Abstract

   Liver metastasis is associated with poor prognosis and resistance to
   immune checkpoint inhibitors. Functional modulation of Kupffer cells
   (KCs) holds promise as an alternative immunotherapeutic approach.
   Leveraging their capacity to capture circulating virions, an oncolytic
   virus‐based KC‐targeting strategy is developed that demonstrated
   efficacy and safety in treating multifocal liver metastasis. A single
   intravenous infusion of the M51R mutant vesicular stomatitis virus
   (VSV‐M51R), but not wild‐type (WT) VSV, induced significant tumor
   regression in mouse models of forced liver metastasis, independent of
   direct oncolysis. The ineffectiveness of VSV‐WT is attributed to its
   induction of massive KC apoptosis, whereas VSV‐M51R replicated
   transiently within KCs without compromising viability. Instead,
   VSV‐M51R promoted KC proliferation in tumor‐adjacent areas, enhancing
   their access to tumor foci and cross‐presentation of tumor antigens.
   This led to robust activation of hepatic anti‐tumor CD8^+ T‐cell
   responses, which required mitochondrial antiviral signaling
   protein‐dependent type I interferon triggering in KCs. Depletion of KCs
   abolished the T cell stimulating and anti‐tumor effects of VSV‐M51R.
   Furthermore, simultaneous blockade of programmed cell death‐ligand
   1(PD‐L1) during VSV‐M51R treatment achieved remarkable synergistic
   efficacy in treating monotherapy‐resistant late‐stage liver metastasis.
   These findings underscore the pivotal role of KCs in systemic oncolytic
   virotherapy and offer a potentially applicable strategy for treating
   advanced liver metastasis.

   Keywords: cross‐presentation, intravital microscopy, Kupffer cell,
   liver metastasis, oncolytic virus, type I IFN, VSV
     __________________________________________________________________

   Targeting Kupffer cells (KC) with a single intravenous infusion of the
   oncolytic virus VSV‐M51R rather than VSV‐WT induces effective tumor
   regression in various types of liver metastatic cancers in mice.
   VSV‐M51R promotes KC proliferation and enhances their antigen
   cross‐presentation capacity without compromising viability, leading to
   the induction of robust tumor antigen‐specific CD8^+ T cell responses
   that can be further potentiated by anti‐PDL1 treatment.

   graphic file with name ADVS-12-e04929-g002.jpg

1. Introduction

   Approximately 90% of cancer‐related deaths are attributed to
   metastasis, a process in which cancer cells spread from primary tumors
   and prosper in distal organs. The liver serves as the primary site for
   the seeding and growth of hematogenously disseminated cancer cells,
   particularly those originating from the gastrointestinal tract. Liver
   metastasis is generally associated with poor prognosis and typically
   manifests as multifocal, making surgical resection difficult.^[ [40]^1
   , [41]^2 ^] Owing to the inherent tolerogenic properties of the liver,
   liver metastases are also resistant to T‐cell‐based immune checkpoint
   inhibitors,^[ [42]^3 , [43]^4 , [44]^5 ^] underscoring the urgent need
   to develop alternative immunotherapeutic approaches.

   Kupffer cells (KCs) are the most abundant immune cells in the liver.
   These embryonically derived macrophages reside intravascularly to
   ensure prompt immunosurveillance against blood‐borne particles,
   including circulating tumor cells.^[ [45]^6 , [46]^7 , [47]^8 ^] In
   addition to directly phagocytosing metastatic tumor cells via C‐type
   lectin receptors and Fc receptors,^[ [48]^9 , [49]^10 , [50]^11 ^] KCs
   also orchestrate early antitumor innate immune responses, collectively
   restricting the growth of liver‐seeded metastatic tumors.^[ [51]^12 ,
   [52]^13 ^] However, these antitumor functions of KCs are compromised as
   metastases progress, likely due to their reduced accessibility to tumor
   foci or repolarization toward an anti‐inflammatory state.^[ [53]^14 ,
   [54]^15 , [55]^16 ^] The functional reprogramming of KCs has emerged as
   an encouraging immunotherapeutic strategy for treating liver
   malignancies. Our previous study demonstrated that targeted disruption
   of MafB and cMaf expression in KCs could expand peritumoral KCs and
   unleash their tumoricidal capacity, leading to durable T‐cell immunity
   against liver metastasis.^[ [56]^15 ^] Similarly, recent reports
   indicate that immune stimuli, such as IFN‐α or β‐glucan, can target KCs
   to induce T‐cell‐dependent tumor regression.^[ [57]^17 , [58]^18 ^]
   These findings suggest a role of KCs in reinvigorating antitumor T‐cell
   immunity, which underlies the efficient control of liver metastases.
   Although KCs are typically considered tolerogenic antigen‐presenting
   cells (APCs) that favor the induction of T‐cell tolerance,^[ [59]^19 ^]
   efficient T‐cell priming by KC‐mediated cross‐presentation has been
   observed during acute viral infections,^[ [60]^20 ^] suggesting the
   possibility of harnessing viral stimulation to subvert the tolerogenic
   properties of KCs and provokes effective T‐cell immunity against liver
   metastasis.

   Oncolytic viruses (OVs) offer a promising alternative strategy in
   cancer therapies because of their inherent ability to selectively kill
   tumor cells. Local delivery of OVs results in immunogenic cell death in
   tumor cells, yielding significant therapeutic benefits in cancer
   patients.^[ [61]^21 , [62]^22 ^] However, intratumoral administration
   of OVs is technically challenging and less effective for treating
   disseminated metastatic cancers, such as liver metastases, which are
   often multifocal and contain some tumor lesions that are too small for
   injection. While systemic administration of OVs appears to be a more
   rational approach for metastatic cancers, exposure to the bloodstream
   inevitably increases the risk of viral sequestration and inactivation
   by the immune system, leading to reduced viral entry into tumor foci
   and impaired oncolysis.^[ [63]^23 ^] It was recently reported that
   infection of noncancerous cells is also effective in promoting
   T‐cell‐dependent tumor regression egression following intravenous OV
   delivery, although the exact cell type that triggers antitumor immunity
   in response to OVs remains elusive.^[ [64]^24 ^] In fact, KCs are the
   predominant phagocytes capable of taking up the majority of circulating
   viral particles.^[ [65]^25 , [66]^26 , [67]^27 ^] Although this
   filtering function potentially impedes the direct oncolytic effects of
   OVs, the selective entrapment of OVs in KCs could be leveraged to spur
   the antitumor properties of these immune sentinels, leading to
   effective control of metastatic tumors independent of oncolysis.
   However, given that KCs are prone to undergo rapid cell death during
   various types of viral infections,^[ [68]^26 , [69]^28 , [70]^29 ^] a
   suitable viral platform that skews KCs into immunostimulatory
   macrophages while maintaining their integrity is necessary.

   In this study, we tested the hypothesis of diverting OVs to target and
   functionally reprogram KCs for the treatment of multifocal liver
   metastasis using vesicular stomatitis virus (VSV). VSV is a
   prototypical negative‐strand RNA virus and represents one of the most
   frequently used OVs in both preclinical models and clinical trials.
   Owing to the low efficiency of VSV in infecting human cells and its
   high sensitivity to type I interferons (IFN‐α/β), it is relatively
   tolerable when applied systemically.^[ [71]^30 , [72]^31 ^] More
   importantly, VSV can induce strong CD8^+ T‐cell responses by activating
   APCs; thus, it is widely engineered as a viral vector for
   vaccination.^[ [73]^32 ^] This renders VSV an attractive candidate in
   our study, aiming at eliciting T‐cell immunity against liver metastasis
   by targeting KCs, which are the largest APC population in the liver.

2. Results

2.1. VSV‐M51R Efficiently Controls Liver Metastasis Independent of Direct
Oncolytic Activity

   To evaluate the therapeutic potential of systemic VSV delivery in
   treating multifocal liver metastases, we employed a well‐established
   mouse model of forced liver metastasis through intrasplenic injection
   of highly metastatic B16F10 melanoma cells, which subsequently seed the
   liver and grow clonally to form multiple metastatic foci. Intravenous
   administration of 3 × 10^8 PFU of wild‐type (WT) VSV at 7 days
   post‐tumor inoculation, when macrometastases had developed in the
   liver,^[ [74]^15 ^] demonstrated only a marginal therapeutic effect. No
   significant difference in hepatic metastatic tumor burden was observed
   between the treated and untreated groups (Figure [75]1A,B).

Figure 1.

   Figure 1
   [76]Open in a new tab

   VSV‐M51R efficiently controls liver metastasis independent of its
   oncolytic activity. A) Mice were treated with 3×10^8 PFU of VSV on day
   7 following B16F10 tumor cell inoculation and were harvested on day 14.
   B) Quantification of liver weight, the liver‐to‐body weight ratio and
   the total area of metastatic foci on the liver surface (tumor area) in
   A. Data from 8–12 mice per group. C) Mice were treated with 3×10^8 PFU
   of VSV‐M51R on day 7 following B16F10 tumor cell inoculation and were
   harvested on day 14. D) Quantification of liver weight, the
   liver‐to‐body weight ratio and the tumor area in C. Data from 14 mice
   per group. E) Mouse survival rates were monitored. Data from 10 mice
   per group. F) Mice were treated with VSV‐M51R on day 7 after MC38 tumor
   inoculation and harvested on day 14. G) Mice were treated with VSV‐M51R
   on day 10 after Pan02 tumor inoculation and harvested on day 21. H)
   Quantification of liver weight, the liver‐to‐body weight ratio and the
   tumor area in F. Data from 4 mice per group. I) Quantification of liver
   weight, the liver‐to‐body weight ratio and the tumor area in G. Data
   from 5 mice per group. J) Two‐photon intravital images of liver
   metastases from B16F10‐tdTomato tumor‐bearing mice at 12 h and at days
   1, 3, and 5 after VSV‐M51R treatment. The X‐Z (bottom) and Y‐Z (right)
   views for each panel are shown. Scale bar, 100 µm. K) The VSV‐M51R N
   mRNA copies were quantified by qPCR analysis at 12 h and at days 1, 3,
   and 5 after VSV‐M51R injection. Data from 3–4 mice per time point. Data
   are represented as mean ± SEM. ^* p < 0.05; ^** p < 0.01; ^*** p <
   0.001; ns, no significance. Unpaired Student's t test for B, D, H, and
   I, one‐way ANOVA with Tukey's test for K, and two‐sided log‐rank test
   for E.

