Abstract

Background

   The mechanisms by which tumor-derived factors remodel the
   microenvironment of target organs to facilitate cancer metastasis,
   especially organ-specific metastasis, remains obscure. Our previous
   studies have demonstrated that SPP1 plays a key role in promoting
   metastasis of hepatocellular carcinoma (HCC). However, the functional
   roles and mechanisms of tumor-derived SPP1 in shaping the
   pre-metastatic niche (PMN) and promoting lung-specific metastasis are
   unclear.

Methods

   Orthotopic metastasis models, experimental metastasis models, CyTOF and
   flow cytometry were conducted to explore the function of SPP1 in
   shaping neutrophil-dominant PMN and promoting HCC lung metastasis. The
   main source of CXCL1 in lung tissues was investigated via fluorescence
   activated cell sorting and immunofluorescence staining. The expression
   of neutrophils and neutrophil extracellular traps (NETs) markers was
   detected in the lung metastatic lesions of HCC patients and mouse lung
   specimens. The therapeutic significance was explored via in vivo DNase
   I and CXCR2 inhibitor assays.

Results

   SPP1 promoted HCC lung colonization and metastasis by modifying
   pulmonary PMN in various murine models, and plasma SPP1 levels were
   closely associated with lung metastasis in HCC patients.
   Mechanistically, SPP1 binded to CD44 on lung alveolar epithelial cells
   to produce CXCL1, thereby attracting and forming neutrophil-abundant
   PMN in the lung. The recruited neutrophils were activated by SPP1 and
   then formed NETs-dominant PMN to trap the disseminated tumor cells and
   promote metastatic colonization. Moreover, early intervention of
   SPP1-orchestrated PMN by co-targeting the CXCL1-CXCR2 axis and NETs
   formation could efficiently inhibit the lung metastasis of HCC.

Conclusions

   Our study illustrates that HCC-lung host cell-neutrophil interactions
   play important roles in PMN formation and SPP1-induced HCC lung
   metastasis. Early intervention in SPP1-orchestrated PMN via CXCR2
   inhibitor and DNase I is a potential therapeutic strategy to combat HCC
   lung metastasis.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s40164-024-00571-x.

   Keywords: Pre-metastatic niche, Hepatocellular carcinoma, C-X-C motif
   chemokine ligand 1, Lung epithelial cells, SPP1

Background

   Hepatocellular carcinoma (HCC) accounts for approximately 75–85% of
   primary liver cancers and is the third leading cause of cancer death
   worldwide and the second leading cause of cancer death in China [[57]1,
   [58]2]. Many patients lose opportunities for curative therapies due to
   extrahepatic metastasis, which makes the prevention of metastasis a
   critical component of systemic HCC treatment and the identification of
   pivotal factors in HCC metastasis is highly important for prolonging
   patient survival [[59]3]. Thus, it is imperative to promptly identify
   patients at high risk of metastasis and intervene before the onset of
   detectable metastasis [[60]4].

   As an early event of metastasis, the pre-metastatic niche (PMN) plays a
   critical role in organic-specific cancer metastasis, and targeting the
   PMN is an effective anti-metastatic strategy for many cancers [[61]5,
   [62]6]. The process of PMN formation involves three components: primary
   tumor-derived factors or extracellular vesicles (EVs), tumor-mobilized
   bone-marrow-derived cells and organ-specific host cells [[63]7].
   Interestingly, the diverse components and characteristics of the PMN in
   different organs may help explain the organ-specific metastasis [[64]6,
   [65]8]. Notably, the interaction between tumor-secreted factors and
   diverse host cells plays a pivotal role in the development of
   metastatic organotropism in numerous cancer types [[66]9]. However, the
   mechanism of PMN formation in the lung metastasis of HCC is less clear.

   SPP1 is a secreted protein that is overexpressed in various solid
   malignancies. In HCC, SPP1 has been identified as a leading
   pro-metastasis gene [[67]10]. The plasma SPP1 level shows a significant
   association with the number of tumors, Edmondson’s grade, serum AFP
   level, and especially the TNM stage. Meanwhile, elevated plasma SPP1
   level was an independent prognostic factor for overall survival (OS)
   and disease-free survival (DFS) [[68]11, [69]12]. SPP1 provides an
   immunosuppressive environment to support HCC growth by influencing
   macrophage infiltration and polarization in the primary lesion
   [[70]13]. Moreover, SPP1 is recognized as a pivotal factor influencing
   the tumor microenvironment landscape [[71]14]. In breast cancer,
   tumor-secreted SPP1 induces fibroblasts to differentiate into
   myofibroblasts, which subsequently facilitate epithelial-mesenchymal
   transition (EMT) and the metastasis of tumor cells through the
   secretion of CXCL12 [[72]15]. Furthermore, in glioblastoma and
   esophageal squamous cell carcinoma, SPP1 promotes the infiltration of
   M2 macrophages and is strongly correlated with poor patient prognosis
   [[73]16, [74]17]. However, cancer metastasis is a multistep cascade,
   and the function of SPP1 in distant organs and whether SPP1 regulates
   the PMN to fuel organ-specific metastasis of HCC is still unclear.
   Moreover, due to the multiple physiological functions of SPP1, directly
   targeting SPP1 to prevent HCC metastasis is still a major challenge.
   Therefore, clarifying the role of SPP1 in the metastatic process,
   especially in the PMN stage, may provide a potential therapeutic
   strategy.

   Neutrophils are the fastest mobilized immune cells and constitute the
   only type of immune cell that independently predicts poor prognosis in
   patients with HCC [[75]18, [76]19]. Neutrophils play dual roles in
   promoting or combating tumor progression through immunosuppression,
   reactive oxygen species, angiogenesis, or nitric oxide release in
   different stages of cancer metastasis [[77]20–[78]25]. As a product of
   the distinctive cell death process of neutrophils, neutrophil
   extracellular traps (NETs) contain granule proteins, DNA, and histones
   and are widely recognized because of their essential role in tumor
   metastasis. We previously reported that under inflammatory conditions,
   NETs could trap HCC cells and trigger their metastatic potential by
   inducing their resistance to cell death and increasing their
   invasiveness [[79]26]. However, the relationship between primary tumors
   and NETs formation in the lung, the most common extrahepatic HCC
   metastasis organ, is poorly understood.

   In this study, we demonstrated the crucial role of the cascade reaction
   among tumor-derived SPP1, lung epithelial cells and neutrophils in the
   formation of the PMN and HCC lung metastasis. Specifically,
   tumor-derived SPP1 acted on lung epithelial cells via the CD44/CXCL1
   axis to increase local neutrophil infiltration. Recruited neutrophils
   formed abundant NETs to trap HCC cells and increase their metastatic
   capacity. Moreover, we found that SPP1 was correlated with MPO-DNA
   levels in HCC patients, which provided theoretical support for the
   prediction of HCC lung metastasis by measuring plasma SPP1 levels and
   expanding the range of plasma SPP1 applications. Moreover, early
   intervention in SPP1-orchestrated PMN by targeting the CXCL1-CXCR2 axis
   and NETs is a promising strategy to prevent HCC lung metastasis.

