Abstract

   The chemokine CXCL6 is identified as a pivotal regulator of biological
   processes across multiple malignancies. However, its function in
   cholangiocarcinoma (CCA) is underexplored. Tumor profiling for CXCL6 is
   performed using a public database. Both in vitro and in vivo
   experiments are utilized to evaluate the oncogenic effects of CXCL6 on
   CCA. Additionally, RNA‐Seq is employed to detect transcriptomic changes
   related to CXCL6 expression in CCA cells and neutrophils. Molecular
   docking, fluorescence colocalization, and Co‐IP are used to elucidate a
   direct interaction between JAKs and CXCR1/2. Additionally, LC‐MS
   lipidomics and explored the impact of CXCL6 on immunotherapy in vivo.
   CXCL6 is upregulated in CCA tissues and promoted the proliferation and
   metastasis of CCA. Mechanistically, CXCL6 regulated the
   CXCR1/2‐JAK‐STAT/PI3K axis in CCA via autocrine signaling, leading to
   lipid metabolic reprogramming, and promoted neutrophil extracellular
   traps (NETs) formation by activating the RAS/MAPK pathway in
   neutrophils. Eventually, NETs formation induced immunotherapy
   resistance in CCA by blocking CD8^+T cell infiltration. CXCL6 modulates
   CCA progression through the CXCR1/2‐JAK‐STAT/PI3K axis and reshaping
   its lipid metabolism. CXCL6 also mediates immunotherapy resistance
   through NETs, which may be a potential therapeutic target and biomarker
   for CCA.

   Keywords: Cholangiocarcinoma, CXCL6, Lipid metabolism, Neutrophil
   extracellular traps, Immunotherapy
     __________________________________________________________________

   CXCL6's function in cholangiocarcinoma is underexplored. Here it is
   discovered that CXCL6 is upregulated in cholangiocarcinoma tissues and
   promotes proliferation and metastasis of it. CXCL6 functions through
   the CXCR1/2‐JAK‐STAT/PI3K axis and reshaping tumor lipid metabolism.
   CXCL6 also mediates immunotherapy resistance through NETs by blocking
   CD8^+T cell infiltration, which may be a potential therapeutic target
   and biomarker for cholangiocarcinoma.

   graphic file with name ADVS-12-2503009-g001.jpg

1. Introduction

   The global incidence of CCA is rising, particularly in Asia.^[ [54]^1 ,
   [55]^2 ^] CCA originates from bile duct epithelial cells and is usually
   further subdivided into intrahepatic cholangiocarcinoma (iCCA),
   periportal cholangiocarcinoma (pCCA), and distal cholangiocarcinoma
   (dCCA).^[ [56]^3 ^] Due to its insidious onset, most patients are
   diagnosed at the advanced stage, with only 15–30% eligible for surgery
   and the recurrence rate exceeding 50%.^[ [57]^4 ^]

   This being the case, systemic drug therapy has become the mainstay for
   advanced CCA. Chemotherapy based on gemcitabine (GEM) was traditionally
   the most common treatment,^[ [58]^5 , [59]^6 ^] although advances in
   sequencing technology enabled mutated genes related to CCA to be
   systematically analyzed, inspiring a wave of targeted treatments. Drugs
   targeting FGFR2, IDH1/2, HER2, and IDH2 have been widely reported and
   even put into clinical use in CCA.^[ [60]^7 ^] Meanwhile, combining
   chemotherapy with immunotherapy opened up a new era, such that GEM plus
   cisplatin (GC) in combination with immunotherapy has now become the
   first‐line treatment for CCA.^[ [61]^8 ^] Nevertheless, the latest
   results from the TOPAZ‐1 clinical trial indicate an overall response
   rate of only 26.7% with GC plus the anti‐PD‐L1 antibody durvalumab,
   which is still far from satisfactory.^[ [62]^9 ^] Further research is
   needed to understand the tumor immune microenvironment of CCA and
   identify suitable strategies to overcome immune resistance.

   The TIME of CCA is a highly complex, dynamic system comprising tumor
   cells, immune components, and blood vessels, among other
   constituents.^[ [63]^10 ^] These different components interact with
   each other in ways that greatly affect the occurrence, development, and
   immune resistance of CCA,^[ [64]^11 ^] which is characterized as a
   “cold tumor” with dense desmoplastic stroma that may secrete
   angiogenesis‐promoting molecules such as vascular endothelial growth
   factors (VEGF).^[ [65]^12 ^] Thus, targeting angiogenesis is a
   potential anti‐CCA option and the entry point of our study.

   Immune cells are also considered functionally important in mediating
   CCA immunotherapy resistance. As well as CD8^+ T cells, which have been
   extensively studied, innate immune cells like tumor‐associated
   neutrophils (TANs) have also been shown to be closely related to tumor
   development and overall survival (OS) of patients.^[ [66]^13 ^] The
   number of TANs has been associated with the prognosis of CCA surgery
   patients but the functional mechanisms have not been fully analyzed.^[
   [67]^14 ^] However, recent research revealed that TANs can form NETs in
   the extracellular space and greatly reshape the TIME,^[ [68]^15 ^]
   suggesting an important way in which TANs might induce immunotherapy
   resistance in CCA.

   Furthermore, chemokines may function as the bridge between CCA and
   neutrophils. The C‐X‐C motif chemokine ligand (CXCL) subfamily is an
   important class of chemokines with close connections to tumor
   biological processes;^[ [69]^16 ^] CXCLs act as chemotactic factors for
   TANs, mediating their recruitment into the TIME to promote tumor drug
   resistance.^[ [70]^17 ^] These chemokines also directly regulate
   processes in cancer cells. For example, CXCL1, CXCL2, and CXCL8
   reportedly affect the proliferation of lung cancer cells and also
   influence angiogenesis, which may lead to tumor infiltration of
   neutrophils and macrophages and induce anlotinib resistance.^[ [71]^18
   , [72]^19 ^]

   By binding to C‐X‐C motif chemokine receptors (CXCRs), CXCLs activate
   several tumor‐associated pathways including MAPK, STAT, TGF‐β, and
   β‐catenin, thereby modulating the development and metabolism of tumor
   cells,^[ [73]^20 , [74]^21 ^] hinting at a possible mechanistic link
   between chemokine signaling and CCA treatment resistance. Notably,
   overexpression of CXCL6 was reported in high‐stemness malignant CCA
   cells;^[ [75]^22 ^] it was also found to promote tumor progression in
   hepatocellular carcinoma (HCC) and non‐small cell lung cancer,^[
   [76]^23 , [77]^24 ^] and to regulate neutrophil infiltration in gastric
   cancer.^[ [78]^25 ^] However, the function and mechanism of CXCL6 in
   CCA have not been investigated in detail.

   Here, we aimed to explore the functions and mechanisms of CXCL6 in CCA
   cells and the TIME. We demonstrated that expression of CXCL6 regulated
   proliferation, metastasis, and prognosis in CCA through modulating the
   CXCR1/2‐JAK‐STAT/PI3K axis. Eventually, CXCL6 could reshape lipid
   metabolism in CCA cells, and its overexpression also related to CCA
   immunotherapy resistance since CXCL6 induced NETs formation and led to
   reduced infiltration of CD8^+ T cells. Overall, our results position
   CXCL6 as a pivotal regulator in CCA development and immunotherapy
   resistance by reshaping lipid metabolism and inducing the formation of
   NETs.