   We initially hypothesized that the limited therapeutic efficacy of
   VSV‐WT was due to its insufficient immune stimulatory capacity. The
   M51R mutant VSV (VSV‐M51R) is a naturally occurring variant in which
   the 51th amino acid (methionine) of the viral matrix (M) protein is
   replaced with arginine. This mutation abolishes M protein‐mediated
   inhibition of host protein synthesis, thereby greatly enhancing the
   immune‐stimulatory properties of the virus.^[ [77]^33 ^] In contrast to
   VSV‐WT, a single intravenous infusion of 3 × 10^8 PFU of VSV‐M51R
   significantly retarded the growth of metastatic melanoma in the liver
   (Figure [78]1C,D). Direct comparison between VSV‐WT and VSV‐M51R
   treatment in the same cohorts of tumor‐bearing mice confirmed the
   superior therapeutic outcomes of VSV‐M51R, as evidenced by substantial
   reductions in both the number and size of hepatic metastatic tumors
   (Figure [79]S1A,B, Supporting Information). Importantly, while
   untreated mice succumbed to metastatic tumor progression within 22
   days, mice treated with VSV‐M51R exhibited prolonged survival,
   extending beyond 30 days. In contrast, VSV‐WT treatment did not confer
   a significant survival advantage (Figure [80]1E). Similarly, systemic
   injection of VSV‐M51R significantly reduced the hepatic tumor burden of
   metastatic MC38 colorectal cancers and Pan02 pancreatic cancers
   (Figure [81]1F–I).

   To interrogate whether the observed therapeutic effects of VSV strains
   depend on their direct oncolytic activity, we monitored viral
   replication within metastatic tumors at various time points following
   treatment with recombinant VSV‐WT or VSV‐M51R, both of which encode
   green fluorescent protein (GFP). While active viral replication, as
   indicated by GFP expression, was readily observed in VSV‐infected
   B16F10 melanoma cells in vitro (Figure [82]S1C, Supporting
   Information), no discernable GFP signal was detected within metastatic
   tumors at days 1, 3, and 5 after treatment with either VSV‐WT or
   VSV‐M51R, as assessed by two‐photon intravital microscopy
   (Figure [83]1H; Figure [84]S1D, Supporting Information). Consistently,
   the mRNA copy number of the VSV nucleocapsid (N) gene in tumor tissues
   was comparable to that in liver tissues from uninfected mice
   (Figure [85]1I; Figure [86]S1E, Supporting Information), further
   confirming the absence of viral replication in metastatic tumors during
   the course of viral treatment. Notably, although metastatic Pan02
   tumors responded to VSV‐M51R virotherapy, these cells were refractory
   to VSV‐M51R infection in vitro (Figure [87]S1F, Supporting
   Information), which is consistent with a previous report showing that
   Pan02 cells are resistant to VSV‐ΔM51‐mediated cell lysis,^[ [88]^34 ^]
   suggesting that virus‐mediated direct oncolysis contributes minimally
   to tumor regression. Furthermore, surface exposure of calreticulin
   (ecto‐CRT), a hallmark of immunogenic cell death typically associated
   with virus‐induced oncolysis of tumor cells,^[ [89]^35 ^] was observed
   in VSV‐M51R‐infected B16F10 cells in vitro but not in metastatic tumor
   cells isolated from VSV‐M51R‐treated mice (Figure [90]S1G–I, Supporting
   Information). Taken together, these data suggest that systemic delivery
   of VSV‐M51R effectively controls liver metastasis through mechanisms
   independent of tumor cell infection and direct oncolysis.

2.2. KC Resistance to Apoptosis Dictates the Therapeutic Efficacy of VSV
Virotherapy

   The absence of VSV replication in metastatic liver tumors is consistent
   with previous studies showing that KCs capture and sequester most
   circulating viral particles,^[ [91]^26 ^] leading us to hypothesize
   that systemically delivered VSV‐M51R may primarily exert its
   therapeutic effects by targeting KCs. To test this hypothesis, we first
   tracked hepatic viral distribution by visualizing GFP expression
   following intravenous injection of VSV‐M51R‐GFP. Intravital microscopy
   and 3D reconstruction revealed that bright GFP expression was
   predominantly observed in KCs (Figure [92]2A). KC expression of virally
   encoded GFP peaked at 12 h post infection and declined rapidly
   thereafter, becoming nearly undetectable after 48 h (Figure [93]2A–C).
   Similar results were observed in tumor‐bearing mice, where VSV‐M51R
   replication was largely restricted to KCs (Figure [94]S2A, Supporting
   Information). Notably, the number of Tim‐4^+ resident KCs remained
   unchanged during the course of VSV‐M51R infection (Figure [95]2A,D).
   Fate mapping of Clec4f‐creER: R26‐LSL‐tdTomato mice, which enabled
   permanent labeling of resident KCs upon tamoxifen (TAM) injection,
   further confirmed that the resident KCs were intact during VSV‐M51R
   challenge (Figure [96]2E,F). These data suggest that, upon systemic
   administration, VSV‐M51R primarily infects KCs in the liver, where the
   virus undergoes transient replication before being rapidly cleared
   without compromising the integrity of the resident KC pool.

Figure 2.

   Figure 2
   [97]Open in a new tab

   KC resistance to apoptosis dictates the therapeutic efficacy of
   virotherapy. A) Intravital liver images showing hepatic viral
   replication at the indicated time points after VSV‐M51R‐GFP treatment.
   Scale bars: top panel, 100 µm; middle panel: 10 µm; bottom panel: 5 µm.
   B) Quantification of the ratio of infected KCs per field of view (FOV)
   in A. C) Quantification of the composition of GFP^+ cells per FOV at 12
   h in A. D) Quantification of KC numbers per FOV in A. E) Intravital
   images showing resident KCs at 0 and 96 h post‐VSV‐M51R treatment in
   TAM‐pulsed Clec4f‐creER: R26‐LSL‐tdTomato mice. Scale bar, 100 µm. F)
   Quantification of the number of resident KCs per FOV in E. G)
   Intravital images showing Tim4‐labeled KCs at the indicated time points
   after VSV‐WT‐GFP infection. Scale bar, 100 µm. H) Quantification of KC
   numbers per FOV in G. I) Representative flow cytometric plot of KCs at
   12 h after VSV‐WT and VSV‐M51R infection. J) Quantification of TIM4^+
   resident KCs. Data from 3–4 mice per group. K) Representative
   intravital images of PI and Annexin V staining 6 h post‐VSV‐WT
   infection. Scale bar, 100 µm. L) Time‐lapse intravital imaging of KCs
   starting at 6 h post VSV‐WT infection in TAM‐pulsed Clec4f‐creER:
   R26‐LSL‐tdTomato mice. Scale bar, 10 µm. M) Experimental design for N
   and O. Clec4f‐iDTR and control C57BL/6 mice were treated with DT or
   VSV‐M51R on day 7 following B16F10 tumor inoculation. N) Representative
   liver photos were taken on day 14. O) Liver weight, the ratio of liver
   weight to body weight and the tumor area were quantified. Data from 6
   mice per group. Data are represented as mean ± SEM. ^* p < 0.05; ^** p
   < 0.01; ^*** p < 0.001; ns, no significance. Unpaired Student's t test
   for F and O and one‐way ANOVA with Tukey's test for B, D, H, and J. For
   B, D, F, and H, each dot represents one FOV, with FOVs randomly
   selected from 3–4 mice per group.

   In sharp contrast to VSV‐M51R treatment, we observed a drastic
   reduction in Tim‐4^+ KCs, peaking at 12 h after intravenous injection
   of VSV‐WT (Figure [98]2G,H), a time point at which VSV‐M51R displayed
   the highest replication within KCs. Flow cytometric analysis confirmed
   a significant decrease in F4/80^+TIM4^+ KCs after VSV‐WT treatment
   (Figure [99]2I,J), with their numbers not being restored until day 7
   post infection (Figure [100]S2B,C, Supporting Information). To
   understand the mechanism of VSV‐WT‐induced KC loss, we applied Annexin
   V and propidium iodide (PI) staining in vivo to determine whether KC
   loss was due to apoptosis or necrosis. The majority of the remaining
   KCs at 6 h after VSV‐WT injection were stained with Annexin V rather
   than PI (Figure [101]2K; Figure [102]S2D, Supporting Information).
   Time‐lapse intravital imaging revealed that Annexin V^+ apoptotic KCs
   frequently underwent fragmentation, and the resulting cell debris was
   rapidly flushed away from the bloodstream, leading to KC loss
   (Figure [103]2L). Consistently, pretreatment with ZVAD‐FMK, a
   pan‐apoptosis inhibitor, completely prevented KC loss, whereas the
   necrosis inhibitor Necrostatin‐1 had no effect. Furthermore, the
   caspase‐9 inhibitor Z‐LEHD‐FMK, which selectively blocks the intrinsic
   apoptotic pathway,^[ [104]^36 ^] inhibited KC loss (Figure [105]S2E,F,
   Supporting Information). Overall, these findings suggest that VSV‐WT
   infection induces massive KC death by activating the intrinsic
   apoptotic pathway.

   The correlation between KC loss and limited therapeutic outcomes during
   VSV‐WT treatment suggested that KCs are indispensable for the
   virus‐mediated control of liver metastasis. To test this hypothesis, we
   administered diphtheria toxin (DT) into tumor‐bearing Clec4f‐iDTR mice
   concurrently with VSV‐M51R injection (Figure [106]2M). This led to the
   ablation of resident KCs after they had internalized the injected viral
   particles, thereby preventing potential morbidity and mortality
   associated with uncontrolled viral dissemination in the absence of KCs.
   Indeed, the antitumor effect of VSV‐M51R was completely abolished by KC
   depletion (Figure [107]2N,O; Figure [108]S2G,H, Supporting
   Information). We next investigated whether preventing KC apoptosis
   could restore the therapeutic efficacy of VSV‐WT. For this purpose, we
   exploited our previously reported recombinant bacteria‐based BIL‐CRISPR
   approach^[ [109]^15 ^] to simultaneously disrupt the expression of Bak
   and Bax in KCs in vivo, which are core components that activate the
   intrinsic apoptotic pathway.^[ [110]^36 ^] As expected, disruption of
   Bak and Bax prevented KC loss and enhanced the antitumor effects of
   VSV‐WT treatment (Figure [111]S2I–K, Supporting Information).
   Furthermore, we generated Clec4f‐cre: R26‐LSL‐Bcl2 mice, in which the
   anti‐apoptotic protein Bcl2 was selectively overexpressed in KCs. Bcl2
   overexpression markedly inhibited KC loss (Figure [112]S2L,M,
   Supporting Information) and significantly improved the therapeutic
   efficacy of VSV‐WT (Figure [113]S2N, Supporting Information).
   Collectively, these data highlight the critical role of KCs in
   virus‐mediated antitumor effects, and demonstrate that KC resistance to
   apoptosis dictates the therapeutic efficacy of VSV virotherapy against
   liver metastasis.