Methods

Human specimens, animal models, and cell lines

   In this study, we collected four cohorts, and all patients were
   diagnosed as HCC according to histological examination. The tissues and
   plamsa samples of HCC patients were obtained from Hepatobiliary
   Surgery, Huashan Hospital, Fudan University with informed patient
   consent in accordance with a protocol approved by the Ethics Committee
   of Huashan Hospital, Fudan University (Shanghai, China). The first
   cohort included non-lung metastatic HCCs (NLM, n = 18) and lung
   metastatic HCCs (LM, n = 9), and the lung metastasis occurrence was
   diagnosed by a typical imaging appearance in chest CT scans. Another
   independent cohort of plasma samples (n = 21), used for myeloperoxidase
   (MPO) and SPP1 concentration correlation analysis, were obtained from
   HCC patients without radiotherapy, chemotherapy, or curative
   hepatectomy. Another independent cohort of lung metastatic HCC samples
   (n = 12) was obtained from patients who underwent metastasectomy for
   pulmonary metastatic lesions. The tissue microarray and cohort (n = 90)
   used for RNAscope In Situ Hybridization and OS/RFS analysis were
   purchased from Shanghai Outdo Biotech Company (Cat# HLivH180Su14,
   Shanghai, China). For animal experiments, 6–8 weeks old of male
   C57BL/6, BALB/c and BALB/c nude mice were purchased from GemPharmatech
   Co., Ltd (Jiangsu, China) and fed in a pathogen-free vivarium at 25° C
   temperature, 50–60% moisture. All animal experiments were performed
   following the guidelines for the care and were approved by the
   institutional review board of the Department of Laboratory Animal
   Science, Fudan University (Shanghai, China). The mouse alveolar
   epithelial cell line MLE-12 was purchased from Procell Life Science &
   Technology (Wuhan, China); the HEK293T cell line, mouse HCC cell line
   H22, and human non-small cell lung cancer cell line A549 were obtained
   from the Chinese Academy of Science (Shanghai, China). Mouse HCC cell
   line Hepa1-6 and human HCC cell line MHCC97H were obtained from the
   Liver Cancer Institute, Fudan University (Shanghai, China). HEK293T,
   Hepa1-6, MHCC97H and A549 were cultured in DMEM (Gibco, Cat# 11965118)
   supplemented with 10% FBS (Gibco, Cat# 10270106) and 1% penicillin and
   streptomycin (Beyotime, Cat# C0222) at 37 °C in 5% CO[2]. MLE-12 was
   cultured in DMEM/F12 (Gibco, Cat# 11320033) supplemented with 2% FBS
   and 1% penicillin and streptomycin at 37 °C in 5% CO[2].

Mice model: SPP1-modified metastasis and treatment model

   Hepa1-6^vector/SPP1 CM (500 μl/mouse, QD, intraperitoneally (i.p)) or
   SPP1 protein (10 μg/mouse, QD, i.p) was injected for 7 days before
   tumor cell implantation to imitate a cancer-related inflammation
   condition prone to generate NETs. A neutrophil-depleting antibody
   (12.5 μg/mouse, QD, i.p, Biox Cell, Cat# BE0075-1) was given at the
   same time as CM or SPP1 to eliminate neutrophils. Anti-SPP1
   neutralizing antibody (2 μg/mouse, QD, i.p, R&D System, Cat# AF808) was
   given at the same time as CM or SPP1 to neutralize secretory SPP1. For
   the treatment exploration, DNase I (200 U/mouse, QD, i.p, Solarbio,
   Cat# D8071), GSK484 (5 mg/kg, twice-weekly, i.p, MCE, Cat# HY-100514),
   and SB225002 (2 mg/kg, QD, i.p, MCE, Cat# HY-16711) were given to
   degrade NETs and inhibit neutrophil recruitment, separately.

Mice model: establishment of spontaneous metastasis in SPP1-modified models

   In 7 days after establishment of the SPP1-modified model in mice, the
   subcutaneous xenograft tumors were removed and minced into 1–2 mm^3
   sections and washed in precooled 1✕PBS. Then, mice were anesthetized
   with intraperitoneal injection of tribromoethanol (250 mg/Kg, Selleck,
   Cat#S4508) and laparotomy with a 3–4 cm incision along the xiphoid
   process was made to expose the left hepatic lobe. Using a sterile
   cotton swab to gently roll out the left hepatic lobe and keeping it
   moist with normal saline. The minced tumor fragment was directly
   inserted into the subcapsular parenchyma of the left hepatic lobe using
   sterile ophthalmic tweezer, followed by closure of the abdominal
   incision with layered sutures. Mice were sacrificed after 5 weeks and
   the tumors, normal liver and lung were removed. Specimens were fixed in
   4% paraformaldehyde (PFA, Beyotime, Cat# P0099), embedded into paraffin
   blocks, or flash-frozen in liquid nitrogen and stored at − 80 °C to
   obtain tissue protein or RNA. Serial sectioning of the lung was
   performed, and H&E staining was used for the determination of
   metastatic burden.

Mice model: establishment of experimental metastasis in SPP1-modified models

   In 7 days after the establishment of the SPP1-modified model in mice,
   2 × 10^6 Hepa1-6 cells were intravenously (i.v) injected into C57BL/6,
   2 × 10^6 MHCC97H cells were i.v injected into BALB/C nude mice and
   1 × 10^6 H22 cells were i.v injected into BALB/C mice to form
   experimental metastasis model, and the in vivo fluorescence imaging was
   performed using In vivo imaging system (IVIS) Spectrum (PerkinElmer)
   with Living Image software. The mice were sacrificed and lung
   metastasis burden was assessed in day 35, day 40 and day 16,
   separately. The percentage of metastasis events or the number of lung
   metastatic lesions was directly assessed on lung sections using H&E
   staining.

CyTOF

   CyTOF analysis was conducted by PLT-Tech Inc. (Hangzhou, China),
   involving the dissociation of lung tissue into single cells using DNase
   I, collagenase IV, and hyaluronidase. The cells were then stained with
   Pt-194 and incubated with DNA Intercalator-Ir to distinguish between
   singly nucleated cells and doublets. Subsequently, a permeabilization
   buffer was utilized, and the cells were subjected to an intracellular
   antibody panel before being rinsed. Subsequently, the cells underwent
   rinsing, and the signals were analyzed utilizing a CyTOF system
   (Helios™, Fluidigm). Identification of immune cell types was achieved
   through nonlinear dimensionality reduction (t-distributed stochastic
   neighbor embedding), followed by density clustering. The antibodies for
   CyTOF were listed in Table S1.

Neutrophil isolation and NETs formation

   The procedure of neutrophil isolation and NETs formation was described
   in our previous study [[80]26]. Briefly, to isolate mouse neutrophils,
   bone marrow cells were harvested in sterile Hank’s balanced salt
   solution (Beyotime, Cat# C0218) and gently laid on the top of a 2-layer
   Percoll gradient (65% and 55%, Cytiva, Cat# 17089101), followed by
   centrifugation at 1000 g for 30 min at room temperature (RT) without
   brake. Neutrophils were collected from the interface of 65% and 55%
   fractions. To isolate human neutrophils, whole blood samples were
   freshly collected from HCC male volunteers with heparinized tubes and
   laid on top of Polymorphprep (Axis Shield, Cat# 1114683) and
   centrifuged at 500 g for 30 min at RT without brake. For NETs
   formation, freshly isolated neutrophils were seeded on poly-L-lysine
   (Beyotime, Cat# C0313) coated coverslips (NEST, Cat# 801007) in 24-well
   plates and stimulated with recombined human or mouse SPP1 for 4–6 h to
   form NETs. In NETs formation under CM stimulation assays, a 1:1 mixture
   of RPMI 1640 (Gibco, Cat# 61870036) and CM was added. In all assays,
   20 nM PMA (Sigma, Cat# P8139) was used as a positive control while
   1XPBS was a negative control. Next, coverslips were stained with
   SytoxGreen (Invitrogen, Cat# S7020) and Hoechst (Beyotime, Cat# C1029)
   for 20 min. Zeiss inverted fluorescence microscopes (Leica) were used
   to detect and image fluorescence. In some situations, NETs were fixed
   with 4% PFA and stained with DAPI (Beyotime, Cat# C1002) and SytoxGreen
   for further IF detection.

Migration assays

   The procedure of two-chamber migration assays was previously
   described [[81]21]. For neutrophil migration assay, 5 × 10^5 freshly
   isolated neutrophils in serum-free RPMI 1640 were added to the upper
   chamber (3.0 µm, Corning, Cat# 3415) and a 1:1 mixture of RPMI 1640 and
   CM was added to the lower chamber as the chemoattractant and incubated
   at 37 °C in 5% CO[2] for 3 h. As for tumor cell migration assay,
   4 × 10^4 Hepa1-6 cells in serum-free DMEM were added to the upper
   chamber (8.0 µm, Corning, Cat# 3422) and DEME (supplemented with 10%
   FBS) was added to the lower chamber and incubated at 37 °C in 5% CO2
   for 48 h. The migrated cells in the lower chamber were fixed by 4% PFA
   fix solution (Beyotime, Cat# P0099) and then counted by 0.2% crystal
   violet staining (Beyotime, Cat# C0121).