2. Results

2.1. CXCL6 Is Upregulated in CCA Tumor Tissue and Correlates with Poor
Prognosis and Enhanced Malignancy

   Angiogenesis‐related checkpoints are considered to be major drivers of
   CCA progression and development.^[ [79]^26 ^] Therefore, in addition to
   containing vascularization, antiangiogenic agents may remodel the TIME
   and potentiate the efficacy of chemotherapy and immunotherapy.^[
   [80]^27 ^] To identify key genes contributing to angiogenesis and
   progression in CCA, we explored The Cancer Genome Atlas (TCGA) database
   (TCGA‐CHOL) to look for the intersection of genes that were highly
   expressed in CCA (log[2]|FC| ≥ 1, P ≤ 0.05), prognosis‐related (HR > 1,
   P ≤ 0.1), and associated with angiogenesis (i.e., genes belonging to
   the “HALLMARK_ANGIOGENESIS” dataset from the Molecular Signatures
   Database). CXCL6 was the sole gene fulfilling all criteria (Figure
   [81]1A–C; Figure [82]S1A, Supporting Information) and was further
   validated as a prognostic marker for DSS in CCA (Figure [83]S1B,
   Supporting Information).

Figure 1.

   Figure 1
   [84]Open in a new tab

   CXCL6 is upregulated in CCA tumor tissue and relates to poor prognosis
   and greater malignancy. A) Identification of CXCL6 based on TCGA data.
   B) CXCL6 is highly expressed in tumor tissue, based on TCGA. C) High
   expression of CXCL6 is associated with poorer OS, based on TCGA. D)
   Western blot analysis of CXCL6 expression in CCA tumor and paired
   non‐tumor tissues. E) RT‐qPCR analysis of CXCL6 expression in CCA tumor
   and paired non‐tumor tissues. F) Comparison of CXCL6 expression in
   tumor and paired non‐tumor tissues in the “Surgery cohort” TMA. G)
   Representative images of high/low CXCL6 expression specimens in the
   “Surgery cohort” TMA; scores were calculated based on the intensity and
   percentage of stained cells. H,I) Upregulation of CXCL6 in CCA was
   correlated with worse OS and EFS in the “Surgery cohort” TMA. J)
   Western blot analysis showing that knocking‐down or overexpression of
   CXCL6 affects the expression of E‐cadherin, N‐cadherin, vimentin, and
   PCNA in CCA cells.

   Pan‐cancer analysis based on TCGA data revealed
   elevated CXCL6 expression across multiple malignancies, especially in
   CCA (P ≤ 0.01, Figure [85]S1C, Supporting Information). We corroborated
   this finding by comparing 10 paired surgical resection tissue specimens
   from CCA patients in Zhongshan Hospital. Western blot and RT‐qPCR
   results showed elevated CXCL6 levels in tumor tissues versus paired
   non‐tumor tissue (Figure [86]1D,E). Differential tumor expression was
   further confirmed by IHC staining of the “Surgery cohort” tissue
   microarray (TMA) from Zhongshan Hospital, allowing us to reach the same
   conclusion that CXCL6 is overexpressed in tumor tissue (P < 0.001,
   Figure [87]1F). Using the same stained TMA to analyze the clinical
   significance of CXCL6 (Figure [88]1G) revealed that patients with
   higher expression had significantly worse OS (P = 0.028, Figure [89]1H)
   and EFS (P = 0.046, Figure [90]1I).

   To explore its effects on tumorigenesis at the cellular level, we first
   quantified CXCL6 protein levels in a panel of five common CCA cell
   lines (Figure [91]S1D, Supporting Information). HuCCT1 and RBE cells
   exhibited the highest CXCL6 expression, whereas CCLP1 showed the
   lowest. Thus, we chose to knock down CXCL6 in HuCCT1 and RBE cell lines
   and overexpress it in CCLP1 cells. We established sh‐CXCL6 (CXCL6
   knockdown via three targeting and one nontargeting control [NC] shRNAs)
   and CXCL6‐OE (CXCL6 overexpression via plasmids) cell lines and
   confirmed the transfection efficiency of lentivirus and plasmid by
   western blot (Figure [92]S1E, Supporting Information). Western blot
   analysis of knockdown and overexpression cell lines revealed that CXCL6
   is associated with metastasis (E‐cadherin and N‐cadherin), invasion
   (vimentin), and proliferation (PCNA) in CCA (Figure [93]1J).

2.2. CXCL6 Promotes CCA Development and Sensitizes Gemcitabine Effects Both
In Vitro and In Vivo

   We next conducted functional assays in vitro to evaluate the role of
   CXCL6 in cellular proliferation and CCA development. Silencing CXCL6
   impaired the clonogenic growth of HuCCT1 (P < 0.001) and RBE
   (P < 0.001) cells, while CXCL6 overexpression increased the colony
   formation of CCLP1 cells (P < 0.001; Figure [94]2A). CCK‐8 cell
   proliferation experiments showed the same trend (Figure [95]2B), as did
   EdU incorporation assays to verify the impact of CXCL6 knockdown or
   overexpression on cellular proliferation (Figure [96]2C). Wound healing
   and transwell assays with the sh‐CXCL6 and CXCL6‐OE lines jointly
   demonstrated that CXCL6 has positive effects on the migration of CCA
   cells (Figure [97]2D,E). To investigate its association with
   angiogenesis, we collected conditioned medium from CCA cell cultures
   and used this to culture HUVECs in a tube formation assay. ELISA
   confirmed reduced CXCL6 levels in sh‐CXCL6 supernatants and elevated
   levels in CXCL6‐OE supernatants (Figure [98]S2A–C, Supporting
   Information). When cultured with sh‐CXCL6 conditioned medium, the tube
   formation capacity of HUVECs was significantly attenuated, whereas it
   was enhanced in the presence of CXCL6‐OE conditioned medium (Figure
   [99]S2D, Supporting Information).

Figure 2.

   Figure 2
   [100]Open in a new tab

   CXCL6 promotes CCA proliferation, migration, and angiogenesis in vitro.
   A) Colony formation assay with HuCCT1, RBE, and CCLP1 cell lines. B,C)
   CCK‐8 and EdU proliferation assays were used to assess the effect of
   CXCL6 overexpression. D,E) Wound healing and transwell assays indicated
   that CXCL6 promotes metastasis of CCA cells.

   We subsequently performed in vivo analysis in BALB/c nude mice to
   assess the function of CXCL6, using sh‐NC or sh‐CXCL6 HuCCT1 cells to
   focus specifically on its impact on tumor cells. The results showed
   that CXCL6 depletion significantly reduced tumor weight (P < 0.001) and
   volume (P = 0.001; Figure [101]3A,B; Figure [102]S3A, Supporting
   Information). H&E and IHC staining confirmed decreased malignancy
   (Ki‐67) and angiogenesis (CD31, VEGF) in sh‐CXCL6 tumors
   (Figure [103]3C). The above results together illustrate that CXCL6 may
   significantly impact the malignancy, proliferation, and angiogenesis in
   CCA tumors.

Figure 3.

   Figure 3
   [104]Open in a new tab

   CXCL6 regulates CCA proliferation and migration through the
   JAK‐STAT/PI3K‐AKT pathway. A,B) Representative images and
   quantification of subcutaneous HuCCT1 xenograft tumors established in
   BALB/c nude mice, with or without CXCL6 knockdown. C) Expression levels
   of CK7, Ki67, CD31, VEGF, and CXCL6 in xenograft tumors were determined
   by IHC staining. D) Volcano plot of differential gene expression
   between sh‐NC and sh‐CXCL6 HuCCT1 cells. E) Results of KEGG pathway
   enrichment analysis based on RNA‐Seq of CCA cells (i.e., sh‐CXCL6 vs
   sh‐NC HuCCT1 cells). F,G) GSEA results based on the RNA‐Seq data.

