMarker/) and corroborated by
   information from published articles. Additionally, for more robust cell
   type identification, we employed the SingleR package (version 2.0.0)
   [200]^43. This package further aided in the accurate determination of
   cell types by aligning our single-cell RNA sequencing data with
   reference datasets.

Functional Enrichment Analysis and TRM Scores

   Upon completing the annotation of each cell type, we undertook a
   functional enrichment analysis of differentially expressed genes across
   various cell clusters. This analysis, essential for elucidating the
   biological processes and potential functions of distinct cell types,
   employed Gene Ontology (GO) and KEGG (Kyoto Encyclopedia of Genes and
   Genomes) pathways. To conduct this analysis, we utilized the
   clusterProfiler package (version 3.17.0) [201]^44 along with the
   org.Hs.eg.db package (version 3.11.4). The p-value cutoffs for both GO
   and KEGG analyses were set at 0.05 to ensure the significance of our
   findings. The top ten terms derived from these results were visually
   represented in the form of barplots or dotplots, offering a clear and
   concise graphical representation of the key functional attributes
   associated with each cell type.

   For TRM scores with gene set analysis, we utilized the 'AddModuleScore'
   function within the Seurat package, based on the TRM related markers
   [202]^45. This approach involved the application of specific gene sets
   of interest, which were sourced from previously published studies or
   datasets. We then computed a score for each cell, derived from the
   expression levels of genes within each gene set, thereby enabling a
   detailed cell-by-cell analysis of gene expression patterns.

Pseudotime Analysis

   We performed trajectory analysis using the Monocle package (version
   2.28.0) [203]^46, allowing us to track the developmental progression of
   cells in a pseudo-temporal order. This analysis encompassed various T
   subsets, including TRM, TEX, and naïve T cells. Specific parameters
   were set for each cell group: a lower detection limit of 0.5, a minimum
   expression threshold of 0.1, and a requirement for the gene to be
   expressed in at least 10 cells. Visualization of cell trajectories was
   achieved using the plot_cell_trajectory function, mapping out potential
   developmental pathways of cells based on pseudotime, Seurat cluster
   assignments, and additional metadata. This provided a dynamic view of
   cellular differentiation and maturation processes.

Cell-Cell Communication Analysis

   In our investigation into intercellular interactions, we employed the
   CellChat package (version 1.6.1) [204]^47, focusing on ligand-receptor
   interactions based on the KEGG signaling pathway database and recent
   experimental studies. The analysis comprised several crucial steps,
   starting with the identification of differentially expressed signaling
   genes to highlight those with significant expression variations. This
   was followed by calculating the ensemble average expression, providing
   a comprehensive view of gene expression patterns across different cell
   types. The final step involved assessing the probability of
   intercellular communication, pivotal in understanding the complex
   signaling networks and interactions among various cell populations in
   our study.

Bulk RNA Sequencing Analysis

   To validate our findings, we sourced transcriptome data from the NCBI
   Gene Expression Omnibus (GEO). The specific datasets accessed were
   [205]GSE145780 [206]^48 and [207]GSE203453 [208]^49, encompassing gene
   expression profiles obtained from microarrays of liver transplant
   biopsies. These biopsies included samples from both rejection and
   non-rejection scenarios, providing a comprehensive view of gene
   expression alterations associated with these different transplant
   outcomes.

Immune Infiltration Estimation

   To delineate the intricate immune cell infiltration within tissue
   samples, we employed the CIBERSORT algorithm [209]^50. Developed by
   Newman et al. from the Alizadeh Lab, this method is esteemed for its
   efficacy in decoding complex immune landscapes based on gene expression
   profiles. CIBERSORT's robust computational approach allows for a
   detailed and accurate estimation of the immune cell composition within
   our tissue samples, providing valuable insights into the immunological
   milieu of LT.

Histological Staining

   Fresh liver tissue samples were initially fixed in 4% paraformaldehyde
   and subsequently embedded in paraffin. These samples were then
   sectioned into 4 μm thick slices. For morphological assessment, the
   sections underwent staining using hematoxylin and eosin (H&E).
   Additionally, reticular fiber staining was performed to evaluate tissue
   architecture, and Masson's trichrome staining was applied to
   specifically detect and highlight areas of fibrosis within the liver
   tissues.

Immunohistochemical (IHC) Staining Analysis

   IHC staining was conducted on liver tissue sections to target specific
   antigens, including CD3, CD4, CD8A, CD69, CD103, and PD1. Initially,
   the tissue slides underwent deparaffinization and dehydration.
   Following this, they were immersed in Tris-EDTA antigen retrieval
   buffer at pH 6.0 and pH 9.0 and subjected to heat-induced antigen
   retrieval. To neutralize endogenous peroxidases and block nonspecific
   antigens, 3% hydrogen peroxide (H2O2) and 3% bovine serum albumin (BSA)
   were applied, respectively. Subsequent to these preparatory steps, the
   slides were sequentially incubated with primary antibodies: anti-CD3,
   anti-CD4, anti-CD8A, anti-CD69, anti-CD103, and anti-PD1. For signal
   detection, a TSA kit from Nanjing Freethinking Biotechnology Co., Ltd.
   (China) was utilized. Nuclei counterstaining was performed using DAPI.
   Finally, the stained slides were scanned using a Pannoramic MIDI slice
   scanner (3Dhistech, Hungary), and image analysis was carried out
   employing the HALO 2.0 Area Quantification algorithm (Indica Labs;
   Corrales, NM) at Nanjing Freethinking Biotechnology Co., Ltd. (China).

Flow Cytometry

   For flow cytometric analysis, cells were stained with a panel of
   fluorochrome-conjugated antibodies. This panel included anti-CD3, CD4,
   CD8A, CD69, CD103 and PD1, all sourced from BioLegend. Staining
   procedures were carried out in strict accordance with the
   manufacturer's protocols. Following staining, the samples were
   subjected to analysis using FlowJo software (verision 10.8.1), enabling
   detailed examination and quantification of various cell populations
   based on their fluorescent properties and antibody binding
   characteristics.

Statistical analysis

   Statistical analysis and graphical representation of the data were
   performed with R software (version 4.2.3). The data are presented as
   mean ± standard error of the mean (SEM). An unpaired student t test was
   used to determine the statistical significance between two groups.
   Statistical significance was indicated as *P < 0.05, ** P < 0.01, *** P
   < 0.001.

Supplementary Material

   Supplementary figures.
   [210]thnov14p4844s1.pdf^ (6.4MB, pdf)

   Supplementary table 1 meta data.
   [211]thnov14p4844s2.xlsx^ (12.7KB, xlsx)

   Supplementary table 2 human CD4T.
   [212]thnov14p4844s3.xlsx^ (110.8KB, xlsx)

   Supplementary table 3 human CD8T.
   [213]thnov14p4844s4.xlsx^ (75.5KB, xlsx)

   Supplementary table 4 GO KEGG.
   [214]thnov14p4844s5.xlsx^ (103.7KB, xlsx)

Acknowledgments