Immunofluorescence staining (IF)

   The day before staining, cells were seeded on coverslips in 24-well
   plates and treated with indicated conditions. Subsequently, the
   coverslip was washed by 1XPBS for 3 times and fixed with 4% PFA for
   20 min at RT. Then coverslips were permeabilized with 0.1% Triton X-100
   (Sigma, Cat# X100) for 15 min at RT, washed with 1✕PBS, and blocked
   with PBS containing 1% BSA (Beyotime, Cat# ST023) for 1 h at RT.
   Staining the coverslips with the primary antibody in blocking buffer at
   4 °C overnight, the antibodies were listed in Table. S4. After washing,
   coverslips were incubated with corresponding fluorescence-conjugated
   secondary antibodies for 1 h protected from light. DAPI was used for
   followed nuclear staining. Zeiss inverted fluorescence microscopes
   (Leica) were used to detect and image fluorescence and images were
   analyzed with ImageJ software.

Quantification and statistical analysis

   Datasets were analyzed with GraphPad Prism software. The student’s
   t-test, Chi-squared test, Fisher exact test, and one- or two-way ANOVA
   test were used to analyze the data as in the figure legends.
   Differences were considered significant at  P< 0.05.

Results

Tumor-secreted SPP1-orchestrated PMN favors HCC lung metastasis

   To assess the potential role of SPP1 in the lung metastasis of HCC, we
   initially validated the upregulation of SPP1 mRNA expression in the
   TCGA-HCC cohort (Figure S1A-B). A significant increase in SPP1
   expression in liver tumor tissues compared with that in corresponding
   noncancer tissues was confirmed by quantitative real-time polymerase
   chain reaction (qRT-PCR) (Figure S1C). Considering that SPP1 can be
   secreted by various matrix cells in the tumor microenvironment, we
   detected SPP1 expression via RNAscope In Situ Hybridization and
   demonstrated that the tumorous secretion of SPP1 was dominant (Figure
   S1D-F). Moreover, SPP1 expression was positively associated with
   reduced OS and RFS in HCC patients according to the results of RNAscope
   In Situ Hybridization (Figure S1G). To further elucidate the
   relationship between SPP1 and HCC metastasis, we analyzed a Huashan
   cohort of HCC patients with lung metastasis (LM) or non-lung metastasis
   (NLM) (Table S2). Interestingly, SPP1 expression in HCC tissues was
   elevated in the LM group compared with the NLM group, and the LM group
   presented higher SPP1 concentrations in the circulation (Fig. [82]1A,
   [83]B). Meanwhile, higher SPP1 expression levels in HCC tissues were
   positively correlated with plasma SPP1 concentration (Figure S1H). In
   the multivariate analysis, SPP1 was identified as an independent risk
   factor for lung metastasis (Table S3). Together, these results indicate
   that SPP1 upregulation is closely associated with lung metastasis in
   HCC.

Fig. 1.

   [84]Fig. 1
   [85]Open in a new tab

   SPP1 is closely associated with the lung metastasis of HCC. A ELISA
   detection of plasma SPP1 concentration in samples of NLM (n = 18) and
   LM (n = 9). B qRT‒PCR analysis of SPP1 mRNA levels in tissues of NLM
   (n = 9) and LM (n = 6). The P values were calculated by Student’s
   t-test. C, D SPP1 protein levels in plasma (C) and pro-metastatic gene
   expression in the lungs (D) of mice at 7 days after SPP1 treatment
   (n = 6 per group). The P values were calculated by Student’s t-test. E
   Tissue-clearing technique showed early colonization of Hepa1-6^GFP in
   the lungs at 1 day after i.v. injected in PBS or SPP1 pre-treated mice
   (n = 3 per group). The P values were calculated by Student’s t-test. F
   The effect of SPP1 on HCC metastasis was evaluated in the SPP1-PMN
   model of Hepa1-6 (n = 7 per group) and H22 cell lines (n = 6 per
   group). The P values were calculated by two-tailed Fisher exact test
   (for Hepa1-6) and Student’s t-test (for H22), respectively. G–J
   Experimental schematics of modified orthotopic spontaneous metastasis
   model (G). Quantification analysis of SPP1 protein levels in plasma (H)
   and pro-metastatic gene expression in the lungs (I) of mice at 14 days
   after orthotopic liver transplantation (n = 5 per group). Shown were
   representative images of lung metastatic loci of Hepa1-6^vector/SPP1 in
   the SPP1-orchestrated spontaneous metastasis model (J, n = 5 per
   group). The P values were calculated by a student’s t-test (for H-I)
   and two-tailed Fisher exact test (for J), respectively. K I.v.
   injection of Hepa1-6^vector or Hepa1-6^SPP1 cells with matched
   CM-treatment for lung metastasis analysis. Representative images and
   the ratio of lung metastasis were shown (n = 7 per group). The P values
   were calculated by the two-tailed Fisher exact test. L The effect of
   anti-SPP1 neutralizing antibody on HCC metastasis was evaluated in the
   experimental metastasis model (n = 7 per group). The P values were
   calculated by the two-tailed Fisher exact test. For all tests,
   significance is determined with a 95% confidence interval (ns,
   P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001)

   Given the important role of the PMN in the metastatic colonization of
   tumor cells [[86]8], we further investigated the potential effects of
   tumor-derived SPP1 on PMN formation and HCC lung metastasis. We
   initially procured recombinant mouse SPP1 (Figure S1I) and administered
   it to immunocompetent mice. Circulating levels of SPP1 in the murine
   models reached a steady state 7 days after the intraperitoneal
   injection (i.p.) of SPP1, similar to what has been observed in clinical
   pathological conditions (Fig. [87]1C) [[88]12]. The RNA sequencing
   (RNA-seq) and qRT-PCR data revealed increased expression of various
   PMN-related genes and inflammation-associated factors in pulmonary
   tissue 7 days after i.p. injection of SPP1 (Fig. [89]1D, Figure S1J,
   Table S4). Notably, these alterations were not detected in other organs
   (Figure S1K-M), indicating a profound pro-metastatic remodeling effect
   of SPP1 on the lung (referred to as SPP1-PMN). PMN fosters early tumor
   cell colonization, a process that is instrumental in facilitating
   metastasis. Similarly, SPP1-PMN was observed to enhance early
   colonization of murine Hepa1-6^GFP cells by intravenous injection (i.v)
   (Fig. [90]1E) [[91]27], which was concomitant with an increase in
   experimental metastasis in both i.v. injected Hepa1-6 and H22 cells, a
   suspended murine HCC cell line from malignant ascites (Fig. [92]1F). To
   better mimic the continuous and dynamic nature of PMN remodeling by
   tumorous SPP1 during HCC progression and the subsequent metastasis
   cascade, we overexpressed SPP1 in Hepa1-6 cells (Figure S1N-O) and
   established a modified Hepa1-6^SPP1 or Hepa1-6^vector orthotopic
   spontaneous metastasis model with matched conditioned medium (CM) to
   accelerate lung remodeling [[93]28]. We harvested lung tissues from
   this model 2 weeks after Hepa1-6 implantation (Fig. [94]1G), a
   timepoint generally recognized as the PMN stage (referred to as
   Hepa1-6^SPP1- or Hepa1-6^vector-PMN) [[95]29, [96]30]. PMN was readily
   remodeled by SPP1 overexpression, as evidenced by the higher
   circulating SPP1 level and PMN-related genes in Hepa1-6^SPP1-PMN than
   in the control group (Fig. [97]1H, [98]I), mirroring the scenario
   observed with SPP1 administration. Consistent with the remodeling of
   the PMN, SPP1 overexpression was also accompanied by a greater
   frequency of spontaneous pulmonary metastasis (Fig. [99]1J), as opposed
   to metastasis in other organs. Moreover, overexpression of SPP1 in
   Hepa1-6 had no effect on the subcutaneous and orthotopic tumor growth
   (Figure S1P). Notably, similar to SPP1-PMN, Hepa1-6^SPP1 CM alone also
   efficiently led to an increased experimental metastasis through PMN
   remodeling (Fig. [100]1K). To further prove that the pro-metastatic
   effects of SPP1 were mediated through PMN remodeling rather than
   augmenting the invasiveness of the primary tumor, we generated an
   N-terminal signal peptide deleted SPP1 mutant in Hepa1-6
   (Hepa1-6^SPP1△1–48, also known as SPP1 isoform-2), which preserves the
   intracellular expression and bioactivity of full-length SPP1 while
   simultaneously inhibiting its extracellular secretion (Figure S2A-B)
   [[101]31]. In vitro, both the overexpression of SPP1 and SPP1△1–48 had
   a negligible effect on the proliferation and migration of Hepa1-6 cells
   (Figure S2C-E), which minimized the possible autocrine or paracrine
   effects of SPP1 on HCC tumor cells directly. Nonetheless, the
   attenuation of SPP1 secretion with SPP1△1–48 significantly reduced the
   lung metastasis as compared with the full-length SPP1 isoform but had
   minimal influence on the growth of subcutaneous and orthotopic tumors
   (Figure S2F-G). Moreover, systemically neutralizing SPP1 effectively
   mitigated the remodeling of PMN remodeled by Hepa1-6^SPP1 CM and
   attenuated the augmented experimental metastasis of i.v Hepa1-6^SPP1
   cells (Fig. [102]1L). This effect was also validated in an
   immunodeficient mouse model, where SPP1 neutralization reduced
   experimental lung metastasis induced by MHCC97H, a highly metastatic
   human HCC cell line with robust high SPP1 expression (Figure S2H).
   Taken together, these results indicate that secretory SPP1 plays a
   dominant role in the lung metastasis of HCC accompanied by PMN
   remodeling.