   Since GEM‐based chemotherapy is still a first‐line choice in
   unresectable CCA,^[ [105]^28 ^] we next asked whether CXCL6 could
   modulate its therapeutic efficacy. In vitro, CXCL6 knockdown synergized
   with GEM to suppress proliferation (Figure [106]S3B,C, Supporting
   Information) and migration (Figure [107]S3D,E, Supporting Information)
   of CCA cells. Treating CCA cell lines with 200 nm GEM indicated (via
   γ‐H2AX assay) that CXCL6 knockdown aggravated drug‐induced DNA damage
   in both HuCCT1 and RBE cells (Figure [108]S4A, Supporting Information).
   This sensitization effect of CXCL6 knockdown was confirmed in vivo by
   establishing a mouse xenograft model. As before, sh‐NC and sh‐CXCL6
   HuCCT1 cells were injected subcutaneously in BALB/c nude mice, and
   details of the dosing regimen are shown in Figure [109]S4B (Supporting
   Information). While GEM treatment clearly reduced tumor weight and
   tumor volume, silencing of CXCL6 further enhanced its efficacy (Figure
   [110]S4C–E, Supporting Information). Taken together, the foregoing
   results illustrate that silencing CXCL6 may overcome GEM resistance in
   CCA.

2.3. CXCL6 Regulates CCA Proliferation and Migration through the
JAK‐STAT/PI3K‐AKT Pathway

   To explore potential mechanisms underlying CXCL6‐mediated regulate
   proliferation and migration in CCA, RNA‐Seq was performed on sh‐NC and
   sh‐CXCL6 HuCCT1 cells, identifying 233 significantly upregulated and
   171 significantly downregulated genes (with q‐value ≤ 0.05 and
   log|FC| ≥ 1; Figure [111]3D). Applying KEGG pathway analysis identified
   five significantly enriched pathways, among which the PI3K‐AKT and
   JAK‐STAT pathways were most significantly enriched in sh‐CXCL6 cells
   versus sh‐NC cells (Figure [112]3E). GSEA similarly highlighted
   significant enrichment of the “PI3K‐AKT signaling pathway” and
   “JAK‐STAT signaling pathway” (Figure [113]3F,G). The JAK‐STAT pathway,
   activated by cytokines,^[ [114]^29 ^] has been implicated in
   carcinogenesis and development in multiple tumors^[ [115]^30 , [116]^31
   ^] and is also able to regulate bile duct cell proliferation and
   function.^[ [117]^32 ^] Similarly, PI3K‐AKT is a prominent signal
   transduction pathway activated in both solid and hematologic tumors.^[
   [118]^33 ^] In CCA, this pathway is known to play a central role in
   proliferation, cell cycle regulation, and metabolism.^[ [119]^34 ^] We
   therefore assumed that CXCL6 may exert its effects in CCA through
   JAK‐STAT and PI3K‐AKT signaling.

   To validate the pathway enrichment results, we applied western blot
   with sh‐CXCL6 HuCCT1 and RBE cells and CXCL6‐OE CCLP1 cells to detect
   crucial factors in both pathways, observing reduced activation with
   CXCL6 knockdown and increased pathway activity upon CXCL6
   overexpression (Figure [120]S5A,B, Supporting Information). To seek
   further confirmation, cells were treated with the JAK1/2 inhibitor
   ruxolitinib or the PI3K inhibitor 3‐methyladenine (3‐MA). While CXCL6
   overexpression enhanced in vitro migration (Figure [121]S6A,B,
   Supporting Information) and proliferation (Figure [122]S6C–E,
   Supporting Information) in these cells, both effects were weakened by
   the addition of ruxolitinib or 3‐MA. These data supported the
   involvement of the JAK‐STAT and PI3K‐AKT pathways in CXCL6 oncogenic
   function.

2.4. CXCL6 Activates the CXCR1/2‐JAK‐STAT/PI3K Axis in CCA

   We further elucidated how CXCL6 regulates JAK‐STAT and PI3K‐AKT
   signaling in CCA by considering that CXCL6, as a secretory chemokine,
   must bind to its membrane surface receptors CXCR1/2 to regulate
   cellular functions.^[ [123]^35 ^] Expression of both receptors has been
   reported on neutrophils, endothelial cells, and tumor cells,^[ [124]^36
   , [125]^37 ^] where they are involved in neutrophil recruitment,
   inflammatory disease, and tumor progression.^[ [126]^37 , [127]^38 ,
   [128]^39 ^] The JAK family proteins can bind to and activate cytokine
   receptors by phosphorylating those receptors upon ligand binding,
   leading to the formation of dimeric Janus kinase cytokine receptor
   complexes^[ [129]^40 ^] that may further recruit and phosphorylate
   proteins like STAT or PI3K.^[ [130]^41 , [131]^42 ^] Based on this, we
   hypothesized that in CCA cells, CXCL6 might exert its observed effects
   through the CXCR1/2‐JAK‐STAT/PI3K axis. While the binding of JAKs to
   CXCR1/2 has not previously been proven, we conducted the following
   experiments to test our hypothesis.

   Immunofluorescence staining in paraffin‐embedded CCA tissue revealed
   strong spatial colocalization of JAK1/CXCR1 and JAK1/CXCR2 (Figure
   [132]4A). Co‐IP assays with JAK1 as bait revealed interactions with
   CXCR1 and CXCR2 in HuCCT1 and RBE cells (Figure [133]4B). Reciprocal
   Co‐IP with CXCR1/2 as bait further validated these interactions
   (Figure [134]4C,D). We also applied molecular docking to evaluate
   possible binding between JAK1 and the CXCL6 receptors. Using GRAMM,^[
   [135]^43 ^] we conducted semiflexible docking between JAK1/CXCR1 and
   JAK1/CXCR2; the results showed stable predicted binding of JAK1 to both
   receptors (Figure [136]S7A,B, Supporting Information). The computed
   JAK1/CXCR1 binding energy was −35.1 kcal mol^−1 and the interface area
   was 3523 Å^2 (Figure [137]S7C, Supporting Information), while the
   equivalent values for JAK1/CXCR2 were −8.1 kcal mol^−1 and 3130.9 Å^2,
   respectively (Figure [138]S7D, Supporting Information). Overall, the
   above results indicate the binding of JAK1 to both CXCL6 receptors. To
   functionally validate receptor involvement, we employed reparixin and
   SB225002 as selective antagonists for CXCR1 and CXCR2, respectively.^[
   [139]^44 , [140]^45 ^] Treatment with either inhibitor impaired
   phosphorylation of JAK (Figure [141]S7E, Supporting Information). In
   CXCL6‐OE CCLP1 cells, reparixin and SB225002 were able to inhibit
   malignancy (Figure [142]S7F, Supporting Information), proliferation
   (Figure [143]S7G–I, Supporting Information), and migration (Figure
   [144]S7J,K, Supporting Information).

Figure 4.