2.3. Peritumoral KCs Proliferate and take up Tumor Antigens Following
VSV‐M51R Treatment

   To explore the mechanisms by which VSV‐M51R induces KC‐dependent
   antitumor effects in liver metastases, we performed bulk RNA sequencing
   (RNA‐seq) on KCs isolated from metastatic B16F10 melanoma‐bearing mice
   that were either untreated or treated with VSV‐M51R for 3 days. By this
   time point, VSV‐M51R had been completely cleared from KCs, and any
   potential interference from the viral genome and acute antiviral
   responses had subsided, enabling a more accurate assessment of the
   impact of the virus on KCs. The RNA‐seq data revealed a total of 574
   differentially expressed genes (DEGs) between KCs from untreated and
   VSV‐M51R‐treated tumor‐bearing mice, with 377 genes upregulated and 195
   genes downregulated after VSV‐M51R treatment (Figure [114]3A).
   Intriguingly, Gene Ontology (GO) analysis revealed that pathways
   related to DNA replication and cell division were most prominently
   enriched (Figure [115]3B). Notably, common proliferation marker
   genes,^[ [116]^37 ^] including Mki67, Top2a, Stmn1, Cdk1, Pcna, Bub1,
   Plk1, and Foxm1, were upregulated in KCs from VSV‐M51R‐treated mice
   (Figure [117]3C). Additionally, genes associated with cyclins, the
   minichromosome maintenance (MCM) complex and the cell division cycle
   (CDC) family—all of which are essential for cell cycle progression—were
   prominently upregulated (Figure [118]S3A–C, Supporting Information).
   Moreover, the expression of the transcription factors Mafb (encoded by
   mafb) and c‐Maf (encoded by maf), which are known to suppress the
   proliferative potential of macrophages, including KCs,^[ [119]^15 ,
   [120]^38 ^] was reduced (Figure [121]3C). These data strongly suggest
   that VSV‐M51R treatment induces active cell proliferation in KCs. In
   line with these findings, we observed an approximately twofold increase
   in the overall number of KCs from day 3 to day 7 after VSV‐M51R
   treatment (Figure [122]3D,E).

Figure 3.

   Figure 3
   [123]Open in a new tab

   Peritumoral KCs proliferate and take up tumor antigens following
   VSV‐M51R treatment. A) Volcano plot showing DEGs between KCs from
   control and VSV‐M51R‐treated tumor‐bearing mice. B) GO functional
   enrichment analysis. C) Heatmaps of upregulated genes related to cell
   proliferation. C, untreated control; M, VSV‐M51R‐treated. D)
   Representative flow cytometric plot of KCs at the indicated time points
   after VSV‐M51R treatment. E) Quantification of the KC number in D. Data
   from 8–13 mice. F) KC localization in proximity to liver metastases at
   the indicated time points after control or VSV‐M51R treatment. G)
   Quantification of the number of KCs containing zsGreen^+ particles per
   FOV in F. H) Quantification of KC density in the peritumoral region
   (0–200 µm away from the tumor margin) or I) distant tumor areas (600 µm
   away from the tumor margin) in F. J) Representative intravital images
   showing binucleated KCs in B16F10‐ZsGreen tumor‐bearing
   Clec4f‐ntdTomato mice on day 3 after VSV‐M51R treatment. Scale bars,
   left panel, 100 µm; middle panel, 10 µm; right panel, 5 µm. K)
   Quantification of the number (left) and ratio (right) of binucleated
   KCs in the tumor‐adjacent area. Data are represented as mean ± SEM. ^*
   p < 0.05; ^** p < 0.01; ^*** p < 0.001; ns, no significance; unpaired
   Student's t test for G, H, I, and K; one‐way ANOVA with Tukey's test
   for E. Each dot of G, H, I, and K represents an FOV, with 8–18 FOVs
   randomly selected from 3–4 mice per group.

   Our previous study revealed that disrupting c‐Maf and Mafb expression
   in KCs led to their preferential expansion in peritumoral areas.^[
   [124]^15 ^] The downregulation of these two transcription factors
   suggests that KC proliferation in response to VSV‐M51R treatment may
   follow a similar spatial pattern. Indeed, in vivo imaging revealed a
   progressive increase in the abundance of peritumoral KCs from day 1 to
   day 7 post virotherapy (Figure [125]3F,H), whereas the KC density in
   tumor‐distant areas was unchanged (Figure [126]3I). Accordingly, using
   Clec4f‐ntdTomato mice in which tdTomato is specifically expressed in
   the nucleus of KCs, we observed an increased number of dividing KCs in
   tumor‐adjacent areas on day 3 after VSV‐M51R treatment
   (Figure [127]3J,K). Furthermore, while Tim‐4^+ resident KCs in
   untreated mice barely infiltrated into tumor foci, the increased
   abundance of peritumoral KCs in VSV‐M51R‐treated mice led to increased
   tumor infiltration of these macrophages (Figure [128]3F). 3D
   reconstruction further revealed that tumor‐infiltrating KCs formed
   intimate interactions with liver metastases and actively took up
   material from ZsGreen‐tagged metastatic melanoma cells
   (Figure [129]3F,G). Taken together, these data suggest that VSV‐M51R
   treatment induces KC proliferation in peritumoral areas, likely
   enhancing their capacity to access tumor antigens.

2.4. VSV‐M51R Treatment Augments KC Cross‐Presentation of Tumor Antigens

   Viral stimulation has the potential to subvert the tolerogenic
   properties of KCs, potentially enhancing their ability to cross‐prime
   hepatic CD8^+T cell responses.^[ [130]^18 , [131]^20 ^] The increased
   acquisition of tumor‐derived materials by KCs following VSV‐M51R
   treatment prompted us to investigate whether their cross‐presentation
   capacity is also enhanced. Indeed, pathway analysis of our RNA‐seq data
   revealed enrichment of genes related to antigen processing and
   presentation in KCs from VSV‐M51R‐treated mice (Figure [132]4A).
   Specifically, genes encoding MHC class I (MHC‐I) molecules such as B2m,
   H2‐K1, H2‐Q5, H2‐Q6, and H2‐Q7, as well as molecular chaperones
   involved in antigen processing, such as Hspa8 and Hsp90a,^[ [133]^39 ,
   [134]^40 ^] were significantly upregulated in KCs after VSV‐M51R
   treatment (Figure [135]4B). Consistently, the surface expression of
   MHC‐I, MHC‐II, and the costimulatory receptors CD80 and CD86 was
   elevated on KCs from VSV‐M51R‐treated mice compared with those from
   untreated tumor‐bearing mice (Figure [136]4C,D). We also examined the
   expression of coinhibitory molecules that can impair T‐cell priming by
   KCs. Notably, complement receptor of the immunoglobulin superfamily
   (CRIg), also known as VSIG4, a potent T‐cell‐inhibitory molecule highly
   and constitutively expressed on KCs,^[ [137]^41 ^] was significantly
   downregulated following VSV‐M51R treatment, with approximately
   two‐thirds of Tim‐4^+ resident KCs completely losing CRIg expression
   (Figure [138]4E,F). However, programmed death 1 ligand 1 (PD‐L1)
   expression was elevated on KCs (Figure [139]4E,F), likely due to
   virus‐stimulated interferon responses, which are known to induce PD‐L1
   expression.^[ [140]^42 ^] Additionally, the upregulation of MHC class I
   molecules, costimulatory receptors and molecular chaperones was
   similarly observed in primary human KCs stimulated with VSV‐M51R in
   vitro (Figure [141]S3D–F, Supporting Information). Hence, these data
   strongly suggest that VSV‐M51R treatment may enhance the
   cross‐presentation capacity of KCs. To verify this, we established
   liver metastases in mice using B16F10 cells or MC38 cells expressing
   chicken ovalbumin (OVA) as a surrogate tumor antigen (B16F10‐OVA or
   MC38‐OVA) and assessed the KC expression of MHC‐I bound to SIINFEKL, a
   dominant CD8^+ T‐cell epitope derived from OVA, at 3 days post‐VSV‐M51R
   treatment. The results revealed a significant increase in both the
   proportion and expression intensity of KCs presenting SIINFEKL/H‐2K^b
   following VSV‐M51R treatment (Figure [142]4G,H; Figure [143]S3G,H,
   Supporting Information), confirming the augmented cross‐presentation of
   tumor antigens by KCs.

Figure 4.

   Figure 4
   [144]Open in a new tab

   VSV‐M51R treatment augments KC cross‐presentation of tumor antigens. A)
   KEGG pathway enrichment analysis of DEGs between KCs from control and
   VSV‐M51R‐treated tumor‐bearing mice. B) Heatmaps of upregulated genes
   related to antigen processing and presentation. C, untreated control;
   M, VSV‐M51R‐treated. C) Representative histograms of CD80, CD86, MHC‐I
   and MHC‐II expression on KCs on day 3 post‐VSV‐M51R treatment. D) Mean
   fluorescence intensity (MFI) of CD80, CD86, MHCI and MHCII on KCs in C.
   Data from 4–9 mice per group. E) Representative histograms of CRIg and
   PD‐L1 expression on KCs on day 3 post‐VSV‐M51R treatment. F) MFI of
   CRIg and PD‐L1 on KCs was quantified. Data from 10–11 mice per group.
   G) Representative flow cytometric plot of SIINFEKL/H‐2K^b expression in
   KCs on day 3 following VSV‐M51R treatment. H) Quantification of the
   proportion and MFI of SIINFEKL/H‐2K^b in KCs. Data from 10 mice per
   group. Data are represented as mean ± SEM. ^* p < 0.05; ^** p < 0.01;
   ^*** p < 0.001; ns, no significance (unpaired Student's t test).