Neutrophils are essential in SPP1-orchestrated lung PMN

   To further elucidate the mechanism by which SPP1 orchestrated lung PMN,
   we initially conducted mass cytometry (CyTOF) with 42-metal
   isotope-tagged monoclonal antibody to evaluate the alterations in all
   major immune cells and certain stromal cell types in the lung after
   SPP1 administration for 7 days (SPP1-PMN, related to Fig. [103]1D). We
   identified 23 distinct cellular clusters and the following principal
   cell subsets based on the expression levels of lineage markers: CD8^+ T
   cells, CD4^+ T cells, B cells, macrophages, gamma-delta T cells,
   natural killer cells, neutrophils, dendritic cells, fibroblasts, and
   epithelial cells (Fig. [104]2A, Figure S3A-B). Compared with the
   control group, the SPP1-PMN group presented a significant increase in
   the neutrophil population, whereas no discernible changes were observed
   in other immune cell clusters (Fig. [105]2B). In accordance with the
   CyTOF results, flow cytometry (FC) revealed that neutrophils were the
   dominant cell type whose abundance was altered in SPP1-PMN (Figure
   S3C-D). Moreover, the ImmuCC algorithm applied to RNA-seq data of
   SPP1-PMN (related to Figure S1J) also predicted a pronounced trend
   toward increased neutrophils, but not other cells (Figure S4A)
   [[106]32, [107]33]. Immunohistochemistry (IHC) analysis of pulmonary
   neutrophils (Fig. [108]2C, [109]D), FC analysis of bronchoalveolar
   lavage fluid (BALF) (Fig. [110]2E), and myeloperoxidase (MPO) activity
   analysis (Fig. [111]2F) further confirmed that the number of
   neutrophils was strongly increased in SPP1-PMN.

Fig. 2.

   [112]Fig. 2
   [113]Open in a new tab

   SPP1 promotes HCC lung metastasis in a neutrophil-dependent manner. A
   TSNE plot projection of single cells from PBS-PMN and SPP1-PMN, colored
   by 23 clusters. The lungs were harvested after i.p. injecting SPP1 for
   7 days (related to Fig. [114]1D) (n = 4 per group). B The
   quantification of the distinct immune cell subtypes in PBS- and
   SPP1-PMN group. C, D IHC staining of neutrophils (Ly6G^+) in the lung
   tissues (C). To verify, the number of neutrophils is counted in five
   random fields and took up the average per sample (D, n = 5 per group).
   The P values were calculated by a student’s t-test. Scale bar, 100 μm.
   E, F FC analysis of neutrophils (CD45^+CD11b^+Ly6G^+) in the BALF and
   MPO activity in the lungs from SPP1-PMN model (n = 5 per group). The P
   values were calculated by a student’s t-test. G, H FC analysis of
   immune cells in lung tissues from the orthotopic model of
   Hepa1-6^vector/SPP1 mice (n = 4 per group). Lungs were harvested at
   14 days after orthotopic inoculation. Percentages of neutrophils
   (CD45^+CD11b^+Ly6G^+), macrophages (CD45^+CD11b^+F4/80^+), NK cells
   (CD45^+ NK1.1^+), CD4 + T cells (CD45^+CD3^+CD4^+CD8^−), CD8 + T cells
   (CD45^+CD3^+CD4^−CD8^+), B cells (CD45^+B220^+) were shown. The P
   values were calculated by a student’s t-test. I, J Representative IHC
   images and quantification of Ly6G staining and in the lungs of Hepa1-6
   orthotopic implantation model and anti-SPP1 treatment model (n = 5 per
   group). Scale bar, 100 μm. The P values were calculated by a student’s
   t-test. K, L Representative H&E images and quantification of lung
   metastasis loci of Hepa1-6^SPP1 (K) or H22 (L) in SPP1-associated
   experimental metastasis model after depleting neutrophils by anti-Ly6G
   antibody. For all tests, significance is determined with a 95%
   confidence interval (ns, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001;
   ****P < 0.0001)

   In addition to the setting of SPP1-PMN, FC analysis of Hepa1-6^SPP1-PMN
   confirmed a considerable increase in the neutrophil population,
   alongside a varied reduction in other immune cells including
   macrophages, natural killer cells, T cells, and B cells, compared with
   that of Hepa1-6^vector-PMN (Fig. [115]2G, [116]H), and IHC confirmed
   the increase in neutrophils in the Hepa1-6^SPP1 group (Fig. [117]2I).
   Moreover, a marked increase in the number of pulmonary neutrophils was
   observed in Hepa1-6^SPP1 CM-treated lungs, and this increase was
   dampened by the SPP1 neutralizing antibody (Fig. [118]2J), which was
   consistent with the change in metastasis (Fig. [119]1L). To further
   determine the necessity of neutrophils in SPP1-orchestrated PMN, we
   used an anti-Ly6G antibody to deplete neutrophils within lung PMN
   induced by Hepa1-6^SPP1 CM or SPP1 and observed a significantly
   mitigated experimental lung metastasis of Hepa1-6 and H22 cells,
   respectively (Fig. [120]2K, [121]L). Similarly, the experimental
   metastasis of human MHCC97H cells was also abrogated by neutrophil
   depletion (Figure S4B). Taken together, these findings strongly
   indicate that neutrophils are essential in SPP1-orchestrated lung PMN.

Neutrophil recruitment to SPP1-orchestrated PMN depends on lung epithelial
CXCL1 secretion

   Orthotopic implantation of Hepa1-6^SPP1 did not significantly affect
   circulating neutrophil counts compared to control group(Figure S5A),
   whereas in vitro assays revealed that CM from SPP1-stimulated lung
   tissues enhanced neutrophil chemotaxis, a response not induced in the
   setting of CM from untreated lung tissues or SPP1 alone (Fig. [122]3A),
   which implied that SPP1-orchestrated PMN may indirectly facilitate
   pulmonary neutrophil recruitment, which is dependent on local lung
   microenvironment.

Fig. 3.