   Figure 4
   [145]Open in a new tab

   CXCL6 functions through activating the CXCR1/2‐JAK‐STAT/PI3K axis in
   CCA. A) Representative images and analysis plots from
   immunofluorescence experiments, demonstrating colocalization of
   CXCR1/JAK1 and CXCR2/JAK1 in CCA tissue. B) Co‐IP results using JAK1 as
   bait indicated interactions between JAK1/CXCR1 and JAK1/CXCR2. C,D)
   Co‐IP results using CXCR1/2 as bait supported CXCR1/JAK1 and CXCR2/JAK1
   interactions in HuCCT1 and RBE cells. E) HIF‐1α IHC staining of nude
   mouse tumor tissue.

   Since it was reported that the JAK‐STAT and PI3K‐AKT pathways both
   activate HIF‐1α expression,^[ [146]^46 , [147]^47 ^] we applied IHC
   staining in nude mouse tumors and found that HIF‐1α expression was
   lower in sh‐CXCL6 cells (Figure [148]4E). In conclusion, the
   CXCR1/2‐JAK‐STAT/PI3K axis is a potential mechanism through which CXCL6
   plays its role in CCA.

2.5. CXCL6 Reshapes CCA Lipid Metabolism

   The JAK‐STAT and PI3K‐AKT pathways regulate biological processes in
   tumor cells in various ways.^[ [149]^48 , [150]^49 ^] Notably, the
   JAK‐STAT pathway was shown to boost de novo fatty acid synthesis and
   fatty acid uptake in the TIME,^[ [151]^48 , [152]^50 ^] while PI3K‐AKT
   activation can cause lipid peroxidation in gastric cancer.^[ [153]^51
   ^] To explore these possibilities in CCA, we applied GO‐BP (Gene
   Ontology – Biological Processes) analysis to the RNA‐Seq data mentioned
   above and noted that lipid metabolism‐related pathways were most highly
   enriched (Figure [154]5A). Subsequent BODIPY and Oil Red O staining of
   murine CCA xenografts demonstrated that CXCL6 knockdown markedly
   reduced intratumoral lipid accumulation (Figure [155]5B,C), suggesting
   CXCL6‐driven lipid metabolic remodeling

Figure 5.

   Figure 5
   [156]Open in a new tab

   CXCL6 reshapes CCA lipid metabolism. A) GO pathway enrichment of
   RNA‐Seq data from CCA cells (sh‐NC vs sh‐CXCL6 HuCCT1 cells). B,C)
   BODIPY and Oil Red O staining of murine xenograft CCA tumors. D)
   Volcano plot of differential metabolites in LC‐MS lipidomics. E) Heat
   map of the top 30 differential metabolites (sh‐NC vs sh‐CXCL6 HuCCT1
   cells). F) KEGG analysis based on the top 30 differential metabolites.
   G) Network of these metabolites and their association with cellular
   pathways.

   Applying LC–MS lipidomics to analyze sh‐NC and sh‐CXCL6 HuCCT1 cells
   allowed us to identify 342 differential metabolites (Figure [157]5D).
   We listed the top 30 metabolites with the largest differences and found
   that in sh‐CXCL6 cells, the most downregulated metabolites belonged to
   the phosphatidylcholine class (Figure [158]5E); phosphatidylcholines
   are reported to promote tumor cell proliferation and impair CD8^+ T
   cell function.^[ [159]^52 , [160]^53 ^] KEGG pathway analysis of these
   metabolites highlighted enrichment in cancer‐associated lipid pathways,
   including “Glycerophospholipid metabolism,” “Linoleic acid metabolism,”
   and “Arachidonic acid metabolism”^[ [161]^54 , [162]^55 , [163]^56 ^]
   (Figure [164]5F). Constructing a network revealed that “DG 12:0/14:0,”
   “DG (18:0/20:1/0:0),” and “DG (18:0/0:0/18:0)” were linked to the most
   pathways (Figure [165]5G). Importantly, diacylglycerol (DG) is known as
   a double‐edged sword affecting tumor progression and can also regulate
   the TIME by activating CD8^+ T cells and enhancing the cytotoxicity of
   natural killer (NK) cells.^[ [166]^57 , [167]^58 ^] In sum, CXCL6 may
   significantly affect the content of key metabolites in CCA, which can
   further regulate tumor progression and TIME in return.

2.6. CXCL6 Expression Correlates with Immunotherapy Resistance and TAN
Infiltration in CCA

   The above results indicate that CXCL6 exerts direct regulation on CCA
   cells and could therefore affect GEM treatment sensitivity, leading us
   to ask whether CXCL6 expression levels are associated with CCA
   immunotherapy response. Through IHC staining of the “Conversion therapy
   cohort” TMA, we found that CXCL6‐low patients (50% cutoff) benefited
   more from immunotherapy, experiencing longer OS and EFS (Figure
   [168]6A–C). After collecting baseline peripheral blood from 26 CCA
   patients prior to immunotherapy (Table [169]S3, Supporting
   Information), serum CXCL6 levels were measured by ELISA and showed that
   nonresponders tended to have higher serum concentrations of CXCL6
   (Figure [170]6D). Higher CXCL6 levels were also associated with shorter
   OS and EFS after immunotherapy (Figure [171]6E,F). An animal
   subcutaneous xenograft study based on CCA cells of murine origin
   (mIC‐23) was conducted to corroborate the clinical findings, in
   C57BL/6J mice according to the experimental design shown in
   Figure [172]6G. CXCL6 expression was knocking down using siRNAs, while
   si‐CXCL6‐3 was chosen for in vivo experiment (Figure [173]S8A,
   Supporting Information). Silencing of CXCL6 boosted the efficacy of
   anti‐PD‐1 therapy, significantly reducing tumor weight and volume
   (Figure [174]6H–J; Figure [175]S8B, Supporting Information).

Figure 6.

   Figure 6
   [176]Open in a new tab

   CXCL6 expression is associated with CCA immunotherapy resistance and
   TAN infiltration level. A) Representative images of CXCL6 IHC staining
   in the “Conversion therapy cohort” TMA. B,C) OS, EFS analysis. D) ELISA
   measurements of CXCL6 in baseline peripheral serum. E,F) Comparison of
   OS and EFS between CXCL6‐high and CXCL6‐low patients. G) Experimental
   design and dosing regimen for the C57BL/6J mouse subcutaneous tumors.
   H–J) Representative images of subcutaneous tumors along with
   quantification of tumor weight and volume. K) Representative images of
   CXCL6 and CD10 IHC staining in the TMA. L–M) Linear regression analysis
   of CXCL6 and CD10 expression in the “Surgery cohort” and “Conversion
   therapy cohort” TMAs. N) Workflow for neutrophil isolation, coculture,
   and RNA‐Seq analysis.

   All in all, these results strongly indicate that CXCL6 may lead to CCA
   immunotherapy resistance, so we further explored the potential
   mechanisms. Specifically, CXCR1/2 is expressed on TANs where they
   regulate TIME recruitment and NETs formation.^[ [177]^59 , [178]^60 ^]
   Considering that CXCL6 is a chemokine that binds to these receptors, we
   studied its influence on TANs, applying IHC staining to detect the
   relationship between CXCL6 and TANs (CD10^+) infiltration in two CCA
   TMAs. In both the “Surgery” and “Conversion therapy” cohorts, CXCL6
   protein levels were positively associated with TANs density (R = 0.422,
   P < 0.001; R = 0.420, P = 0.015, Figure [179]6K–M). We also analyzed
   the correlation of CXCL6 with infiltration of other immune cells (T
   cells, macrophages, and Tregs), but no significant relationship was
   apparent in any case (Figure [180]S8C–G, Supporting Information).