2.5. VSV‐M51R Treatment Elicits Robust KC‐Dependent Antitumor CD8^+T Cell
Responses

   Given that VSV‐M51R treatment significantly increased the capacity of
   KCs to both take up and cross‐present tumor antigens, we hypothesized
   that this effect would lead to efficient induction of T‐cell responses
   against liver metastases. Profiling the composition of hepatic immune
   effector cells revealed a prominent increase in both the percentage and
   number of hepatic CD8^+T cells on day 7 post‐VSV‐M51R treatment,
   whereas natural killer (NK) cells and CD4^+T cells remained unchanged
   compared with those in the untreated group (Figure [145]5A,B). Further
   analysis of the CD8^+T cell population demonstrated an increase in the
   frequency of CD44^hiCD62L^lo effector T cells (T[EFF]), accompanied by
   a proportional decrease in CD44^−CD62L^hi naïve T cells (T[N]) and
   CD44^hiCD62L^hi central memory T cells (T[CM]) among hepatic CD8^+T
   cells, indicating a status of heighted activation (Figure [146]5C).
   Correspondingly, the ability of hepatic CD8^+ T cells to secrete
   effector molecules, such as IFN‐γ and Granzyme B, was greater than that
   of the untreated group (Figure [147]5D,E). To assess the impact of
   virotherapy on tumor antigen‐specific CD8^+T cell responses, we
   measured SIINFEKL/H‐2K^b tetramer^+ CD8^+T cells in the livers of
   B16F10‐OVA‐ or MC38‐OVA‐bearing mice and observed an increase in the
   percentage and effector function of these cells on day 3 post‐VSV‐M51R
   treatment (Figure [148]S4A–H, Supporting Information). Moreover,
   endogenous antigen‐specific CD8^+T responses against several reported
   neoepitopes in MC38 tumors, including Rpl18,^[ [149]^43 ^] Adpgk and
   Dpagt1,^[ [150]^44 ^] as measured by pMHC‐tetramer staining, were
   significantly increased in the livers of MC38 tumor‐bearing mice
   following VSV‐M51R treatment (Figure [151]S4I,J, Supporting
   Information). Additionally, intravital imaging of tumor‐bearing
   CD4‐cre‐ZsGreen reporter mice, in which the hepatic ZsGreen^+ cells
   were composed primarily of CD8^+T cells after VSV‐M51R treatment
   (Figure [152]S5A, Supporting Information), revealed enhanced T‐cell
   infiltration into the foci of liver metastases (Figure [153]5F,G).
   While T cells in untreated mice were mostly arrested at the tumor
   periphery, intratumoral T cells in VSV‐M51R‐treated mice exhibited
   increased motility (Figure [154]5G), a feature associated with active
   tumor immune surveillance by CD8^+T cells.^[ [155]^45 ^] Taken
   together, these findings demonstrate that VSV‐M51R treatment provokes
   robust antitumor CD8^+ T‐cell responses in liver metastases.

Figure 5.

   Figure 5
   [156]Open in a new tab

   VSV‐M51R elicits robust KC‐dependent antitumor CD8^+ T‐cell responses.
   A) Representative flow cytometric plots of hepatic NK, T, CD8^+ T and
   CD4^+ T cells at 7 days after VSV‐M51R treatment. B) Quantification of
   the proportions and numbers of NK, T, CD8^+ T and CD4^+ T cells. Data
   from 7–8 mice per group. C) Representative flow cytometric plots of
   T[N], T[CM], and T[EFF] cells at 7 days post‐VSV‐M51R treatment, gated
   on hepatic CD8^+T cells. D) Representative flow cytometric plots of
   IFN‐γ and Granzyme B expression by hepatic CD8^+ T cells. E)
   Quantification of the ratio of IFN‐γ‐ or Granzyme B‐expressing CD8^+ T
   cells in D. Data from 6–8 mice per group. F) Representative intravital
   images showing T‐cell infiltration in liver metastases at 7 days
   post‐VSV‐M51R treatment. G) The density, average velocity and
   displacement of intratumoral T cells were quantified. Each dot
   represents one tumor focus; n = 15–16 foci from 3 mice per group. H)
   Experimental design for I‐J. I) Liver photos were taken on day 7 after
   VSV‐M51R treatment with and without CD8^+T cell depletion. J) Liver
   weight and the liver‐to‐body weight ratio were quantified. Data from 4
   mice per group. K) Quantification of the proportions and numbers of
   hepatic CD8^+ T and CD4^+ T cells at 7 days post‐VSV‐M51R treatment in
   the presence or absence of KCs. Data from 6 mice per group. L)
   Representative flow cytometric plots and quantification of the ratio of
   IFN‐γ‐ or Granzyme B‐expressing CD8^+ T cells. Data from 6 mice per
   group. Data are expressed as mean ± SEM. ^* p < 0.05; ^** p < 0.01;
   ^*** p < 0.001; ns, no significance. Unpaired Student's t test for E,
   G, J, and L, and one‐way ANOVA with Tukey's test for B and K.

   To validate the importance of CD8^+ T‐cell responses in
   virotherapy‐induced tumor regression, we depleted T cells in
   tumor‐bearing CD4‐cre‐iDTR mice by administering DT prior to VSV‐M51R
   treatment. T‐cell depletion almost completely abrogated the therapeutic
   effect of VSV‐M51R (Figure [157]S5B, Supporting Information). A similar
   loss of therapeutic efficacy was observed when CD8^+T cells were
   selectively depleted with anti‐CD8α (Figure [158]5H–J), highlighting
   the indispensable contribution of CD8^+ T cells to VSV‐M51R‐mediated
   tumor control. Considering the elevated IFN‐γ production by CD8^+ T
   cells following VSV‐M51R treatment, we next investigated its
   contribution to the observed therapeutic effect. Although untreated
   IFN‐γ‐deficient (GKO) mice showed outgrowth of B16F10 metastases in the
   liver, reflecting an essential immune surveillance function of IFN‐γ in
   restricting the growth of liver metastases, VSV‐M51R treatment failed
   to reduce the tumor burden in these GKO mice (Figure [159]S5C,
   Supporting Information). These findings suggest that CD8^+ T cells may
   exert their tumoricidal activity at least in part through IFN‐γ
   production.

   Given the enhanced cross‐presentation of tumor antigens by KCs, we
   sought to explore the contribution of these APCs to the
   VSV‐M51R‐mediated induction of antitumor CD8^+ T‐cell responses. To
   address this, we established B16F10 liver metastases in Clec4f‐iDTR
   mice and concurrently depleted KCs upon VSV‐M51R treatment (Figure
   [160]S5D, Supporting Information). Compared with those of control mice,
   the proportions and numbers of CD8^+ T cells but not CD4^+ T cells in
   the livers of KC‐depleted mice were significantly lower
   (Figure [161]5K, [162]S5E). Moreover, the activation of hepatic CD8^+ T
   cells (Figure [163]S5F,G, Supporting Information), as well as their
   production of IFN‐γ and Granzyme B, was also strikingly diminished in
   the absence of KCs (Figure [164]5L). These findings confirm the
   indispensable role of KCs in eliciting efficient antitumor CD8^+ T‐cell
   responses. In summary, VSV‐M51R treatment significantly enhances the
   ability of KCs to take up and cross‐present tumor antigens, thereby
   triggering robust antitumor T‐cell responses that effectively control
   liver metastases.

2.6. Mitochondrial Antiviral Signaling Protein (MAVS)‐Dependent IFN‐I
Signaling Licenses KCs for Cross‐Presentation

   Type I interferon (IFN‐I) is a well‐known stimulator of T‐cell
   cross‐priming.^[ [165]^46 ^] KCs have been reported to be the primary
   early producers and responders of IFN‐I during systemic viral
   infections,^[ [166]^47 , [167]^48 ^] and we demonstrated that VSV‐M51R
   treatment strongly stimulated IFN‐I production in KCs (Figure
   [168]6A,B), suggesting that the increased induction of KC
   cross‐presentation following virotherapy was likely dependent on IFN‐I
   signaling. To test this hypothesis, IFN‐I receptor‐deficient (Ifnar^−/−
   ) mice bearing metastatic B16F10‐OVA melanomas were systemically
   administered VSV‐M51R (Figure [169]6C). As expected, IFNAR deficiency
   resulted in significantly decreased expression of OVA peptide‐loaded
   MHC‐I on KCs, as measured by surface staining for the SIINFEKL/H‐2K^b
   pMHC complex, indicating impaired cross‐presentation of tumor antigens
   (Figure [170]6D,E). To further investigate whether IFN‐I directly acted
   on KCs, we generated partial bone marrow chimeras in which a proportion
   of CD45.1 resident KCs were replaced by their Ifnar^−/− CD45.2 bone
   marrow‐derived counterparts (Figure [171]6F). These mice were treated
   with VSV‐M51R 7 days after intrasplenic inoculation of B16F10‐OVA
   tumors. Compared with IFNAR‐competent KCs, IFNAR‐deficient KCs
   presented much lower surface expression of SIINFEKL/H‐2K^b pMHC
   complexes (Figure [172]6G,H), suggesting that IFN‐I triggering of KCs
   is required for enhancing their cross‐presentation of tumor antigens.

Figure 6.