   [123]Fig. 3
   [124]Open in a new tab

   SPP1 promotes neutrophil recruitment by stimulating lung epithelial
   cells to increase CXCL1 secretion. A CM from different groups was
   placed in the lower chambers. Primary neutrophils were isolated and
   then placed in the upper chambers and their migration into the lower
   chambers was counted. The P values were calculated by one-way ANOVA. B
   qRT-PCR showed the expression of neutrophil-related chemokines and
   cytokines in lung tissue in SPP1-PMN (upper panel, related to
   Fig. [125]1C) and in Hepa1-6^vector/SPP1-PMN (lower panel). The P
   values were calculated by Student’s t-test. C qRT‒PCR analysis of Cxcl1
   expression in epithelial cells, neutrophils, macrophages, fibroblasts,
   T cells, and B cells purified from the lung tissues in the
   Hepa1-6^vector/SPP1-PMN model. D Representative images (right panel)
   and quantitative analysis of SPP1-his distribution in different tissues
   of mice with i.p. administration of SPP1-his. Scale bar, 50 μm. E, F IF
   staining of SPP1-his and lung epithelial cell (SFTPC^+) colocalization
   (E) and SPP1-positive rates (F) in different cells of mice with i.p.
   administration of SPP1-his (related to Figure S5C, Supporting
   Information). Scale bar, 20 μm. G Representative IF image of SPP1-his
   internalization by MLE-12 cells. Scale bar, 10 μm. H qRT‒PCR showed the
   neutrophil-related chemokines and cytokines in MLE-12 cells treated
   with various concentrations of SPP1 (upper panel) or cocultured with
   Hepa1-6^vector/SPP1 (lower panel). The P values were calculated by
   Student’s t-test. I Migrated cell numbers of murine neutrophils
   recruited by CM of MLE-12 cells, SPP1-pretreated MLE-12 cells or MLE-12
   cells treated with 1 µg/ml SPP1 plus CXCR2 inhibitor (SB225002, 40 nM)
   or by MLE-12 CM plus 1 µg/ml SPP1. The P values were calculated by
   one-way ANOVA. J IHC images and quantification of neutrophils in the
   Hepa1-6-PMN models following CXCR2 inhibitor treatment (n = 6 per
   group). The P values were calculated by Student’s t-test. Scale bar,
   100 μm. For all tests, significance was determined with a 95%
   confidence interval (ns, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001;
   ****P < 0.0001)

   To elucidate this phenomenon, we examined the expression of several
   cytokines and chemokines in the Hepa1-6^SPP1- or Hepa1-6^vector-PMN in
   a modified orthotopic model, as well as the SPP1-PMN. Several
   neutrophil-recruiting chemokines, including C-X-C motif chemokine
   ligand 1 (Cxcl1), were significantly upregulated, whereas the
   expression of macrophage-related chemokines, including C–C motif
   chemokine ligand 1, remained unchanged (Fig. [126]3B, Figure S5B). We
   further sorted various types of resident cells in Hepa1-6^SPP1-PMN and
   found that epithelial and neutrophils were the main sources of
   increased Cxcl1 compared with those in the control group
   (Fig. [127]3C). To further elucidate how SPP1 recruited neutrophils
   through lung epithelial cells, SPP1 labeled with his-tag (SPP1-his) was
   purified and then i.p injected into the mice. We found that his-tagged
   SPP1 was predominantly distributed throughout the lung but not other
   organs, supporting the specific role of SPP1 in reshaping the lung PMN
   (Fig. [128]3D). Furthermore, IF staining revealed that exogenous SPP1
   targeted mainly type II alveolar epithelial cells (AECII, Surfactant
   Protein C + cells (SFTPC^+)), with fewer associations with neutrophils
   (MPO^+ cells) and no association with other cells (Fig. [129]3E,
   [130]F, Figure S5C), which was congruent with the results of qRT-PCR
   (Fig. [131]3C).

   Upon stimulation with SPP1-his, the murine lung epithelial cell line
   MLE-12 was directly combined with the exogenous SPP1 (Fig. [132]3G).
   Moreover, qRT-PCR revealed that Cxcl1 expression was significantly
   upregulated in MLE-12 cells after SPP1 stimulation or co-culture with
   Hepa1-6^SPP1 cells (Fig. [133]3H). Consistently, the upregulation of
   Cxcl1 was also induced by SPP1 or recombinant human SPP1 (rhSPP1) in
   freshly purified AEC II from mouse lungs and human A549 cells,
   respectively (Figure S5D-E). In accordance with elevated Cxcl1
   expression, CM from MLE-12 cells pre-treated with SPP1 exhibited a
   greater increase in neutrophil chemotaxis than MLE-12 CM or a
   combination of SPP1 and MLE-12 CM. This effect was abrogated by
   blockage of C-X-C motif chemokine receptor 2 (CXCR2, a well-known CXCL1
   receptor) via SB225002 (Fig. [134]3I). In the orthotopic model, the
   increased neutrophil infiltration in Hepa1-6^SPP1-PMN was also
   attenuated by CXCR2 blockade (Fig. [135]3J). Collectively, these data
   suggest that HCC-derived SPP1 facilitates neutrophil recruitment
   through stimulating CXCL1 expression by lung epithelial cells, thereby
   forming a neutrophil-enriched PMN that favors the lung-specific
   metastasis of HCC.

SPP1 facilitates neutrophil recruitment by activating the lung epithelial
CD44/STAT3/CXCL1 signaling axis in PMN

   SPP1 is known to interact primarily with two types of receptors, CD44
   and various integrins, including αv (β1, β3, β5), α8β1, and α5β1
   [[136]34]. Our RNA-seq analysis (Figure S1J, Table S4) revealed
   elevated CD44 and certain integrins in SPP1-PMN, suggesting a potential
   role for SPP1 and its receptors (Fig. [137]4A) and qRT-PCR confirmed
   that Cd44, Itgav, Itga5, and Itgb1 were the predominant SPP1 receptors
   in both mouse lung tissues and cultured MLE-12 cells (Figure S6A-B).
   Among these receptors, depletion of CD44 via lentiviral shRNA negated
   the SPP1-induced upregulation of CXCL1 in MLE-12 cells at both the mRNA
   and protein levels. In contrast, interfering with ITGB1/ITGAV/ITGA5
   with neither shRNA nor the integrin inhibitor MK-0429 failed to
   replicate these findings (Fig. [138]4B–H, Figure S6C-F). IF staining
   confirmed the colocalization of CD44 and SPP1-his in MLE-12 cells, and
   the knockdown of CD44 markedly decreased SPP1 binding in vitro
   (Fig. [139]4I). Additionally, SPP1 stimulation was found to upregulate
   CD44 expression in MLE-12 cells (Figure S6G-H). In accordance with
   CXCL1 expression, the knockdown of CD44 effectively abolished the
   increase in neutrophil chemotaxis toward SPP1-treated MLE-12 cells,
   whereas the knockdown of ITGB1, ITGAV, or ITGA5 did not (Fig. [140]4J),
   confirming that SPP1 regulated CXCL1 expression via CD44, not
   integrins. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway
   analysis of RNA-seq data from SPP1-treated MLE-12 cells highlighted the
   significant enrichment of the JAK/STAT signaling pathway (Figure S6I),
   which has been reported as a downstream effector of the CD44 receptor
   and induce CXCL1 expression [[141]35, [142]36]. Consistent with these
   findings, SPP1 stimulation increased the phosphorylation of STAT3 in
   MLE-12 cells, and the STAT3 inhibitor Stattic reduced SPP1-induced
   CXCL1 expression (Fig. [143]4K, [144]L). Moreover, in vivo experiments
   also showed that SPP1 administration significantly upregulated STAT3
   phosphorylation and CXCL1 expression (Fig. [145]4M). Similar outcomes
   were observed in the A549 cell line (Figure S6J-L).

Fig. 4.