2.7. CXCL6 Promotes NETs Formation and Impairs CD8^+ T Cell Infiltration

   To explore the effects of CXCL6 on neutrophils, peripheral neutrophils
   were isolated from blood and cocultured with sh‐NC or sh‐CXCL6 HuCCT1
   cells (Figure [181]6N); TANs were identified using flow cytometry as
   previously reported^[ [182]^61 ^] (Figure [183]S8H, Supporting
   Information). By RNA‐Seq, we then identified 240 differently expressed
   genes (with log[2]|FC| ≥ 1, q‐value ≤ 0.05) between TANs cultured at
   the two different concentrations of CXCL6 (Figure [184]S9A, Supporting
   Information).

   Applying KEGG analysis showed that the “MAPK signaling pathway” and
   “Ras signaling pathway” ranked highest among differentially expressed
   genes (Figure [185]7A), while GO analysis revealed “Extracellular
   region” and “Extracellular space” as the two highest‐ranked “Cellular
   component” terms (Figure [186]7B). Activation of the RAS/MAPK pathway
   in CXCL6‐exposed neutrophils was verified by western blot (Figure
   [187]S9B, Supporting Information). This pathway was reported to be a
   key upstream regulator of NETs formation,^[ [188]^62 , [189]^63 ^] and
   CXCR1/2 are apparently also involved.^[ [190]^64 ^] Thus, our results
   strongly suggested a role for CXCL6 in NETs formation derived from
   TANs. mIHC staining of murine mIC‐23 tumors revealed that CXCL6
   knockdown greatly reduced the formation of NETs (Figure [191]7C),
   confirming this as a potential relevant mechanism. Considering that the
   generation of reactive oxygen species (ROS) is the key starting point
   for NET formation,^[ [192]^65 ^] we measured ROS levels in TANs after
   coculturing with HuCCT1 cells and found that ROS levels decreased in
   sh‐CXCL6 groups (Figure [193]S9C,D, Supporting Information). Taken
   together, the above evidence reveals that CXCL6 may mediate the
   formation of NETs in the CCA TIME.

Figure 7.

   Figure 7
   [194]Open in a new tab

   CXCL6 promotes NETs formation and inhibits CD8^+ T cell infiltration in
   the TIME. A) KEGG enrichment analysis of significantly differentially
   expressed genes in TANs cocultured with sh‐NC versus sh‐CXCL6 HuCCT1
   CCA cells. B) GO enrichment analysis of the same differentially
   expressed genes. C) mIHC staining of two NETs markers, neutrophil
   elastase (NE) and myeloperoxidase (MPO), in murine C57BL/6J tumors.
   D–F) mIHC staining for CD4, CD8, and CK7 in murine C57BL/6J tumors and
   statistical analysis. G) Experimental design and dosing regimen for the
   rescue experiment. H) Flow cytometry of CD8^+CD45^+ in rescue
   experiment.

   A previous study illustrated how NETs contribute to blockage of tumor T
   cell infiltration,^[ [195]^66 ^] so we also analyzed T cell populations
   in murine C57BL/6J tumors by mIHC staining (Figure [196]7D). Analysis
   of the data showed no significant difference in CD4^+ T cell between
   “si‐NC+IgG”, “si‐CXCL6+IgG” and ‘si‐NC+Anti‐PD‐1′ groups
   (Figure [197]7E), while ‘si‐CXCL6+Anti‐PD‐1′ showed an increase. As for
   CD8^+ T cell, knocking down of CXCL6 and anti‐PD‐1 treatment slightly
   increased the number of CD8^+ T cell, the combination treatment showed
   the strongest recruitment of CD8^+ T cell (Figure [198]7F). Since CD8^+
   T cell exhaustion reportedly causes immunotherapy resistance in CCA,^[
   [199]^67 ^] we concluded that CXCL6 may underlie this resistance by
   reducing CD8^+ T cell infiltration through NETs formation.

   To further confirm the interaction between NETs, formation and CD8^+ T
   cell infiltration, we conducted a rescue experiment as depicted in
   Figure [200]7G. Cl‐amidine was chosen to weaken the formation of NETs
   as reported.^[ [201]^68 , [202]^69 ^] which was a direct inhibitor of
   PAD4, a key inducer of NETs.^[ [203]^70 ^] Cl‐amidine reversed
   CXCL6‐driven tumor growth (Figure [204]S9E, Supporting Information).
   Based on flow cytometry results, we found that the overexpression of
   CXCL6 greatly rose the formation of NETs, shown by ROS levels (Figure
   [205]S9F, Supporting Information). At the meantime, the infiltration
   numbers of CD8^+T cell were significantly reduced (Figure [206]7H). It
   was more persuasive that Cl‐amidine could reverse the effects of CXCL6
   on NETs, as well as restore the infiltration of CD8^+T cell. We also
   examined the IFNγ (Figure [207]S9G, Supporting Information) and GzmB
   (Figure [208]S9H, Supporting Information) levels in CD8^+T cell,
   results showed that NETs could reduced the killing effects of CD8^+T
   cell. The same conclusion could be observed from the mIHC staining of
   MPO, NE, and CD8 based on the same samples of tumor tissue (Figure
   [209]S10A, Supporting Information). In all, the above results together
   show that CXCL6 induces the formation of NETs in CCA, which blocks the
   infiltration and killing function of CD8^+T cell and eventually causes
   immunotherapy resistance.

3. Discussion

   Accumulating studies suggest the involvement of the CXCLs cytokine
   family in the progression of multiple malignancies.^[ [210]^71 ^] As
   secreted factors, CXCLs can target a variety of cell types to exert
   different functions, and their roles have been particularly studied in
   HCC. For instance, Wu et al. reported that CXCL6 overexpression in HCC
   cells activates the JAK‐STAT3 pathway in hepatocytes through paracrine
   signaling.^[ [211]^72 ^] Among liver cancers, the incidence of CCA is
   second only to HCC, yet the molecular mechanisms of CXCLs in CCA have
   barely been explored.

   Our study identifies CXCL6 as a multifunctional oncoprotein in
   CCA. First, we discovered CXCL6 as a pro‐angiogenic factor, observing
   that it can raise VEGF levels in CCA (Figure [212]3C) and increase
   HIF‐1α expression (Figure [213]4E). As well as inducing a hypoxic
   environment, HIF‐1α acts as an upstream regulator of vascular growth
   factors.^[ [214]^73 ^] Second, we found that high CXCL6 expression in
   CCA cells accelerates tumor progression and metastasis through
   autocrine mechanisms. We also tested whether CXCL6 silencing could
   sensitize cells to GEM, given that this is still a first‐line therapy
   for CCA. In vivo and in vitro experiments showed that GEM had stronger
   cytotoxic effects on CXCL6‐depleted CCA cells (Figure [215]S3B–E;
   Figure [216]S4, Supporting Information).

   Gene enrichment analysis indicated that metabolic balance in CCA might
   be altered by CXCL6, especially lipid metabolism (Figure [217]5A).
   Lipid metabolism is one of the most prominent metabolic changes in
   cancer^[ [218]^74 ^] and can generate the energy, biofilm components,
   and signaling molecules required for tumor cell proliferation,
   invasion, and metastasis while influencing the tumor microenvironment
   and responses to cancer treatments.^[ [219]^75 ^] Lipidomic analysis
   enabled us to profile CXCL6‐related lipid metabolism changes in CCA and
   uncovered several core metabolites (Figure [220]5E–G), in particular
   the DGs ‘DG 12:0/14:0’, ‘DG (18:0/20:1/0:0)’, and ‘DG (18:0/0:0/18:0)’,
   which were found at the center of the metabolic network.