   Figure 6
   [173]Open in a new tab

   MAVS‐dependent IFN‐I signaling licenses KCs for cross‐presentation. A)
   mRNA expression levels of Ifnα and Ifnβ in liver tissues at the
   indicated time points after VSV‐M51R infection. B) mRNA levels of Ifnα
   and Ifnβ in KCs sorted from VSV‐M51R‐infected mice. Data from 3 mice
   per group. C) Liver metastasis of OVA‐expressing B16F10 melanoma was
   established in C57BL/6 or Ifnar^−/− mice, followed by intravenous
   treatment with VSV‐M51R. D) Representative flow cytometric plots and E)
   quantification of SIINFEKL/H‐2K^b expression in KCs two days after
   VSV‐M51R treatment. Data from 4 mice per group. F) Experimental design
   for partial reconstitution of KCs. Sublethally irradiated CD45.1 mice
   were reconstituted with bone marrow cells from Ifnar^−/− CD45.2 mice
   for 4 weeks, followed by the establishment of liver metastases and
   VSV‐M51R treatment. G) Representative flow cytometric plots and H)
   quantification of SIINFEKL/H‐2K^b expression on CD45.1 (WT) or CD45.2
   (Ifnar^−/− ) KCs after VSV‐M51R treatment. Data from 6 mice per group.
   I) Representative flow cytometric plots and J) quantification of
   SIINFEKL/H‐2K^b expression in KCs from cGas^−/− , Tlr3^−/− , Mavs^−/−
   and WT C57BL/6 mice 3 days after VSV‐M51R treatment. Data from 5–8 mice
   per group. K) Representative flow cytometric plots and L)
   quantification of SIINFEKL/H‐2K^b tetramer^+ CD8^+ T cells in the
   livers of cGas^−/− , Tlr3^−/− , Mavs^−/− and WT C57BL/6 mice. Data from
   5–8 mice per group. Data are represented as mean ± SEM. ^* p < 0.05;
   ^** p < 0.01; ^*** p < 0.001; ns, no significance. Unpaired Student's t
   test for B, E, and H and one‐way ANOVA with Tukey's test for A, J, and
   L.

   To exclude the possibility that IFN‐I directly stimulates T cells
   instead of APCs during cross‐priming,^[ [174]^49 ^] we adoptively
   transferred WT or Ifnar ^−/− T cells into the lymphopenia Rag1^−/− mice
   (Figure [175]S6A, Supporting Information). T cells of both genotypes
   were successfully reconstituted in Rag1^−/− recipients (Figure
   [176]S6B, Supporting Information) and showed comparable responses
   following VSV‐M51R treatment of liver metastases (Figure [177]S6C–G,
   Supporting Information), further suggesting that KCs, rather than T
   cells or B cells, were the primary targets of IFN‐I stimulation during
   virotherapy. Notably, we were unable to directly detect T‐cell
   responses in tumor‐bearing Ifnar^−/− mice, as these mice rapidly
   succumbed to VSV‐M51R within three days after intravenous
   administration, possibly due to unrestrained viral infections. Multiple
   intracellular signaling pathways contribute to IFN‐I production during
   VSV infection in different cell types, such as the RIG‐I‐MAVS,^[
   [178]^50 ^] cGAS‐STING^[ [179]^51 ^] and endosomal toll like receptor 3
   (TLR3) pathways.^[ [180]^52 ^] We tested the involvement of each of
   these pathways in VSV‐M51R‐augmented cross‐priming by KCs. While KCs
   from cGas^−/− and Tlr3 ^−/− mice were able to efficiently cross‐present
   tumor antigens upon VSV‐M51R treatment, KCs from Mavs ^−/− mice were
   significantly impaired (Figure [181]6I,J). Accordingly, the induction
   of tumor antigen‐specific CD8^+ T‐cell responses in the liver was
   largely diminished in Mavs ^−/− mice compared with that in cGas^−/− or
   Tlr3 ^−/− mice (Figure [182]6K,L). No significant morbidity or
   mortality was observed in any of these tumor‐bearing mouse strains upon
   VSV‐M51R treatment, suggesting that the blunted IFN‐I production was
   sufficient to protect against viral infection, although it compromised
   KC cross‐priming of hepatic antitumor T cells. These data collectively
   suggest that the MAVS‐IFN‐I signaling pathway is required to license
   KCs to efficiently cross‐prime hepatic antitumor CD8^+ T cells during
   VSV‐M51R treatment of liver metastases.

2.7. PD‐L1 Blockade Potentiates the VSV‐M51R‐Mediated Antitumor Effect
Against Advanced Liver Metastasis

   Although we showed that a single intravenous infusion of VSV‐M51R at
   the early stage of liver metastasis was sufficient to induce tumor
   regression (Figure [183]1), the therapeutic effects of VSV‐M51R
   virotherapy on late‐stage liver metastasis, which represents a more
   clinically relevant scenario, remain unclear. To this end, we
   administered VSV‐M51R on day 12 after intrasplenic inoculation of
   B16F10 melanoma cells to mimic advanced liver metastasis, as untreated
   mice typically start reaching the experimental endpoint around day
   15.^[ [184]^15 ^] However, the results indicated that a single infusion
   of VSV‐M51R had a minimal therapeutic effect on advanced liver
   metastases, demonstrating that VSV‐M51R virotherapy alone is
   insufficient to achieve tumor regression in late‐stage liver metastasis
   (Figure [185]S7A, Supporting Information).

   VSV‐M51R treatment induced a striking upregulation of PD‐L1 expression
   on KCs (Figure [186]4E,F). Given that PD‐1 expression gradually
   increases on hepatic CD8^+T cells during the progression of liver
   metastasis (Figure [187]7A,B), the intensified PD‐1/PD‐L1 interaction
   may impede the antitumor CD8^+T cell responses triggered by
   virotherapy, particularly in late‐stage liver metastasis. Therefore, we
   combined VSV‐M51R with anti‐PD‐L1 to treat advanced liver metastasis.
   While PD‐L1 blockade alone had no therapeutic effect on liver
   metastases, which is consistent with the well‐documented resistance of
   liver metastasis to immune checkpoint inhibitors,^[ [188]^3 , [189]^4
   ^] the combination treatment significantly reduced the hepatic
   metastatic tumor burden and prolonged the survival of tumor‐bearing
   mice compared to VSV‐M51R treatment alone, with a proportion of mice
   achieving near‐complete tumor eradication (Figure [190]7C–F; Figure
   [191]S7B, Supporting Information). This enhanced therapeutic efficacy
   was accompanied by a greater than 10‐fold increase in hepatic CD8^+ T
   cells, with a smaller increase in CD4^+T cells (Figure [192]7G).
   Further analysis of hepatic CD8^+ T cells revealed significantly
   increased production of effector molecules, such as IFN‐γ and Granzyme
   B (Figure [193]7H,I). Collectively, these results demonstrate that
   simultaneous blockade of PD‐L1 potentiates the therapeutic efficacy of
   VSV‐M51R virotherapy against advanced liver metastases by
   reinvigorating robust antitumor T‐cell responses.

Figure 7.

   Figure 7
   [194]Open in a new tab

   PD‐L1 blockade potentiates VSV‐M51R virotherapy against advanced liver
   metastasis. A) Representative flow cytometric plot and B)
   quantification of PD‐1 expression on hepatic CD8^+ T cells during the
   progression of B16F10 liver metastases. Data from 4 mice per group. C)
   Experimental design for combinatory treatment of late‐stage liver
   metastasis using VSV‐M51R and anti‐PD‐L1 antibodies. D) Mice were
   treated with VSV‐M51R and anti‐PD‐L1 on day 12 following B16F10 tumor
   inoculation and were harvested on day 19 as illustrated. E) Liver
   weight, the ratio of liver weight to body weight and the tumor area
   were quantified. Data from 5–6 mice per group. F) Mouse survival rates
   were monitored. Pooled data from 10–15 mice per group. G)
   Quantification of the proportions and numbers of hepatic CD4^+ and
   CD8^+ T cells. Data from 5 mice per group. H) Representative histograms
   and I) quantification of IFN‐γ and Granzyme B expression in hepatic
   CD8^+ T cells after combined VSV‐M51R and anti‐PD‐L1 treatment. Data
   from 5 mice per group. Data are expressed as mean ± SEM. ^* p < 0.05;
   ^** p < 0.01; ^*** p < 0.001; ns, no significance. Unpaired Student's t
   test for G and I, one‐way ANOVA with Tukey's test for B and E.
   Two‐sided log‐rank test for F.

3. Discussion

   Systemic infusion of immunostimulatory recombinant virus holds promise
   for targeting and functionally revitalizing KCs in the treatment of
   liver malignancies. However, KCs are susceptible to death upon various
   viral infections, particularly when high viral doses are delivered
   intravenously to meet therapeutic requirements. Therefore, an
   appropriate viral platform capable of stimulating the antitumor
   function of KCs while retaining their integrity is crucial.
   Additionally, the potential risks of viral persistence and viral genome
   integration should be minimized to improve safety for clinical usage.
   In these regards, VSV‐M51R has emerged as an ideal candidate for
   developing KC‐targeted virotherapies. Recombinant VSV has demonstrated
   safety in clinical trials even with systemic administration.^[ [195]^53
   , [196]^54 ^] The M51R mutation further reduces the potential toxicity
   of VSV by abrogating the virus's ability to inhibit host cell antiviral
   responses while simultaneously enhancing its immunostimulatory
   capacity.^[ [197]^33 ^] Moreover, VSV can be easily manipulated,
   allowing the engineering of genetic payloads into the virus. The
   self‐limited replication of VSV‐M51R within KCs could enable temporal
   and controllable expression of these therapeutic payloads, further
   increasing its therapeutic efficacy. Therefore, our findings highlight
   VSV‐M51R as a feasible, versatile and nontoxic viral platform for
   developing KC‐targeted immunotherapies against liver malignancies.

   Although VSV‐WT relies primarily on the M protein to activate the
   intrinsic pathway of apoptosis in cancer cells,^[ [198]^55 ^] the
   VSV‐M51R mutant can alternatively trigger Fas‐ and Caspase‐8‐mediated
   extrinsic pathways of apoptosis in host cells.^[ [199]^56 , [200]^57 ^]
   Moreover, adenovirus induces massive KC necrosis following high‐dose
   intravenous injection.^[ [201]^29 ^] It remains unclear why KCs are
   resistant to VSV‐M51R‐induced cell death but are vulnerable to other
   viral infections; one possibility is that their intrinsic
   susceptibility to different types of cell death may vary. It is
   conceivable that other VSV mutants capable of both sparing KCs from
   death and enhancing their antitumor activity may also offer therapeutic
   benefits against multifocal liver metastasis. Understanding the
   underlying molecular mechanisms could not only guide the development of
   KC‐targeting viral platforms but also provide insights into KC fate
   determination, as the loss of resident KCs is frequently observed
   during liver inflammation and is associated with disease progression.^[
   [202]^58 ^]

   It has been increasingly recognized that OVs can function beyond their
   direct cancer cell cytotoxicity by targeting noncancerous cells to
   elicit potent antitumor immunity.^[ [203]^24 , [204]^59 , [205]^60 ^]
   This is particularly true for intravenous OV delivery because of the
   widespread viral distribution. Recent studies have demonstrated that
   systemic administration of VSV‐∆M51 efficiently promotes tumor
   regression in infection‐resistant subcutaneous tumors, an effect that
   relies on the activation of conventional type 1 DCs (cDC1s) and their
   cross‐priming of antitumor T cells in tumor‐draining lymph nodes.^[
   [206]^24 , [207]^61 , [208]^62 ^] These findings reinforce previous
   findings that VSV infection of myeloid cells in secondary lymphoid
   organs (SLOs) is critical for antiviral T‐cell priming.^[ [209]^63 ^]
   Although these studies reported no substantial VSV replication in KCs,
   our findings clearly indicate that VSV‐M51R can transiently infect and
   replicate within KCs early after intravenous administration. This
   discrepancy may be attributed to differences in viral strains,
   infectious dosages or observation time points. Importantly, we observed
   that VSV‐M51R elicited robust, KC‐dependent intrahepatic anti‐tumor
   CD8^+T cell responses, suggesting that the mechanisms involved in
   T‐cell activation in liver metastases may differ from those in
   subcutaneous tumor models.