   [146]Fig. 4
   [147]Open in a new tab

   SPP1 increases CXCL1 expression in MLE-12 cells through the CD44/STAT3
   axis. A Heatmap of the SPP1 receptor genes detected by RNA-seq in the
   lung tissue after i.p. administration of SPP1 for 7 days. B–E
   Immunoblotting of CXCL1 expression in MLE-12 shCD44 (B) cells, MLE-12
   shITGAV (C) cells, MLE-12 shITGB1 (D) cells, MLE-12 shITGA5 (E) cells
   treated with SPP1 for 48 h. F–H Immunoblot (F), ELISA (G) and qRT-PCR
   (H) showed CXCL1 expression in MLE-12 shCD44 cells treated with SPP1 or
   MLE-12 cells treated with SPP1 and/or integrin inhibitor (MK-0429,
   20 nM). The P values were calculated by a student’s t-test. I
   Representative IF images of SPP1-his and CD44 colocalization in MLE-12
   shNC and MLE-12 shCD44 cells. Scale bar, 10 μm. J Relative migrated
   cell numbers of murine neutrophils recruited by CM of CD44-, ITGB1- or
   ITGAV-knockdown MLE-12 cells treated with SPP1. The P values were
   calculated by a student’s t-test. K Changes in p-STAT3, STAT3, and
   CXCL1 in MLE-12 cells under SPP1 stimulation. L qRT‒PCR (upper panel)
   or immune blot (lower panel) analysis of CXCL1 expression in MLE-12
   cells treated with SPP1, Stattic (20 μM), and SPP1 plus Stattic. The P
   values were calculated by one-way ANOVA. M Immunoblotting of CXCL1,
   STAT3, p-STAT3 in lung tissues of mice at 7 days after PBS or SPP1
   treatment. N–Q Overlapping of genes resulting from the comparison of
   upregulated genes in mouse lung tissue in SPP1-PMN mice model and in
   SPP1-pretreated MLE-12 cells (N). qRT-PCR (O) and ELISA (P) showed the
   SPP1 expression in SPP1-pretreated MLE-12 cells. The pattern
   diagram (Q) of MLE-12 self-amplified SPP1 expression under exogenous
   SPP1 stimulation. The P values were calculated by a student’s t-test.
   For all tests, significance was determined with a 95% confidence
   interval (ns, P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001;
   ****, P < 0.0001)

   Intriguingly, we noted that SPP1 (encoding SPP1) was among the
   upregulated genes in the RNA-seq data of both SPP1-stimulated MLE-12
   cells and SPP1-PMN (Fig. [148]4N). The triggered increase in SPP1 level
   in SPP1-treated MLE-12 cells was verified by qRT-PCR and ELISA
   (Fig. [149]4O–P). These findings suggest that tumor-derived SPP1 exerts
   a local self-amplification through lung epithelial cells, which further
   increases CXCL1 secretion and neutrophil infiltration in the
   lung-specific PMN (Fig. [150]4Q).

NETs facilitate HCC lung metastasis in SPP1-initiated PMN

   To further elucidate the mechanism by which neutrophil-dominant PMN
   drives HCC lung metastasis, we studied the phenotypic changes of
   neutrophils in SPP1-orchestrated lungs. RNA-seq analysis of
   SPP1-stimulated murine neutrophils revealed that these cells could not
   be classified into classic anti-tumor N1-like or pro-tumor N2-like
   neutrophils (Figure S7A, Table S5). Instead, significantly upregulated
   genes enriched in NETs formation were revealed by KEGG pathway
   enrichment analysis (Fig. [151]5A). We have previously shown that NETs
   promote HCC metastasis by enhancing early seeding and promoting a
   tumorous inflammatory response [[152]26, [153]37]. We thus speculated
   that NETs may participate in the SPP1-orchestrated PMN in HCC. In
   vitro, SPP1 and CM from Hepa1-6^SPP1 cells led to significant NETs
   formation in murine neutrophils (Fig. [154]5B, [155]C), and this effect
   could be abrogated by the administration of GSK484 (which inhibits NETs
   formation) or DNase I (which destroys formed NETs) (Fig. [156]5C)
   [[157]38, [158]39]. Consistently, NETs formation in human peripheral
   blood neutrophils was also increased by rhSPP1 in vitro (Fig. [159]5D).
   The ERK and AKT pathways have been reported to be involved in NETs
   formation, and KEGG analysis of SPP1-stimulated neutrophils also
   revealed the top enriched MAPK (ERK-involved) signaling pathway
   (Fig. [160]5A). Moreover, in vitro inhibition of the ERK pathway,
   rather than the AKT pathway, significantly reduced NETs formation
   (Figure S7B-C), indicating that SPP1 facilitated mouse NETs formation
   by activating the ERK pathway in neutrophils.

Fig. 5.

   [161]Fig. 5
   [162]Open in a new tab

   SPP1 induces NETs formation in PMN. A KEGG enrichment bubble diagrams
   of upregulated genes in SPP1-stimulated neutrophils. B, C
   Representative IF images of NETs formation (B) from mice neutrophils
   with different administrations. NETs are stained with Sytoxgreen
   (green), and DNA is stained with Hoechst (blue). To quantify, NETs
   extension from five random views is shown (C). The P values were
   calculated by one-way ANOVA. Scale bar, 100 μm. D Representative IF
   images and quantified NETs percentage from human neutrophils treated
   with PBS or rhSPP1. The P values were calculated by Student’s t-test.
   Scale bar, 100 μm. E IF images of DIL-labeled Hepa1-6 cells trapped
   within NETs. F, G Quantified NETs extension in lungs of SPP1-PMN at day
   7 (F, related to Fig. [163]1C) and in lungs of Hepa1-6 orthotopic model
   at day 14 (G, related to Fig. [164]1G). The P values were calculated by
   Student’s t-test. H Representative images of adhered DIL-labeled
   Hepa1-6 cells in SPP1-PMN with the treatment of GSK484 or DNase I, the
   lungs were harvested 1 day after i.v. injected Hepa1-6 cells. To
   quantify, tumor cells were counted in five random views for each mouse
   and took an average count (n = 8 per group). The P values were
   calculated by one-way ANOVA. Scare bar, 100 μm. I, J The effect of
   DNase I or GSK484 on experimental metastasis model of Hepa1-6^SPP1.
   Representative images of H3cit + and MPO + signals in lung tissues (I)
   and lung metastasis lesions (J). The P values were calculated by
   one-way ANOVA. Scale bar, 100 μm. K The effect of DNase I or GSK484 on
   SPP1-PMN model of H22. The lungs were harvested at 14 days after i.v.
   injected (n = 5 per group). The P values were calculated by one-way
   ANOVA. For all tests, significance was determined with a 95% confidence
   interval (ns, P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001;
   ****P < 0.0001)

   For additional in vivo evidence of NETs in SPP1-orchestrated PMN, IF
   was performed on SPP1-PMN following i.v. injection of DIL-stained
   Hepa1-6 cells (related to Fig. [165]1D), which visualized the increased
   well-like structural NETs trapping Hepa1-6 cells (Fig. [166]5E,
   [167]F). In line, this finding was corroborated by increased
   citrullinated histone H3 (H3cit, a NETs marker) in SPP1-PMN (Figure
   S7D). In the modified HCC orthotopic model, Hepa1-6^SPP1 PMN also
   presented increased NETs formation, as revealed by two-photon imaging
   and immunoblotting (Fig. [168]5G, Figure S7E). However, neutralizing
   SPP1 effectively prevented the increased pulmonary NETs formation in
   the Hepa1-6^SPP1-CM treated lungs (Figure S7F). We further asked
   whether targeting NETs impaired SPP1-induced HCC PMN development and
   metastasis. As expected, targeting NETs with GSK484 or DNase I
   efficiently decreased the early seeding of Hepa1-6 cells in SPP1-PMN
   (Fig. [169]5H). GSK484 or DNase I efficiently degraded NETs in PMN
   induced by Hepa1-6^SPP1 CM and subsequently diminished the experimental
   lung metastasis of Hepa1-6^SPP1 cells (Fig. [170]5I, [171]J).
   Consistently, H22 metastasis favored by SPP1-PMN was also abrogated by
   targeting NETs (Fig. [172]5K). These results indicate the key role of
   NETs formation in the SPP1-orchestrated HCC PMN.