   The DG metabolite class has a relatively positive function in immune
   cells in the TIME. In T cells, activation of the T cell receptor (TCR)
   triggers the production of DGs and stimulates the transcription of
   genes related to T cell expansion and cytotoxicity.^[ [221]^76 ,
   [222]^77 ^] However, DG enrichment is considered a hallmark of
   hyporesponsive infiltrating CD8^+ T cells,^[ [223]^78 ^] while genetic
   ablation of DG kinases (DGKs) enhances the cytotoxic activity of NK
   cells.^[ [224]^58 ^] While targeted inhibitors for DGKα and DGKζ have
   been developed and investigated preclinically for tumor
   immunotherapy,^[ [225]^79 ^, [226]^80 ^] the precise functions of DGs
   in tumor cells remain controversial.^[ [227]^81 ^] The downstream
   pathways and metabolites of DGs are so complex that more research is
   needed to obtain a proper understanding. In this study, we were unable
   to clarify the detailed functions of the three core DG metabolites but
   may analyze this in subsequent studies.

   The CXCL6 receptors, CXCR1 and CXCR2, are expressed on tumor and immune
   cells, and both are important in CXCL6 function.^[ [228]^82 ^] We
   revealed that JAKs could be activated by CXCR1/2 and further
   phosphorylate STAT3 and PI3K proteins. To our knowledge, this is the
   first study to directly demonstrate the association between CXCR1/2 and
   JAKs, which could help clarify the effects of CXCL6 on CCA cells. In
   addition, CXCR1/2 are important regulator of TANs infiltration.^[
   [229]^59 ^] Xiong et al. detected CXCL1/2‐induced NETs in esophageal
   squamous cell carcinoma and found that this augmented NK cells.^[
   [230]^83 ^] Furthermore, NETs induced by IL‐8 via CXCR1/2 can promote
   gastric cancer progression.^[ [231]^84 ^] Our RNA‐Seq results with
   cocultured neutrophils strongly suggested that NETs formation is
   influenced by CXCL6, with the NETs‐associated RAS/MAPK pathway
   participating in this process as reported.^[ [232]^62 , [233]^63 ^]
   Thus, we hypothesized that CXCL6 might regulate CCA immunotherapy
   efficiency and found, using baseline peripheral blood samples, that
   higher initial CXCL6 levels led to a worse prognosis
   (Figure [234]6D–F). Thus, CXCL6 may function as a biomarker for
   immunotherapy. We also conducted animal experiments in which CXCL6
   silencing in C57BL/6J tumors significantly enhanced the therapeutic
   efficacy of an anti‐PD‐1 mAb (Figure [235]6H–J). Changes in the TIME
   were also detected upon CXCL6 knockdown. Immunofluorescence assays
   confirmed that CXCL6 depletion reduced the formation of NETs and
   increased infiltration of CD8^+ T cell. A rescue experiment was further
   conducted to clarify this assumption. As Cl‐amidine inhibited the NETs
   level caused by CXCL6 overexpression, the infiltration and killing
   effects of CD8^+ T cell were also raised, which could help establish
   the causality between CXCL6 expression, NETs formation, and CD8+ T cell
   infiltration. This conclusion also mirrors other reports in pancreatic,
   breast, and bladder cancers, where NETs have been shown to suppress
   CD8^+ T cell recruitment to the TIME.^[ [236]^70 , [237]^85 , [238]^86
   ^] Meanwhile, CD8^+ T cell exhaustion was reported to mediate
   immunotherapy in CCA,^[ [239]^87 ^] as well as other cancers including
   melanoma, colorectal cancer, and bladder cancer.^[ [240]^88 , [241]^89
   , [242]^90 ^] Hence, we propose that high CXCL6 expression in CCA
   reduces CD8^+ T cell infiltration through NETs formation, eventually
   causing immunotherapy resistance.

   Through multimodal approaches (Figure [243]8A), we delineate CXCL6 as a
   dual driver of CCA progression and immune evasion. In conclusion, the
   present study has demonstrated CXCL6 as a dual regulator of CCA
   progression and immunotherapy resistance. Mechanistic investigations
   showed that CXCL6 activates the CXCR1/2‐JAK‐STAT/PI3K axis in CCA cells
   through autocrine signaling and reprograms lipid metabolism balance.
   Concurrently, CXCL6 induces NETs formation via RAS/MAPK signaling,
   which impedes CD8^+ T cell infiltration and confers immunotherapy
   resistance (Figure [244]8B). Targeting CXCL6 or its downstream
   effectors may offer a novel sensitizing strategy to augment
   chemotherapy and immunotherapy responses in CCA patients.

Figure 8.

   Figure 8
   [245]Open in a new tab

   Graphical abstract of the key mechanisms and experimental workflows. A)
   Experimental workflows of the study. B) Proposed mechanisms of CXCL6 in
   CCA. Figures were created with BioRender.com.

4. Experimental Section

Cell Culture and Extraction

   The HuCCT1, RBE, CCLP1, HCCC‐9810, QBC939, and HUVEC cell lines were
   obtained from the Liver Cancer Institute, Fudan University. The mouse
   CCA cell line mIC‐23 was established in the Liver Cancer Institute of
   Fudan University, as reported.^[ [246]^65 ^] Human CCA cell lines were
   cultured in RPMI 1640 medium containing 10% fetal bovine serum (FBS;
   Gibco, A5670701) and 1% penicillin/streptomycin (Gibco, 15140148) at
   37 °C, in an atmosphere containing 5% CO[2] in an incubator. The mIC‐23
   cell line was cultured in DMEM supplemented with 10% FBS (Gibco,
   A5670701) and 1% penicillin/streptomycin (Gibco, 15140148), with
   incubation as above. HUVEC cell line was cultured in an endothelial
   cell medium (SclenCell, 37969), with incubation as above.

   Peripheral blood (20 mL per person) was collected from three healthy
   volunteers into EDTA‐coated tubes. Whole blood was centrifuged at
   300 × g for 10 min and the middle layer was taken for isolation, then
   neutrophils were isolated using the MojoSort Whole Blood Human
   Neutrophil Isolation Kit (BioLegend, 480152). Antibodies and magnetic
   beads were added to the middle layer in sequence, and each was
   incubated for more than 20 min. Then the tube was placed in the magnet
   for 5 min, and the cells poured out were isolated neutrophils. The
   obtained neutrophils were cultured as for the human CCA cell lines (see
   above). Where indicated, neutrophils were cocultured with HuCCT1 cells
   (sh‐NC/sh‐CXCL6 transfected; see below) at a ratio of 10:1.

Lentivirus, Plasmids, and siRNA Infection

   Cells were infected with lentivirus to deliver shRNA‐control (sh‐NC) or
   shRNA‐CXCL6 (sh‐CXCL6) constructs, using lentiviral preparations
   purchased from Genomeditech (Shanghai, China). Plasmids for CXCL6
   overexpression and small interfering RNA (siRNA) for CXCL6 knockdown
   were sourced from the same supplier. Cells were transfected with
   plasmids and siRNA using Lipofectamine 3000. The proper infection ratio
   was tested before the formal experiment and drug screening was
   conducted after infection. The infection efficiency was confirmed with
   a Western blot afterward.