   In addition to DCs, macrophages, including KCs, also possess the
   ability to cross‐present engulfed antigens.^[ [210]^64 , [211]^65 ^] In
   fact, KCs are fully capable of cross‐priming hepatic T‐cell responses
   during acute viral infections.^[ [212]^20 ^] It is thus temping to
   propose here that VSV‐M51R‐stimulated KCs interact with and
   cross‐present tumor antigens to naïve CD8^+T cells recirculating
   through the bloodstream, leading to their activation and active
   surveillance of metastatic liver tumors. Although the potential
   involvement of DC‐mediated cross‐priming of antitumor T cells in SLOs
   cannot be ruled out, the fact that the spleen was removed in our liver
   metastasis model suggests a limited contribution of this process. In
   support of these findings, a recent study demonstrated that β‐glucan
   treatment stimulates antitumor T‐cell responses against pancreatic
   cancer liver metastasis by activating cross‐presentation in KCs, even
   with an intact spleen in this model.^[ [213]^17 ^] Moreover, while
   wild‐type VSV infection is sufficient to activate the cross‐priming
   function of DCs,^[ [214]^63 , [215]^66 ^] it barely retards the growth
   of liver metastases due to the induction of KC lysis, further
   underscoring the indispensable contribution of KC‐mediated
   intravascular cross‐priming during VSV virotherapy. Nevertheless, it is
   possible that VSV‐M51R treatment provokes the production of
   inflammatory cytokines by KCs, such as IL‐1, IFN‐I and TNFα, which are
   essential for the full maturation and activation of DCs.^[ [216]^24 ,
   [217]^67 ^] Future investigations into the coordination between KCs and
   DCs in eliciting hepatic antitumor T‐cell immunity would be important
   for advancing the development of virotherapy for liver metastasis.

   Local IFN‐I triggering of KCs has been demonstrated in various types of
   systemic viral infections and is essential for restricting viral
   replication and dissemination.^[ [218]^28 , [219]^47 ^] Here, we showed
   that direct sensing of IFN‐I by KCs can also augment their
   cross‐presentation of internalized antigens during VSV‐M51R
   virotherapy. Antigen cross‐presentation occurs via two distinct
   pathways, the cytosolic pathway and the vacuolar pathway. Although
   macrophages are generally thought to favor the vacuolar pathway for
   cross‐presentation,^[ [220]^68 ^] the upregulation of the molecular
   chaperones Hsp90 (encoded by Hsp90aa1) and Hsp70 (encoded by Hsp8a)
   supports the hypothesis that IFN‐I may stimulate the cytosolic pathway
   of antigen cross‐presentation in KCs. Hsp90 is well documented for
   pulling endosomal antigens out into the cytosol,^[ [221]^40 , [222]^69
   ^] whereas Hsp70 facilitates the translocation of cytosolic antigens
   into the proteasome for degradation.^[ [223]^70 ^] Their coordinated
   action may enable KCs to undergo cytosolic processing of engulfed
   tumor‐derived antigens. In addition, IFN‐I can reduce phagosomal
   acidification and prolong the retention of engulfed antigens, thereby
   promoting the cytosolic pathway of cross‐presentation.^[ [224]^46 ,
   [225]^71 ^] This action may be particularly important for KCs, which
   are likely the most highly phagocytic cells in the body. Alternatively,
   KCs may employ a process called cross‐dressing to acquire preformed
   pMHC complexes directly from tumor cells without undergoing antigen
   processing. This finding is supported by our previous findings showing
   that KCs can “nibble” live tumor cells,^[ [226]^15 ^] a process
   reminiscent of trogocytosis, which facilitates the transfer of the pMHC
   complex from tumor cells to phagocytes.^[ [227]^72 ^] A recent study
   suggested that Poly I:C can enhance cross‐dressing in DCs, possibly in
   an IFN‐I‐dependent manner.^[ [228]^73 ^] Whether and how cross‐dressing
   is involved in the KC‐mediated cross‐priming of antitumor T cells
   merits further investigation.

   In addition to the marked upregulation of MHC‐I and costimulatory
   molecules, VSIG4, a well‐characterized coinhibitory molecule^[ [229]^41
   ^] constitutively expressed on KCs, is significantly downregulated upon
   VSV‐M51R treatment. This downregulation is consistent with previous
   findings that inflammatory stimuli can inhibit VSIG4 expression^[
   [230]^41 ^] or promote its shedding.^[ [231]^74 ^] VSIG4 has been shown
   to directly suppress T‐cell activation and proliferation by mitigating
   TCR signaling upon antigen engagement.^[ [232]^75 ^] Therefore, its
   downregulation could be a critical mechanism by which VSV‐M51R subverts
   the tolerogenic property of KCs and promotes effective T‐cell priming.
   In contrast, PD‐L1 expression is elevated on KCs during VSV‐M51R
   treatment. Given that PD‐1 expression progressively increases on T
   cells as liver metastases grow, this could explain why VSV‐M51R
   treatment is less effective in treating late‐stage liver metastasis and
   provide a rationale for combining VSV‐M51R with PD‐1/PDL1 blockade in
   the treatment of advanced liver metastasis. Although OVs, including
   adenovirus and herpes simplex virus (HSV), have been tested for liver
   metastasis treatment in clinical trials, their overall clinical
   efficacy remains limited.^[ [233]^76 , [234]^77 , [235]^78 ^]
   Specifically, intravenous injection of adenovirus, which has been
   reported to induce rapid death of KCs,^[ [236]^29 ^] has shown no
   clinical benefits in all tested patients, further highlighting the
   importance of KCs in effective virotherapy for liver metastasis.^[
   [237]^77 ^] Taken together, our findings suggest that targeting KCs
   with a combination of intravenous VSV‐M51R administration and immune
   checkpoint inhibitors could represent a clinically applicable approach
   for treating patients with advanced liver metastasis.

4. Experimental Section

Mice

   C57BL/6 mice were purchased from Nanjing GemPharmatech Co. (Nanjing,
   China). Clec4f‐Cre‐tdTomato, iDTR, Cd4‐Cre, Rosa26‐LSL‐zsGreen,
   Rosa26‐LSL‐tdTomato, Ifnγ^−/− , cGas^−/− , Tlr3^−/− , Mavs^−/− , and
   Ifnar1^−/− ko mice were originally obtained from the Jackson
   Laboratory. Clec4f‐CreER and Rosa26‐LSL‐BCL2‐2A‐tdTomato mice were
   generated by Shanghai Model Organisms Center Inc. (Shanghai, China).
   All the mice were maintained in a specific pathogen–free (SPF) facility
   at USTC. Animal experiments were conducted under the guidelines of the
   animal care committee of USTC with the approval number
   #USTCACUC192401034. Unless otherwise stated, 6‐ to 8‐week‐old mice of
   both sexes were used. Age‐ and sex‐matched littermates of the knock‐in
   or knockout mice were used as controls.

Cell Lines and Viruses

   B16F10, 293T, Pan02, and Vero cells were purchased from the China
   Center of Type Culture Collection (CCTCC). The MC38, MC38‐OVA and
   B16F10‐OVA cell lines were kindly provided by Dr. Zhigang Tian (USTC,
   Hefei) and Dr. Shu Zhu (USTC, Hefei). B16F10 cell lines stably
   expressing ZsGreen or tdTomato fluorescent proteins were generated via
   lentiviral transduction. The cell lines were cultured in RPMI 1640
   (Biosharp) or Dulbecco's modified Eagle medium (DMEM, Vivacell)
   supplemented with 10% fetal calf serum (Biological Industries) and 1%
   penicillin/streptomycin (Biosharp). GFP‐encoding VSV‐WT and VSV‐M51R,
   both Indiana strains, were previously reported.^[ [238]^52 ^] For virus
   propagation, Vero cells were cultured in α‐MEM supplemented with 10%
   FBS until they reached 80% confluence. The cell monolayer was infected
   with VSV‐WT or VSV‐M51R at a MOI of 0.01. The supernatant containing
   the VSV particles was collected after 72 h, centrifuged at 3000 × g for
   20 min, and passed through a 0.45 µm filter. Virus quantification was
   performed via the median tissue culture infectious dose (TCID[50])
   assay as described previously.^[ [239]^79 ^] The mRNA copy number of
   the VSV N gene in the tissues was determined via quantitative PCR
   (qPCR). Briefly, total RNA was extracted from tumor and control liver
   tissues using the TRIzol reagent (Invitrogen), and transcribed into
   cDNA via the HiScript III qRT SuperMix Kit (Vazyme Biotech). The
   purified VSV‐N plasmid at a concentration of 50 ng µL^−1 was serially
   diluted to serve as a template for generating the standard curve. SYBR
   green‐based real‐time qPCR amplification was conducted using the
   qTOWER^3 system (Analytik Jena, Germany). The primers for VSV‐N were as
   follows: forward primer: 5′‐TGA‐TAG‐TAC‐CGG‐AGG‐ATT‐GAC‐GAC‐3′; reverse
   primer: 5′‐CCT‐TGC‐AGT‐GAC‐ATG‐ACT‐GCT‐CTT‐3′.