Early intervention in SPP1-orchestrated PMN by targeting CXCR2 and NETs
prevents HCC lung metastasis

   To further validate the roles of the SPP1/CXCL1/NETs axis in HCC lung
   metastasis, we assessed the clinical relevance of SPP1 and CXCL1
   expression in the lung metastasis loci of human HCCs. We found that the
   metastatic lesions with high SPP1 expression presented elevated CXCL1
   levels, and most of them were co-localized with AECIIs (SFTPC^+)
   (Fig. [173]6A–C). Furthermore, we observed increased neutrophil
   infiltration and NETs formation in the high SPP1 expression group
   (Fig. [174]6A, D, E). Meanwhile, the clinical cohort revealed a
   positive correlation between plasma SPP1 concentration and the
   circulating NETs level, which was marked by MPO-DNA (Fig. [175]6F).

Fig. 6.

   [176]Fig. 6
   [177]Open in a new tab

   CXCR2 inhibitor plus DNase I treatment inhibits HCC lung metastasis. A
   IHC staining of SPP1 in HCC lung metastasis lesions from Huashan cohort
   (left). Matched IF staining of CXCL1 + and SFTPC + cells in lung
   tissues (middle), and IF staining of MPO + and H3cit + cells in
   metastatic tumor lesions (right). Scare bar, 25 μm. B–E Quantification
   analysis of CXCL1 (B), percent of CXCL1 + SFTPC + cells (C), NETs
   formation (H3cit^+, D), and Neutrophils (MPO^+ E) in SPP1-high/-low
   groups (n = 6 per group). The P values were calculated by Student’s
   t-test. Scare bar, 25 μm. F Correlation analysis between serum MPO-DNA
   and SPP1 concentration in HCC samples (n = 21, Pearson correlation). G
   Experimental setup. Briefly, GFP-labeled Hepa1-6 (Hepa1-6^GFP) cells
   were i.v. injected into C57BL/6 mice bearing orthotopic Hepa1-6
   (unlabeled) at their PMN stage. For the early and late treatment
   groups, DNase I and CXCR2 inhibitor-SB225002 were given at day 0 and
   day 14, respectively. At the endpoint, the lung metastatic burden of
   Hepa1-6^SPP1−GFP was detected by BLI (n = 6 per group). H Tumor
   metastasis burden was observed and quantified by fluorescence imaging.
   Representative lung images (left) and quantified results (right) were
   shown, respectively. Data were means ± SD. The P values were calculated
   by one-way ANOVA. I, J Quantification analysis (I) and HE images (J) of
   lung metastasis lesions. The P values were calculated by Fisher’s exact
   test, one-tailed. K The immunostaining of MPO + and H3cit + and the
   signal intensity were quantified and analyzed, and representative
   images of indicated staining of lungs were shown. The P values were
   calculated by one-way ANOVA. Scale bar, 50 μm. For all tests,
   significance was determined with a 95% confidence interval (ns,
   P > 0.05; **P < 0.01; ****P < 0.0001)

   Considering the important effects of neutrophils and NETs on HCC lung
   metastasis, we assessed the therapeutic value and intervention timing
   of targeting neutrophil recruitment and NETs formation to block
   SPP1-initiated PMN formation. We established an orthotopic implantation
   model with Hepa1-6^SPP1 to generate an SPP1-initiated PMN, followed by
   i.v loading of small numbers of GFP-labeled Hepa1-6^SPP1
   (Hepa1-6^SPP1−GFP) cells to mimic the dissemination and metastatic
   process (Fig. [178]6G) [[179]40]. Early intervention prior to the
   establishment of the PMN by combining the inhibition of CXCL1-CXCR2
   with SB225002 and destruction of NETs with DNase I effectively
   diminished pulmonary neutrophil infiltration, NETs formation, and
   pre-established lung metastasis. In contrast, late intervention after
   PMN establishment failed to reduce lung metastasis (Fig. [180]6H–K).
   Both early and late interventions had no direct effect on orthotopic
   tumor growth (Figure S8A-B). To further verify the above findings, we
   repeated the above experiments in human MHCC-97H cells. Orthotopic
   implantation and i.v injection models were established in nude mice via
   MHCC-97H cells and luciferase-labeled MHCC-97H cells (MHCC97H^luci),
   respectively (Figure S8C). Administration of the CXCR2 inhibitor and
   DNase I had no effect on primary tumor growth (Figure S8D-E), but only
   early treatment attenuated the lung metastasis burden (Figure S8F-G).

   In conclusion, these data suggest that targeting CXCL1-CXCR2 and NETs
   formation could effectively prevent lung metastasis in SPP1-positive
   HCC patients, but early intervention is crucial for successful
   treatment (Fig. [181]7).

Fig. 7.

   [182]Fig. 7
   [183]Open in a new tab

   Schematic diagram. In HCC, primary tumor-derived SPP1 stimulated lung
   epithelial cells to secrete more CXCL1 through SPP1/CD44/STAT3 axis.
   Elevated CXCL1 levels in the lung microenvironment promoted neutrophil
   infiltration and NETs formation to facilitate HCC lung metastasis

Discussion

   PMN formation is generally acknowledged as a critical step in cancer
   metastasis, but how HCC modifies the PMN and facilitates subsequent
   metastasis remains obscure. Our previous work revealed that SPP1 was a
   leading pro-metastasis gene in HCC [[184]10], but how HCC-derived SPP1
   modified the PMN in target organs, especially in lungs, to fuel the
   metastasis of liver cancer remains to be further investigated. In this
   study, we demonstrated that tumor-derived SPP1 could increase lung
   neutrophil infiltration to form the PMN and subsequently stimulate the
   formation of NETs to promote HCC lung metastasis. Mechanistically, SPP1
   can specifically target lung host cells, known as lung epithelial
   cells, to secrete CXCL1 through the SPP1/CD44/STAT3 axis and recruit
   neutrophils to form NETs.

   SPP1, a pro-metastatic protein, has been widely studied and is closely
   related to immune escape, EMT, and other functions required for tumor
   metastasis [[185]13, [186]41]. In accordance with these findings, we
   found that SPP1 was upregulated in tumor tissue compared with matched
   para-cancerous tissues and was more highly expressed in the tissues and
   serum of patients with lung metastasis. However, we noticed the plasma
   SPP1 levels in some patients in the NLM group were higher than the
   average level of LM group. Previous studies indicated that the SPP1
   plasma levels were influenced by Edmondson grading, TNM staging and the
   number of tumor nodules of HCC patients [[187]11], which made plasma
   SPP1 levels vary from patient to patient in the same group. Meanwhile,
   cancer metastasis is a muti-steps and complex process, SPP1 could
   increase the opportunity of metastasis rather than necessarily lead to
   HCC lung metastasis. Moreover, tumor-derived SPP1 promoted
   neutrophil-dominant PMN formation by specifically targeting lung
   epithelial cells to secrete CXCL1 and subsequently boosted HCC lung
   metastasis through NETs formation. However, some studies have shown
   that SPP1 promotes HCC cell proliferation and migration [[188]42],
   while we found SPP1 overexpression had little effect on Hepa1-6
   proliferation and migration (Figure S2A-C). The discrepancy between our
   study and that of Wu et al. may be due to differences in cell lines,
   and the addition of SPP1 may not be essential for malignant phenotypes
   in some HCC cell lines. Meanwhile, we emphasized the role of secretory
   SPP1 on the PMN rather than the cancer cell itself in this study.