CCA Patients and Tissue Microarray

   The “Surgery Cohort” included 192 CCA patients who underwent surgery
   without pretreatment from January 2018 to December 2020 at the
   Zhongshan Hospital, Fudan University, Shanghai, China. The “Conversion
   Therapy Cohort” included 33 CCA patients who underwent surgery after
   immunotherapy from January 2019 to December 2022 at the same hospital.
   Written informed consent was obtained from all patients prior to
   involvement in the study. The study protocol was approved by the Ethics
   Committee of Zhongshan Hospital (Approval Number B2020‐177R) and was in
   accordance with the Declaration of Helsinki. Clinicopathological
   features of patients in “Surgery Cohort” and “Conversion Therapy
   Cohort” tissue microarray are shown in Tables [247]S1,S2 (Supporting
   Information). For both cohorts, OS was calculated from the surgery or
   first cycle of immune‐based therapy to death due to any cause, or
   censored at the last follow‐up. Disease‐specific survival (DSS) was
   calculated from the surgery or first cycle of immune‐based therapy to
   death due to CCA, or censored at the last follow‐up. Event‐free
   survival (EFS) was calculated from the surgery or first cycle of
   immune‐based therapy to the first occurrence of any of the following
   events: disease progression, local or distant recurrence, death due to
   any cause, or censored at the last follow‐up, etc.

   A paraffin‐embedded CCA TMA was constructed and IHC analysis was
   performed as described below. The antibodies used for IHC staining
   were: anti‐CXCL6 (Invitrogen, PA5‐115276, 1:100), anti‐CD45
   (Servicebio, [248]GB113886‐50), anti‐CD3 (Servicebio, GB13014‐50),
   anti‐CD8a (Servicebio, [249]GB115692‐50), anti‐CD10 (Servicebio,
   [250]GB121120‐50, 1:100), anti‐FOXP3 (Servicebio; #[251]GB112325‐50,
   1:500), and anti‐CD68 (Servicebio, GB11093‐1‐50, 1:100).

Animal Studies

   Male six‐week‐old C57BL/6J and BALB/c nude mice were purchased from
   Charles River (Shanghai, China). Animal care and experimental protocols
   were approved by the Institutional Animal Care and Use Committee
   (IACUC) of Zhongshan Hospital, Fudan University (20240206‐443), and all
   mice were fed in a specific pathogen‐free facility. To establish
   tumors, BALB/c nude mice were injected subcutaneously in the right
   underarm with 5 × 10^6 HuCCT1 cells in RPMI 1640 serum‐free medium
   (200 µL), while C57BL/6J mice were injected in the same manner with
   1 × 10^7 mIC‐23 cells. Two weeks after injection, mice bearing
   subcutaneous tumors were randomized into groups as indicated below. For
   tumor‐bearing BALB/c nude mice, treatments were administered
   intraperitoneally every two days and seven times in total.
   Tumor‐bearing C57BL/6J mice were treated with 10 mg k g^−1 anti‐mouse
   PD‐1 (CD279) (BioXCell, BE0146) or Rat IgG2a isotype control (BioXCell,
   BE0089) by the same route and with the same frequency.

Molecular Docking

   The 3D structures of JAK1 (PDB6GGH), CXCR1 (PDB2LNL), and CXCR2
   (PDB6LFL) proteins were obtained from the RCSB Protein Data Bank
   ([252]www.rcsb.org/). For docking, structures were converted to PDB
   files containing all of the polar residues with hydrogens. The GRAMM
   ([253]https://gramm.compbio.ku.edu/request) semi‐flexible method was
   adopted for protein–protein docking, and each protein pair was uploaded
   to a third‐party website to perform the docking, returning the
   conformation with the lowest binding energy. Models of the resultant
   protein complexes were analyzed using LigPlus and PyMol.

HUVEC Tube Formation Assay

   Matrigel (60 µL; Corning, 356231) was added to each well of a 24‐well
   plate and incubated at 37 °C for 10 min. The HUVEC cell line was
   cultured as mentioned above. Then, HUVEC cells were digested and
   distributed into the plate at 1 × 10^5 cells per well. Cells were
   cultured using a conditioned medium derived from sh‐NC, sh‐CXCL6, and
   CXCL6‐OE tumor cells. Images were taken 12 h later using a basic
   bright‐field microscope.

Flow Cytometry

   The identity of isolated neutrophils was confirmed using flow
   cytometry. Specifically, the collected neutrophils were stained with
   phycoerythrin (PE) conjugated anti‐human CD45 (BioLegend, 368509) and
   FITC‐conjugated anti‐human CD16 (BioLegend, 302005) antibodies at 4 °C
   for 30 min.

   In the rescue experiment, mouse tumor tissue was cut into small pieces
   and placed in appropriate digestive enzymes (TrypLE Express Enzyme,
   ThermoFisher, 12605036). Tissue was digested in a suitable temperature,
   dispersed, and filtered into single‐cell suspension. Then single cell
   suspension was stained by APC‐cy7‐conjungated Zombie (BioLegend,
   423106), percpcy5.5‐conjungated CD45 (BioLegend, 103130),
   APC‐conjungated CD8 (BioLegend, 100712), BV421‐conjungated IFNγ
   (BioLegend, 505830) and PE‐cy7‐conjungated GzmB (BioLegend, 372214).
   Flow cytometry was conducted using a BD FACS Aria II flow cytometer and
   data were analyzed with FlowJo 10.6.2 (FlowJo, LLC).

Western Blot Assay

   Cells and CCA tissue were lysed with RIPA Lysis Buffer (Strong)
   (Yeasen, 20101ES60). Electrophoresis and membrane transfer were
   conducted using SDS–PAGE buffer (Servicebio, G2081‐1L) and transfer
   buffer (Servicebio, G2028‐1L), respectively. Membranes were incubated
   with primary antibodies at 4 °C overnight, washed three times with TBST
   (Servicebio, G0004‐500ML) for 10 min, then incubated with secondary
   antibodies at room temperature for 2 h. Results were detected via the
   FUSION FX Imaging System (Vilber GmbH, France) using the ultrasensitive
   ECL FemtoLight Substrate (Epizyme, SQ201). For sh‐NC and sh‐CXCL6
   transfected HuCCT1 and RBE cells, three repetitions were performed in
   each case for western blot.

Cell Counting Kit‐8 (CCK‐8) Assay

   One thousand cells per well were seeded into 96‐well plates. The assay
   was started by adding 10% CCK‐8 solution (Beyotime, C0037) one day
   later. After incubation in the dark for 1 h, absorbance measurements
   were recorded at 450 nm using a SpectraMax iD3/iD5 microplate reader
   (Molecular Devices, USA). Measurements were repeated every 24 h.

5‐Ethynyl‐2′‐Deoxyuridine (EdU) Assay

   The BeyoClick EdU Cell Proliferation Kit with Alexa Fluor 488
   (Beyotime, C0071S) was used for proliferation assays. Cells in the
   logarithmic growth phase and in good condition were used for the
   experiment. Briefly, cells were treated with 10 µm EdU for 2 h, fixed
   with 4% paraformaldehyde, permeabilized with 0.3% Triton X‐100 for
   15 min, then stained using Click Reaction Buffer containing Azide 555
   for 30 min followed by Hoechst 33342 for 10 min. Images were acquired
   using a THUNDER Imaging System (Leica, Germany).

Colony Formation Assay

   Cells were seeded into 6‐well plates (1000 cells per well) and cultured
   for 15 days. The cell culture medium was changed every three days. For
   analysis, cells were fixed with 4% paraformaldehyde and stained with
   Crystal Violet Staining Solution (1.5 mL; Beyotime, C0121‐100 ml) then
   washed three times with PBS before imaging.