Animal Models and In Vivo Treatment

   An intrasplenic injection‐based multifocal liver metastasis model was
   established as previously described.^[ [240]^15 ^] Briefly, 6‐ to
   8‐week‐old mice were anesthetized with isoflurane, mouse fur at the
   left upper flank was removed with a razor, the spleen was exposed via a
   small incision in the left flank, and tumor cells were injected into
   the spleen using a 30‐gauge needle, followed by splenectomy 5 min after
   injection. Unless otherwise stated, a total of 3 × 10^5 B16F10 cells, 3
   × 10^5 B16F10‐ZsGreen cells, 3 × 10^5 B16F10‐OVA cells, 5 × 10^5 MC38
   cells, 5 × 10^5 MC38‐OVA cells or 1 × 10^6 Pan02 cells were injected in
   50 µL of PBS to establish metastatic liver tumors. For in vivo
   virotherapy of metastatic B16F10 and MC38 tumors, tumor‐bearing mice
   received intravenous (i.v.) administrations of VSV or VSV‐M51R at a
   dose of 3 × 10^8 PFU on day 7 post‐tumor inoculation. Alternatively,
   tumor‐bearing mice were i.v. injected with virus on day 12 to mimic
   late‐stage liver metastasis and were harvested on day 19. For the
   slow‐growing Pan02 metastatic tumors, tumor‐bearing mice were treated
   with VSV‐M51R on day 10 post‐tumor inoculation and harvested on day 21.
   Mice were monitored daily during this period, and an increase in body
   weight greater than 20% was considered to have reached the end point of
   the experiment. For combinatory treatment with anti‐PD‐L1 antibodies,
   mice were additionally injected intraperitoneally (i.p.) with 100 µg of
   anti‐PDL1 (#124 339, Biolegend) at days 11, 13, 15, and 17 after tumor
   inoculation as depicted.

   To deplete KCs, Clec4f‐iDTR mice were i.p. injected with 200 ng DT
   (D0564, Millipore Sigma). To deplete T cells, Cd4‐iDTR mice were i.p.
   injected with 100 ng DT daily for 3 consecutive days. To deplete CD8^+
   T cells, tumor‐bearing mice were i.p. injected with 100 µg of anti‐CD8α
   (BE0061, Bio X Cell) antibody three times at days 5, 8, and 11 after
   tumor inoculation. For the fate‐mapping experiments, a dose of 1 mg
   tamoxifen (Millipore) dissolved in corn oil (Millipore) was orally
   administered to Clec4f‐CreERT2:R26‐ tdTomato mice three days before
   VSV‐M51R injection. For in vivo inhibition of cell death, the
   pan‐caspase inhibitor Z‐VAD‐FMK (10 mg kg^−1), the caspase‐9 inhibitor
   Z‐LEHD‐FMK (6 mg kg^−1) or the necroptosis inhibitor Necrostatin‐1 (5
   mg kg^−1) were dissolved and diluted in 200 µL of saline solution and
   were i.p. injected 30 min before VSV infection.

Intravital Microscopy (IVM)

   Surgical preparation of mice for IVM of the liver was conducted as
   previously described.^[ [241]^15 ^] Briefly, the tail vein of
   anesthetized mice was cannulated with PE‐10 tubing filled with sterile
   saline, allowing for the delivery of fluorescence‐conjugated antibodies
   and additional anesthetics. The following antibodies or dyes were used:
   Annexin V (1.4 µg), anti‐F4/80 (1.8 µg), anti‐TIM4 (2 µg), and PI (5
   µg). Midline laparotomy was performed to exteriorize the left lateral
   lobe of the liver, which was then laid against the glass coverslip
   embedded in a customized temperature‐controllable sample holder. The
   surface of the liver was covered with strips of saline‐moistened
   Kimwipes (Fisher Scientific) to restrict movement. The sample holder
   fit the motorized microscope stage and was maintained at 37 °C during
   the course of the experiments. For spinning‐disk confocal IVM, example
   images were acquired via a Nikon Ti2‐E inverted microscope coupled with
   a Yogokawa CSU‐W1 spinning disk scanner. The microscope was equipped
   with four lasers (405, 488, 561, and 640 nm; Toptica) and four emission
   bandpass filters (447/60, 525/50, 617/73, and 685/40) to offer
   multichannel high‐resolution imaging that was recorded using an sCMOS
   camera (Prime95B, Photometrics). For two‐photon IVM, a Nikon AX R MP
   upright microscope system equipped with a resonance laser scanner head,
   a dual‐output pulsed laser (1040 nm and 660–1320 nm, Coherent
   Discovery) and GaAsP HyD detectors was used. Example images were
   analyzed and quantified via NIS‐Elements AR software (version 5.20.00)
   and ImageJ (Fiji). 3D reconstruction was conducted using Imaris
   (version 7.0, Bitplane).

Cell Isolation and Flow Cytometry

   Mouse livers were cut into small pieces and incubated with a digestion
   mixture comprising 10 mL of prewarmed DMEM supplemented with 0.5
   mg mL^−1 collagenase I (C0130, Millipore Sigma) and 5 U mL^−1 DNase I
   (11 284 932 001, Millipore Sigma). Liver tissues were digested at 37 °C
   with shaking at 200 rpm (Constant temperature incubator shaker,
   #ZWY240, ZhiCheng) for 20 min and then filtered through a 70‐µm cell
   strainer (Biosharp). After the removal of hepatocytes and tissue debris
   by a short centrifugation, the cells were pelleted and washed by
   centrifugation at 400 × g at 4 °C for 5 min. Liver nonparenchymal cells
   (LNPCs) were then obtained following the lysis of red blood cells and
   were resuspended in ice‐cold 1× PBS. For surface marker staining, the
   cells (1 × 10⁶ in 100 µL) were incubated with an Fc blocker (2.4G2, Bio
   X Cell) for 20 min before being stained with a mixture of
   fluorophore‐conjugated antibodies for 30 min at 4 °C in the dark. For
   tetramer staining, the cells were stained with PE or Alexa 647‐labeled
   H‐2K^b‐SIINFEKL tetramer (NIH) for 30 min at 4 °C in the dark prior to
   surface marker staining. For intracellular staining, the cells were
   pretreated with the Cell Activation Cocktail (with Brefeldin A)
   (423 304, BioLegend) for 4 h prior to surface marker staining, followed
   by fixation and permeabilization using the Foxp3 Staining Buffer Set
   (00‐5523‐00, eBioscience). DAPI (Biosharp) or Zombie Violet dyes
   (423 102, BioLegend) were used to exclude dead cells. Cell‐counting
   beads (424 902, BioLegend) were added to the samples before flow
   cytometric detection for cell enumeration. Data acquisition was carried
   out on either the BD Fortessa or Beckman CytoFLEX flow cytometer. Flow
   cytometric data were analyzed using the FlowJo (version 10.4). The
   following gating strategy was used to define the cell populations: KCs
   (CD45^+ Ly6G^− CD11b^low F4/80^+), NK cells (CD45^+ NK1.1^+), T cells
   (CD45^+NK1.1^−CD3^+), CD8^+ T cells (CD45^+NK1.1^−CD3^+ CD8^+) and
   CD4^+ T cells (CD45^+NK1.1^− CD3^+ CD4^+). The antibodies used for flow
   cytometry were as follows: FITC anti‐mouse/human CD11b (M1/70,
   Biolegend), FITC anti‐mouse CD3 (145‐2C11, Biolegend), FITC anti‐mouse
   CD62L (MEL‐14, Biolegend), FITC anti‐mouse CD80 (Armenian Hamster IgG,
   Biolegend), Alexa Fluor 488 anti‐mouse VSIG4 (NLA14, Biolegend), Alexa
   Fluor 488 anti‐mouse F4/80 (BM8, Biolegend), PE anti‐mouse Tim4
   (RMT4‐54, Biolegend), PE anti‐mouse CD4 (RM4‐5, Biolegend), PE
   anti‐mouse IFN‐γ (XMG1.2, Biolegend), PE anti‐mouse H‐2K^b bound to the
   SIINFEKKL Antibody (25‐D1.16, Biolegend), PE anti‐mouse CD274 (10F.9G2,
   Biolegend), PE anti‐mouse CD86 (A17199A, Biolegend), PerCP/Cy5.5
   anti‐mouse CD8a (53‐6.7, Biolegend), PerCP/Cy5.5 anti‐mouse H‐2K^b
   (AF6‐88.5, Biolegend), PE/Cyanine7 anti‐mouse Tim‐4 (RMT4‐54,
   Biolegend), PE/Cyanine7 anti‐mouse/human CD11b (M1/70, Biolegend),
   PE/Cyanine7 anti‐mouse CD45.1 (A20, Biolegend), PE/Cyanine7 anti‐mouse
   NK1.1 (PK136, Biolegend), APC anti‐mouse F4/80 (BM8, Biolegend), APC
   anti‐mouse CD44 (IM7, Biolegend), APC anti‐human/mouse Granzyme B
   Recombinant Antibody (QA16A02, Biolegend), APC anti‐mouse CD3 (17A2,
   Biolegend), Alexa Fluor 647 anti‐mouse Tim‐4 (RMT4‐54, Biolegend),
   APC/Cyanine7 anti‐mouse CD45 (30‐F11, Biolegend), APC/Cyanine7
   anti‐mouse CD45.2 (104, Biolegend), APC/Cyanine7 anti‐mouse CD3ε
   (145‐2C11, Biolegend), Alexa Fluor 700 anti‐mouse Ly‐6G (1A8,
   Biolegend), Alexa Fluor 700 anti‐mouse CD45 (30‐F11, Biolegend),
   Brilliant Violet 421 anti‐mouse F4/80 (BM8, Biolegend), Brilliant
   Violet 421 anti‐mouse CD62L(MEL‐14, Biolegend), Brilliant Violet 605
   anti‐mouse F4/80 (BM8, Biolegend), Brilliant Violet 510 anti‐mouse CD4
   (RM4‐4, Biolegend), Brilliant Violet 510 anti‐mouse I‐A/I‐E
   (M5/114.15.2, Biolegend), Brilliant Violet 785 anti‐mouse/human CD11b
   (M1/70, Biolegend), Anti‐Calreticulin antibody (ab2907, Abcam), Donkey
   Anti‐Rabbit IgG H&L (Alexa Fluor 647, Abcam), APC anti‐human CD68
   (Y1/82A, Biolegend), PE/Cyanine7 anti‐human CD45 (HI30, Biolegend), PE
   Rat IgG1, κ Isotype Ctrl (RTK2071, Biolegend), PE Mouse IgG1, κ Isotype
   Ctrl (MOPC21, Biolegend), FITC Rat IgG2a, κ Isotype Ctrl (RTK2758,
   Biolegend). H‐2K^b chicken ova 257–264 SIINFEKL PE‐labeled tetramer
   (NIH), H‐2K^b chicken ova 257–264 SIINFEKL Alexa 647‐labeled tetramer
   (NIH). Customized PE‐labeled H2K^b‐RPL18 neoepitope KILTFDRL tetramer,
   H2D^b‐Adpgk neoepitope ASMTNMELM tetramer and H2K^b‐Dapgt1 neoepitope
   SIIVFNLL tetramer were obtained from the Gene‐Bio Ltd. (Hefei)