   The "soil and seed" theory, a widely recognized concept of cancer
   metastasis, proposes two critical steps that contribute to the
   development of metastasis: primary tumors with high metastatic
   potential and conducive environments in targeted organs [[189]5]. Of
   particular interest is the formation of PMN, which has garnered
   significant attention. PMN plays crucial roles in attracting
   circulating tumor cells, facilitating their survival, reactivating
   dormant cells, and fueling subsequent outgrowth. In the present study,
   we revealed that neutrophils significantly accumulated in the lung PMN
   after SPP1 stimulation and that the neutrophil-depleting Ly6G antibody
   had a potent ability to suppress lung metastasis. Neutrophil, a pivotal
   component of the PMN, in primary tumors and target organs is closely
   related to more metastatic events and worse clinical outcomes in
   several cancers [[190]21, [191]43]. And certain neutrophil-recruiting
   chemokines, such as CXCL1 and CXCL5, were found to be elevated in the
   serum of patients with HCC [[192]44], also indicating the crucial role
   of neutrophils in HCC progression. However, the function of neutrophils
   in tumor progression is controversial [[193]20]. With the advancement
   of single-cell sequencing technologies, neutrophil heterogeneity in
   cancers has been revealed, and some studies have validated the dual
   function of neutrophils in cancer progression in different subsets
   [[194]45, [195]46]. Therefore, the conflicting conclusions on
   neutrophil function in previous articles cannot be attributed entirely
   to diverse tumor types, and other functions of neutrophil
   subpopulations should also be further studied.

   Previous studies have indicated that exogenous SPP1 induces neutrophil
   chemotaxis [[196]47]. In our research, we found that SPP1 had a slight
   chemotactic effect on neutrophils in vitro, and this effect seemed
   insufficient to cause a significant accumulation of neutrophils in vivo
   (Fig. [197]3A). We further found that AECII was the significant cell
   type that responded to SPP1, and AECII was the main source of
   SPP1-induced CXCL1 elevation. AECII cells are the most abundant cell
   type in the lung and are the progenitor cells of type I alveolar
   epithelial cells, which are closely related to inflammatory diseases
   [[198]48]. AECII can secrete various chemokines and cytokines to
   recruit immune cells to phagocytes and kill pathogens or
   infected/apoptotic cells to maintain lung homeostasis [[199]49]. A
   previous study revealed that AECII could respond to tumor-derived RNA
   to recruit neutrophils by secreting chemokines in the PMN [[200]23]. In
   the present study, AECII increased CXCL1 secretion to recruit more
   neutrophils from circulation under SPP1 stimulation. Moreover,
   loss-of-function studies revealed that CD44 was the primary receptor
   for SPP1 in MLE-12 cells. After injecting his-tag-labeled SPP1, we
   traced the distribution of SPP1 in different organs and observed that
   the lung had the highest accumulation abundance. These data showed that
   the unbalanced distribution of tumor-derived SPP1 may explain the
   organ-specific metastasis of HCC. Lung epithelial cells highly express
   the CD44 receptor and activation of the downstream chemotaxis signaling
   pathway supports PMN formation and promotes subsequent HCC lung
   metastasis.

   Another important question is how recruited neutrophils promote the
   lung metastasis of HCC. In vitro and in vivo studies revealed that more
   NETs were formed under SPP1 stimulation, which was associated with a
   greater lung metastasis burden. The role of NETs in cancer progression
   is still unclear. While some studies suggest an anti-tumor function of
   NETs, more studies have shown the protumor function of NETs, especially
   in supporting tumor metastasis [[201]50]. Our previous study
   demonstrated that NETs can capture HCC cells and increase their
   metastatic potential by activating the TLR4/9-COX2 axis [[202]26].
   Another study by our team evaluated NETs enrichment levels based on 23
   NETs-related genes in 22 cancers and reported that the combination of
   SPP1 upregulation and NETs enrichment was closely related to EMT in
   cancers [[203]37]. Moreover, NETs are also involved in many steps of
   metastasis, for example, Yang et al. showed that CCDC25 on the surface
   of the tumor cell membrane could sense extracellular DNA and fuel
   breast cancer cells metastasize to the liver, which contains enriched
   NETs [[204]22]. The enrichment of NETs in target organs is positively
   correlated with tumor metastasis, and serum NETs DNA, which is usually
   measured by MPO-DNA, has also become a potential biomarker for
   predicting cancer metastasis. In this study, we also found a positive
   correlation between the MPO-DNA and plasma SPP1 concentration. Elevated
   plasma SPP1 levels may predict the risk of HCC lung metastasis.

   SPP1 has been confirmed as an effective molecule promoting tumor
   progression and metastasis in various cancers. While SPP1 is indeed a
   promising therapeutic target, it is currently not feasible to
   intervene. On the one hand, SPP1 has many important biological
   functions, including mucosa repair process [[205]51], regulating bone
   resorption and matrix turnover [[206]52], and activating and regulating
   various immune cells [[207]53], which makes directly targeting SPP1 can
   lead to severe side effects and off-target effect. On the other hand,
   SPP1 is too complex to develop drugs due to transcriptional and
   post-translational modifications, splice variants, and single
   nucleotide polymorphisms. The clinical studies related to SPP1 are
   limited in disease prediction and diagnosis, such as metastatic breast
   cancer ([208]NCT04274504) and mesothelioma ([209]NCT02029105). However,
   we think there are possibilities for future exploration. In this study,
   we revealed exogenous SPP1 could induce endogenous SPP1 expression
   upregulation (Fig. [210]4N-Q). The self-amplification mechanism of SPP1
   could be a potential therapeutic target and requires further study. In
   addition, several studies have indicated the presence of structural
   variations between tumor-derived SPP1 and other SPP1, which suggested a
   potential avenue for directly targeting tumor-derived SPP1 in the
   future [[211]54]. Nonetheless, additional research is necessary to
   substantiate these preliminary observations. In this manner, we elected
   to intervene at the pivotal stage of tumor metastasis, the PMN.

   CXCR2 inhibitors and DNase I have proven safe in clinical practice, and
   they may be effective interventions that target SPP1-induced PMN
   formation and arrest HCC lung metastasis. CXCR2 is one of the most
   important molecules controlling neutrophil activation and migration.
   CXCR2 antagonists were initially designed to cure pulmonary
   inflammatory diseases and are gradually being used in cancer treatment
   with some success in preclinical studies ([212]NCT02499328,
   [213]NCT02583477). DNase I is a classical destroyer of NETs and
   extracellular chromatin and has been used in SLE and cystic fibrosis
   [[214]39]. However, due to its incomplete degradation of NETs, DNase I
   is usually combined with additional agents. Because pulmonary
   neutrophil accumulation in the PMN and NETs promoted HCC lung
   metastasis, we used the CXCR2 antagonists SB225002 and DNase I, which
   target neutrophil infiltration and NETs formation, respectively, and
   showed that early intervention could significantly reduce the lung
   metastasis burden. In contrast, the late intervention failed
   (Fig. [215]6G). These data emphasized a therapeutic window beyond which
   the antimetastatic effects of anti-neutrophil and anti-NETs treatment
   were attenuated. Given the slight effects of NETs on the growth of HCC,
   which are supported by the orthotopic implantation results (Figure
   S8A-B, D-E), we reasoned the efficacy gap is mainly due to the crucial
   role of neutrophil and NETs in capturing and colonizing circulating
   tumor cells during the initial phases of metastasis in target organs
   (Fig. [216]5E). This notion is supported by some other research, which
   has shown that neutrophils are more important in the early stage of
   lung metastasis, whereas macrophages facilitate the proliferation of
   disseminated cells [[217]55, [218]56]. And Albrengues J et al. reported
   that NETs can awaken dormant tumor cells and facilitate metastasis
   [[219]57]. This study increased our understanding of CXCR2 antagonists
   and revealed that early administration of CXCR2 antagonists with DNase
   I may become a potential therapeutic strategy for preventing HCC lung
   metastasis, especially in SPP1-positive HCC patients.

Supplementary Information

   [220]Supplementary Material 1.^ (139.6MB, docx)
   [221]Supplementary Material 2.^ (25.5KB, xls)
   [222]Supplementary Material 3.^ (24KB, xls)
   [223]Supplementary Material 4.^ (20KB, xls)
   [224]Supplementary Material 5.^ (3.6MB, xls)
   [225]Supplementary Material 6.^ (25.2MB, xls)
   [226]Supplementary Material 7.^ (29KB, xls)
   [227]Supplementary Material 8.^ (23KB, xls)

Acknowledgements