Wound Healing assay

   Cells were seeded into 6‐well plates. Once cultures had reached 95–100%
   confluency, 200 µL pipette tips were used to introduce wounds. Results
   were recorded 3 days later using a microscope. Relative migration rates
   were used to make statistical comparisons.

Transwell Assay

   Cells (2×10^4 per sample) were seeded into the upper chamber in
   serum‐free medium and the lower chamber was filled with growth medium
   containing 20% FBS. After 3 days, cells were fixed and stained with
   Crystal Violet Staining Solution (Beyotime, C0121‐100 ml). Cells on the
   lower surface were photographed after removing the redundant
   (non‐migrated) cells. Migration cell numbers were counted and made
   statistical comparisons.

Co‐Immunoprecipitation (Co‐IP) Assay

   Cells were lysed with Lysis Buffer (Beyotime, P2179S‐1) and lysates
   were incubated with anti‐CXCR1 (Absin, 120403), anti‐CXCR2
   (Proteintech, 20634‐1‐AP), or anti‐JAK1 (Cell Signaling Technology,
   3344T) antibodies at 4 °C overnight. Samples were then incubated with
   protein A+G magnetic beads (Beyotime, P2179S‐4) at room temperature for
   1 h and isolated with a magnet. Immunoprecipitated proteins were
   detected by western blot.

Multicolor Immunohistochemistry (mIHC)

   Paraffin‐embedded tissue samples were deparaffinized, followed by
   citrate buffer antigen retrieval at 95 °C and blocking. Samples were
   then incubated with primary antibody at 4 °C overnight and secondary
   antibody at 37 °C for 30 min. A DAB Color Development Kit (Servicebio,
   G1212‐200T) was used for the chromogenic reaction.

Immunofluorescence

   Paraffin‐embedded tissue samples were deparaffinized and blocked with
   5% BSA. The samples were subsequently incubated with primary antibody
   at 4 °C overnight then fluorescent secondary antibody at room
   temperature for 2 h. Nuclei were stained with DAPI. Images were
   acquired using a THUNDER Imaging System (Leica, Germany). Statistical
   analysis was conducted based on ImageJ software.

Reactive Oxygen Species Detection

   Dichlorodihydrofluorescein diacetate (DCFH‐DA) was diluted with
   serum‐free culture medium to a final concentration of 10 µm.
   Neutrophils were cultured with HuCCT1 cells as mentioned above.
   Neutrophils were then collected, resuspended in the diluted DCFH‐DA,
   and incubated at 37 °C for 20 min. Cells were observed by fluorescence
   microscopy using 488 nm excitation wavelength and 525 nm emission
   wavelength.

RNA Sequencing (RNA‐Seq)

   Using the TRIzol reagent (Beyotime, R0016), total RNA was extracted
   from sh‐NC and sh‐CXCL6 transfected HuCCT1 cells, and from neutrophils
   after coculture with sh‐WT or sh‐CXCL6 transfected HuCCT1 cells. The
   concentration and purity of total RNA were checked using a NanoDrop
   2000 instrument (Thermo Scientific, MA, USA). Then, RNA‐Seq analysis
   was performed by LC‐Biotechnology Co., Ltd., Hangzhou, China.

Gene Ontology (GO) Functional and Kyoto Encyclopedia of Genes and Genomes
(KEGG) Pathway Enrichment Analyses

   To analyze the possible key pathways, GO functional and KEGG pathway
   enrichment analyses were conducted and visualized through the
   “clusterProfler” package based on RNA‐seq data.^[ [254]^91 ^] A
   p‐value < 0.05 was considered statistically significant.

Gene Set Enrichment Analysis (GSEA)

   To confirm the enriched KEGG pathways in CCA cells RNA‐seq, GSEA was
   conducted using the “clusterProfler” package. All genes were ranked
   based on their log[2]FC, and gene sets with adjusted p‐values < 0.05
   were considered statistically significant.

Liquid Chromatography–Mass Spectrometry(LC–MS) Lipidomics

   Collected cells were thawed on ice and metabolites were extracted with
   80% methanol buffer. All samples were analyzed by LC‐MS.
   Chromatographic separations were performed using an UltiMate 3000 UPLC
   System (Thermo Fisher Scientific, Germany) equipped with an ACQUITY
   UPLC T3 column (100 × 2.1 mm, 1.8 µm; Waters, Milford, CT, USA) for
   reverse‐phase separation. A TripleTOF 6600 high‐resolution tandem mass
   spectrometer (SCIEX, Framingham, MA, USA) was used to detect
   metabolites eluted from the column. After the acquisition, MS data
   preprocessing was performed using XCMS software, including peak
   picking, peak grouping, retention time correction, second peak
   grouping, and annotation of isotopes and adducts. The online KEGG and
   human metabolome database resources were used to annotate metabolites
   by matching exact molecular mass data (m/z) from samples with database
   values. Statistical analysis was performed in R (version 4.0.0).

Rescue Experiment

   NC/CXCL6‐OE mIC‐23 cells were constructed with plasmids. A total of
   four groups, “NC+Vector”, “NC+ Cl‐amidine”, “CXCL6‐OE+Vector” and
   “CXCL6‐OE+Cl‐amidine” were constructed. To establish tumors, C57BL/6J
   mice were injected subcutaneously in the right underarm with 5 × 10^7
   mIC‐23 cells in RPMI 1640 serum‐free medium (200 µL). Seven days after
   injection, Cl‐amidine (10 mg k g^−1)^[ [255]^92 ^] or vector was
   administrated intra‐abdominally daily for NETs formation inhibition, 7
   times in total. Acquired tumors were used for flow cytometry tests and
   mIHC staining.

Statistical Analysis

   Experimental data were expressed as mean ± standard deviation (SD) and
   analyzed using GraphPad Prism 10.0 (GraphPad Software, San Diego, CA,
   USA); P‐values were calculated via the Student's or Welch's t‐test for
   continuous variables and the chi‐square test for categorical variables.
   Linear correlations were tested by Pearson correlation analysis.
   Survival analysis used the Kaplan–Meier method and log‐rank tests. In
   the figures, the level of statistical significance is indicated as
   follows: * (P < 0.05), ** (P < 0.01), or *** (P < 0.001).

Ethical Approval and Consent to Participate

   Written informed consent was obtained from all patients before
   inclusion in the study. The study protocol was approved by the Ethics
   Committee of Zhongshan Hospital (Approval Number B2020‐177R), and the
   research was conducted in accordance with the declaration of Helsinki.

   Further materials and methods are provided in the supplementary
   materials and methods.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   T.H., Z.‐Y.W, B.X. and C.‐J.Z. contributed equally as joint first
   authors. Conception, design, and development of methodology were done
   by T.H., Z.‐Y.W., H.‐C.S., and C.H. T.H., Z.‐Y.W., B.X., L.‐N.W.,
   H.‐C.S., C.‐J.Z., Z.‐Y.Y., and S.‐Q.Z. dealt with performing
   experiments, data curation, and analysis. B.H., X.‐D.Z., Y.‐H.S., J.Z.,
   J.F., H.‐C.S., and C.H. dealt with administration, study supervision,
   and material support. Writing and visualization were done by T.H. and
   Z.‐Y.W. Review, editing, and validation were done by B.H., X.‐D.Z.,
   Y.‐H.S., H.‐C.S., and C.H. All authors read and approved the final
   version of the manuscript.

Supporting information

   Supporting Information
   [256]ADVS-12-2503009-s001.pdf^ (13.7MB, pdf)

Acknowledgements