Human KC Isolation and Treatment

   Human liver tissues were obtained from donor liver grafts procured from
   brain‐dead individuals who met the Chinese Classification Criteria for
   organ donation after Cardiac Death. Informed consent for the use of
   donor organs in transplantation and scientific research was obtained
   from the donors’ next of kin. All procedures were critically reviewed
   and approved by the institutional ethics board of the First Affiliated
   Hospital of USTC (#AF/SC‐12‐2/04.0‐2024KY‐510). For KC isolation, liver
   tissue was cut into small pieces (1–2 mm^3) and digested in prewarmed
   DMEM containing 2 mg mL^−1 type IV collagenase (C5138, Millipore Sigma)
   at 37 °C with agitation at 100 rpm for 30 min (Constant temperature
   incubator shaker, #ZWY240, ZhiCheng). The samples were vortexed every
   10 min during digestion. The resulting mixture was filtered through a
   100‐µm cell strainer (Biosharp). The cell suspension was then
   centrifuged at 50 × g for 3 min at 4 °C to separate hepatocytes
   (pellets) from non‐parenchymal cells. The supernatant was collected and
   further centrifuged at 500 × g for 5 min at 4 °C. The resulting cell
   pellets were resuspended in 10 mL of DMEM and centrifuged again at 500
   × g for 5 min at 4 °C. The final cell pellets were cultured in DMEM
   supplemented with 10% fetal calf serum and 1% penicillin/streptomycin.
   After a 4‐h incubation at 37 °C with 5% CO[2], non‐adherent cells and
   debris were removed by gently washing with PBS. The adherent cells,
   which were composed predominantly of KCs, were maintained in DMEM
   supplemented with 10% fetal calf serum and 1% penicillin/streptomycin.
   The KC identity of these adherent cells was confirmed via flow
   cytometry. For VSV‐M51R treatment, KCs were seeded into 6‐well plates
   at a density of 5 × 10^5 cells/well. After 12 h of culture to allow
   cell monolayer formation, the cells were infected with VSV‐M51R at a
   MOI of 0.6 and incubated for an additional 24 h. The cells were then
   harvested and subjected to qPCR analysis of the following molecules,
   using the following primers: CD80 sense
   (5′‐CTC‐TTG‐GTG‐CTG‐GCT‐GGT‐CTT‐T‐3′) and CD80 anti‐sense
   (5′‐GCC‐AGT‐AGA‐TGC‐GAG‐TTT‐GTG‐C‐3′); CD86 sense (5′‐
   CCA‐TCA‐GCT‐TGT‐CTG‐TTT‐CAT‐TCC‐3′) and CD86 anti‐sense (5′‐
   GCT‐GTA‐ATC‐CAA‐GGA‐ATG‐TGG‐TC‐3′); HLA‐A sense (5′‐
   AGA‐TAC‐ACC‐TGC‐CAT‐GTG‐CAG‐C‐3′) and HLA‐A anti‐sense (5′‐
   GAT‐CAC‐AGC‐TCC‐AAG‐GAG‐AAC‐C‐3′); HLA‐B sense (5′‐
   CTG‐CTG‐TGA‐TGT‐GTA‐GGA‐GGA‐AG‐3′) and HLA‐B anti‐sense (5′‐
   GCT‐GTG‐AGA‐GAC‐ACA‐TCA‐GAG‐C‐3′); HLA‐C sense (5′‐
   GGA‐GAC‐ACA‐GAA‐GTA‐CAA‐GCG‐C‐3′) and HLA‐C anti‐sense (5′‐
   ACA‐TCC‐TCT‐GGA‐GGG‐TGT‐GAG‐A‐3′); HSP90AA1 sense (5′‐
   TCT‐GCC‐TCT‐GGT‐GAT‐GAG‐ATG‐G‐3′) and HSP90AA1 anti‐sense (5′‐
   CGT‐TCC‐ACA‐AAG‐GCT‐GAG‐TTA‐GC‐3′); HSPA8 sense (5′‐
   TCC‐TAC‐CAA‐GCA‐GAC‐ACA‐GAC‐C‐3′) and HSPA8 anti‐sense (5′‐
   CAG‐GAG‐GTA‐TGC‐CTG‐TGA‐GTT‐C‐3′); GAPDH sense (5′‐
   GTC‐AAG‐GCT‐GAG‐AAC‐GGG‐AA‐3′) and GAPDH anti‐sense (5′‐
   AAA‐TGA‐GCC‐CCA‐GCC‐TTC‐TC‐3′);

Bulk RNA Sequencing

   KCs were isolated via a two‐step in situ perfusion procedure. Briefly,
   the mice were anesthetized with isoflurane, and a 26G catheter
   connected to the pump was inserted into the portal vein. The inferior
   vena cava was cut immediately after catheter insertion. The liver was
   perfused via the portal vein with an initial 20 mL of HBSS (with 0.5 mm
   EGTA) followed by 20 mL of perfusion buffer (HBSS containing calcium
   and magnesium + 0.05% collagenase IV) at a constant flow rate of 5
   mL min^−1. The entire liver was removed, minced and passed through a 70
   µm cell strainer. The cell suspension was centrifuged at 50 × g for 3
   min to remove liver parenchymal cells. The supernatant was then
   centrifuged at 400 × g for 5 min at 4 °C, and the resulting pellet was
   resuspended in 1 mL RBC lysis buffer for 1 min. LNPC was obtained after
   washing and was subsequently resuspended in 1 mL of HBSS buffer for
   antibody staining. Zombie^−CD45^+TIM4^+KCs were sorted using a BD
   FACSAria III with a purity greater than 95%. A total of 20 000 to
   50 000 cells were collected, and total RNA was extracted using the
   RNeasy Micro Kit (QIAGEN, USA). RNA quality was assessed via an Agilent
   Bioanalyzer (Agilent, USA), with RNA integrity numbers (RINs) ranging
   from 6–9. cDNA libraries were prepared from 2 ng of total RNA using the
   Smart‐Seq v2 protocol as previously reported.^[ [242]^80 ^] In brief,
   50 pg of cDNA was processed with the Nextera XT DNA Library Preparation
   Kit (Illumina, USA) via 12–14 PCR cycles according to the
   manufacturer's instructions. The length distribution of the cDNA
   libraries was monitored using an Agilent Bioanalyzer. cDNA libraries
   that passed quality control (QC) were subjected to Illumina sequencing.
   For analysis of the transcriptomic data, poor‐quality reads were first
   filtered via Trimmomatic. The raw reads were aligned to the mouse
   reference genome GRCm38_M13 provided by GENCODE using STAR. The
   FeatureCounts tool was used to assign exonic reads to the corresponding
   genes. The expression data were imported into the R statistical
   environment (R version 4.1.1) for further analysis of differential gene
   expression. GO and KEGG pathway analyses were conducted with the
   “clusterProfiler” package in R.

In Vivo KC Editing

   In vivo editing of KCs was performed using the BIL‐CRISPR approach as
   we described previously.^[ [243]^15 ^] Briefly, two pairs of sgRNA
   oligos targeting different exons of Bax or Bak were designed and
   synthesized as follows: Bax‐exon1: GAC‐GGG‐TCC‐GGG‐GAG‐CAG‐CT;
   Bax‐exon2: GCT‐CTG‐AAC‐AGA‐TCA‐TGA‐AG; Bak‐exon1:
   CCG‐AAG‐GTG‐GGC‐TGC‐GAT‐GA; Bak‐exon2: ACC‐CCG‐AGA‐TGG‐ACA‐ACT‐TG.
   sgRNA oligos were then annealed and ligated sequentially into the
   pX459‐2U6‐BsaI‐SapI vectors using the BsaI‐ and SapI‐restricted
   enzymes. A total of 5–10 µg of sequence‐verified plasmids were
   electroporated into ClearColi competent E. coli (Lucigen). The
   transformed bacteria were plated onto LB agar media in 15 cm petri
   dishes in the presence of ampicillin (100 µg mL^−1). Bacteria were
   gently scraped from the plates after 16–18 h of overnight culture,
   collected into a 50 mL centrifuge tube, weighed, and washed with
   saline. The bacterial number was calculated as follows: total CFU = 30
   × bacterial weight (g) ×10^10. For KC editing, 8‐week‐old mice were
   i.v. administered 10^9 CFU ClearColi containing Bax/Bak dual editing
   plasmids or backbone vectors and waited for 7 days before further
   experiments.

Statistical Analysis

   Every experiment was independently repeated at least twice, with each
   group containing a minimum of three biological replicates. Statistical
   analysis was conducted using the GraphPad Prism software (version 9.0).
   Data were expressed as the mean ± SEM. Unpaired Student's t test was
   used for comparisons between 2 individual groups. One‐way ANOVA with
   Tukey's test was used for multiple group comparisons. Mouse survival
   was analyzed via a 2‐sided log‐rank test. All experiments were repeated
   at least twice, with 3–5 biological samples per group. ^* p < 0.05; ^**
   p < 0.01; ^*** p < 0.001; ^**** p < 0.0001; ns, no significance.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   C.C. designed and conducted the majority of the experiments and
   analyzed the data. Q.Z., J.L., and X.Z. assisted in some experiments;
   J.W., D.G., and L.L. provided valuable experimental resources, critical
   thinking and technical assistance; Z.Z. conceived and supervised the
   study; C.C. and Z.Z. wrote the manuscript.

Supporting information

   Supporting Information
   [244]ADVS-12-e04929-s001.pdf^ (9.8MB, pdf)

Acknowledgements